\n日期：<date> 参会者：<names>\n目的：<one line — why are we here?>\n\n决策\n-\n\n行动项\n- [ ] <task> — <owner> — <due date>\n\n未解决问题\n-\n\n备注 / 背景\n-\n\n下次会议 / 后续跟进：<date or trigger>","outro":"就是这样。会前用它准备，会中填写，会后花五分钟整理，然后发出去。这个结构会替你完成大部分工作。"},"faq":{"title":"常见问题"},"faqs":[{"question":"怎样做好会议记录？","answer":"会前准备一个简单模板，然后只捕捉重要内容：决策、有负责人和截止日期的行动项，以及未解决问题。不要试图逐字转写，留意“我们已经决定”“我来处理”“周五前”这样的信号。会议结束后几分钟内整理记录，然后分享到团队已经工作的地方。"},{"question":"会议记录应该包含什么？","answer":"好的会议记录顶部应包含日期、参会者和目的，然后是四个部分：已做决策、行动项（每项都有负责人和截止时间）、未解决问题，以及值得保留的背景信息。其他内容——寒暄、边想边说、幻灯片里已有的内容——都可以省略。"},{"question":"会议记录最好的格式是什么？","answer":"一致、可复用的格式胜过任何聪明格式。最可靠的结构是四个栏目：决策、行动项、未解决问题、备注。每次使用同一模板，意味着你把注意力放在倾听上，而不是决定页面怎么排版，也让不同会议的记录之后更容易浏览和搜索。"},{"question":"会议记录和会议纪要有什么区别？","answer":"会议纪要通常是正式且常常必须保存的记录，包括出席情况、动议和投票，常见于董事会和法律场景。会议记录更轻量、更实用，目的是帮助团队记住决策并推进行动项。大多数工作会议需要的是好的记录，而不是正式纪要。"},{"question":"怎样一边记笔记一边专心听？","answer":"停止试图把所有内容都写下来。当你只记录决策、负责人和截止时间时，大多数时候双手是空的，你可以真正倾听。能录制并转写通话的会议记录工具会彻底移除这个冲突：你在会议中保持专注，会后再编辑现成草稿。"},{"question":"应该录制会议来做记录吗？","answer":"录制能让你保持参与，并获得准确文字稿，而不是实时打字；但一定要先告知参会者，很多地方要求同意。更大的问题是音频会去哪里。多数云端笔记工具会把会议上传到服务器。对于敏感对话，OpenWhispr这样的本地优先工具会在你自己的设备上录制、转写并标注说话人，因此不会上传任何内容。"},{"question":"为了做记录而录制会议合法吗？","answer":"这取决于参会者所在地区。有些地区只需要一方同意，有些地区要求通话中所有人同意，欧盟GDPR等规则还会把录制的声音视为个人数据。这不是法律建议，但在任何地方最安全的习惯都一样：开始前告诉大家你在录制，并取得他们同意。"},{"question":"使用AI会议记录工具时，我的音频会去哪？","answer":"多数云端笔记工具会把会议音频和文字稿上传到公司的服务器进行处理和保存，有些还会用这些数据训练模型。使用OpenWhispr这样的本地优先工具时，录音、文字稿和说话人标签都留在你自己的设备上，永远不会上传。这对HR、法律或涉及客户数据的敏感会议尤其重要。"},{"question":"主持会议时怎样记笔记？","answer":"当你在主持会议时，你不可能同时打字。要么指定记录员，要么让会议记录工具捕捉文字稿，这样你能专注于现场。如果必须自己处理，就只用短片段记下决策和行动项，然后在会后五分钟内趁记忆新鲜把它们展开。"},{"question":"最好的会议记录应用是什么？","answer":"最好的会议记录应用，是能准确捕捉对话并符合你隐私需求的那个。Otter、Fireflies、Granola等云端工具很方便，但会把你的音频发送到它们的服务器。OpenWhispr是免费的开源会议记录工具，可以在本地转写并识别说话人，因此音频永远不会离开你的设备。"}],"cta":{"title":"不用打一个字也能做会议记录","body":"OpenWhispr是一款免费、开源的会议记录工具，可以录制、转写并标注谁说了什么，而且完全在你的设备上完成。你可以专注参会，会后再编辑现成草稿。支持macOS、Windows、Linux 和 iOS。","starButton":"在GitHub上给我们加星","footnote":"无需账户 · 可离线使用 · 永远开源"},"internalLinks":[{"href":"/blog/local-speaker-diarization","label":"私密会议记录工具"},{"href":"/blog/local-vs-cloud-transcription","label":"本地 vs 云端转写"},{"href":"/blog/best-dictation-tools-windows-2026","label":"Windows最佳听写工具"},{"href":"/blog","label":"所有博客文章"}]},"blogHowWhisper":{"metadata":{"title":"Whisper AI 工作原理：完整指南 | OpenWhispr","description":"详解 Whisper AI 的工作原理：Transformer 架构、68 万小时训练数据、模型大小和准确率基准。开源，并可在设备本地运行。","keywords":["Whisper AI 工作原理","Whisper 语音识别","OpenAI Whisper 解释","Whisper 模型架构","Whisper AI 精度","Whisper Transformer 模型","自动语音识别","Whisper vs 其他 ASR","whisper.cpp 解释"]},"schema":{"title":"Whisper AI 的工作原理：完整指南","description":"深入解析OpenAI的Whisper语音识别模型 — 架构、训练数据、精度基准和实际使用。"},"breadcrumb":{"home":"首页","blog":"博客","article":"Whisper AI 的工作原理"},"hero":{"title":"Whisper AI 的工作原理：完整指南","subtitle":"对OpenAI开源语音识别模型的技术深挖：从梅尔频谱图到Transformer解码。","date":"2026年2月3日","tag":"技术"},"sections":[{"id":"what-is-whisper","title":"Whisper是什么？"},{"id":"architecture","title":"架构"},{"id":"training-data","title":"训练数据"},{"id":"model-sizes","title":"模型尺寸"},{"id":"accuracy","title":"精度"},{"id":"whisper-cpp","title":"whisper.cpp"},{"id":"comparison","title":"ASR对比"},{"id":"applications","title":"应用"},{"id":"sources","title":"来源"}],"intro":{"paragraph":"Whisper是OpenAI开发的自动语音识别(ASR)模型。 它于2022年9月发布，是一个通用语音识别系统，使用从互联网收集的68万小时多语言音频数据训练而成。Whisper采用编码器-解码器Transformer架构，将音频转换为log-mel频谱图，通过神经网络编码器处理，并以自回归方式解码文本token。它支持99种语言的转写、翻译为英文、语言识别和时间戳预测，而且全部在一个统一模型中完成。与多数商业ASR服务不同，Whisper的权重在MIT许可证下<openSourceLink>完全开源</openSourceLink>，由此催生了繁荣的移植、优化和应用生态，包括<whisperCppLink>whisper.cpp</whisperCppLink>这个C/C++实现，让Whisper能够实际用于实时、设备端语音识别。","lastUpdated":"最后更新：2026年2月17日。本文中的定量主张均已与至少两个主要来源交叉核对。"},"factCheck":{"title":"事实核查快照（双来源）","items":[{"label":"Whisper发布时间、训练规模和许可证：","body":"OpenAI于2022年9月发布Whisper，使用68万小时数据训练，并采用MIT许可证。"},{"label":"任务范围和语言支持：","body":"Whisper支持99种语言的转写、翻译成英文、语言识别和时间戳token。"},{"label":"英文基准背景：","body":"约3% WER是在LibriSpeech test-clean上针对英文评估的基准数字，不是通用真实世界错误率。"},{"label":"近期模型更新：","body":"large-v3在许多语言上报告的错误率低于large-v2，turbo则通过更小的解码器计算量专注于降低延迟。"},{"label":"与OpenWhispr的关系：","body":"OpenWhispr通过whisper.cpp使用Whisper，为桌面平台提供本地、离线优先的听写。"}],"sources":["<a1>arXiv论文</a1> · <a2>OpenAI Whisper仓库</a2>","<a1>Whisper模型卡</a1> · <a2>Whisper README</a2>","<a1>large-v2模型卡</a1> · <a2>基准详情</a2>","<a1>large-v3模型卡</a1> · <a2>turbo发布讨论</a2>","<a1>whisper.cpp</a1> · <a2>OpenWhispr</a2>"],"sourceUrls":[["https://arxiv.org/abs/2212.04356","https://github.com/openai/whisper"],["https://github.com/openai/whisper/blob/main/model-card.md","https://github.com/openai/whisper"],["https://huggingface.co/openai/whisper-large-v2","https://github.com/openai/whisper/blob/main/model-card.md"],["https://huggingface.co/openai/whisper-large-v3","https://github.com/openai/whisper/discussions/2363"],["https://github.com/ggml-org/whisper.cpp","https://openwhispr.com"]]},"whatIs":{"title":"Whisper是什么？","paragraph1":"Whisper由OpenAI的Alec Radford、Jong Wook Kim、Tao Xu、Greg Brockman、Christine McLeavey和Ilya Sutskever在论文<paperLink>“Robust Speech Recognition via Large-Scale Weak Supervision”</paperLink>中提出。代码和模型权重于2022年9月以MIT许可证发布在<githubLink>GitHub</githubLink>上。","paragraph2":"在Whisper之前，最先进的语音识别系统通常使用经过整理、人工标注的数据集训练，这些数据集往往局限于特定语言、口音或领域。Whisper打破了这一做法，使用从互联网抓取的音频与文字稿配对数据，规模达到68万小时，且数据非常多样。关键洞见在于，这种“弱监督”方式——使用不完美的机器生成或用户上传文字稿，而不是人工逐条标注的数据——可以产生一个在语言、口音、背景噪声和技术词汇上都非常稳健的模型。","paragraph3":"结果是一个可以处理多种语音任务的单一模型：转写（同语种语音转文本）、翻译（任意语言语音转英文文本）、语言识别，以及带时间戳的语音活动检测。这种多任务能力通过一套特殊token系统嵌入到模型架构中，用这些token指定要执行的任务。","paragraph4":"Whisper的开源发布意义重大。它让高质量ASR变得人人可用，开发者、研究人员和公司都能使用一个可与商业API竞争的模型，而且不需要按分钟付费、不需要把音频发送到云端，也没有供应商锁定。这推动了<whisperCppLink>whisper.cpp</whisperCppLink>、<fasterWhisperLink>Faster Whisper</fasterWhisperLink>等工具，以及建立在它们之上的几十个应用的发展，其中也包括<openWhisprLink>OpenWhispr</openWhisprLink>。"},"architecture":{"title":"架构深挖","intro":"Whisper使用编码器-解码器Transformer架构，这与原始Transformer（Vaswani et al., 2017）以及许多机器翻译系统背后的基本设计相同。编码器处理音频，解码器生成文本。下面是每个阶段的工作方式。","preprocessing":{"title":"音频预处理","intro":"在任何神经网络处理之前，原始音频会被转换成一种声音的可视化表示，称为log-mel频谱图。流程如下：","steps":["音频被重采样为16,000 Hz（16 kHz单声道）。","使用25毫秒窗口和10毫秒步幅（hop length）计算短时傅里叶变换(STFT)。","频谱被投影到80个梅尔频率滤波器组（large-v3为128个）上，这些滤波器通过对数间隔近似人类听觉感知。","应用对数缩放，生成最终的log-mel频谱图。","频谱图被分成30秒块。短于30秒的音频会补零，长音频会按顺序分块处理。"],"outro":"结果是一个形状为<code>(80, 3000)</code>的二维表示——80个梅尔通道，3,000个时间步（30秒，10ms步幅）。它类似一张图像，编码器处理它的方式也很像视觉模型处理像素数据。"},"encoder":{"title":"编码器","intro":"编码器把梅尔频谱图转换成一串学习到的音频表示。它包含：","steps":["两个1D卷积层 — 第一个卷积的kernel size为3、padding为1，后接GELU激活。第二个卷积的kernel size为3、stride为2、padding为1，会把时间维度减半，把3,000个时间步下采样到1,500个位置。","正弦位置嵌入 — 与解码器不同，编码器使用固定的正弦嵌入（不是学习得到的）来编码每个时间步的位置。","Transformer块 — 一叠标准Transformer编码器层，每层包含多头自注意力和带GELU激活的前馈网络，并通过残差连接和层归一化连接。"],"outro":"输出是一串1,500个上下文化音频特征向量——输入音频每20毫秒对应一个向量——解码器在生成文本时会对这些向量进行注意力计算。"},"decoder":{"title":"解码器","intro":"解码器以自回归方式生成文本token：一次生成一个token，每个token都依赖编码后的音频和之前生成的所有token。它包含：","steps":["Token嵌入层 — 将token ID转换为稠密向量表示。","学习到的位置嵌入 — 与编码器的正弦嵌入不同，解码器使用随模型一起训练的位置嵌入。","带交叉注意力的Transformer块 — 每个解码器层有三个子层：masked self-attention（防止模型看到未来token）、对编码器输出的cross-attention（让模型“听”音频），以及前馈网络。"],"outro":"最终的解码器输出会被投影到词表上，生成token概率。解码策略包括贪心搜索、束搜索和基于温度的采样。"},"tokens":{"title":"特殊Token与多任务训练","intro":"Whisper通过一套巧妙的特殊token系统，用单个模型执行多种任务。这些token充当指令。解码器输入遵循如下结构：","startOfTranscript":"— 表示输出序列开始。","language":"— 指定语言（99个语言token之一）。可以自动检测，也可以手动设置。","taskJoin":"或","task":"— 指定是转写原语言，还是翻译成英文。","timestampsPart1":"— 控制是否生成时间戳token。启用时间戳时，模型会生成类似","timestampsJoin":"和","timestampsPart2":"这样的时间偏移token，将文本与音频对齐。","outro":"这种统一token格式意味着，一个模型无需单独的模型头或微调，就能执行转写、翻译、语言检测和带时间戳转写。任务完全由提供给解码器的token序列决定。"}},"trainingData":{"title":"训练数据","paragraph1":"原始Whisper模型（tiny到large-v2）使用从互联网收集的音频和文字稿配对数据训练，总量为680,000小时。这个数据集并未公开。它的关键特征是弱监督：文字稿标签没有经过人工标注者逐条验证，而是来自字幕、闭合字幕以及其他用户生成的音频配套文本。","compositionTitle":"数据组成","compositionIntro":"根据<modelCardLink>官方模型卡</modelCardLink>，68万小时训练集组成如下：","headers":{"segment":"片段","hours":"小时","share":"占比","description":"说明"},"rows":[{"segment":"英语 ASR","hours":"438,000","share":"65%","description":"带英文文字稿的英文音频"},{"segment":"翻译（其他语言到英语）","hours":"126,000","share":"18%","description":"带英文文字稿的非英文音频"},{"segment":"多语言 ASR","hours":"117,000","share":"17%","description":"带母语文字稿的非英文音频（覆盖98种语言）"}],"paragraph2":"训练数据中英文占比很高（65%），这解释了为什么Whisper在英文上的表现明显优于低资源语言。对于2023年11月发布的large-v3模型，OpenAI将训练集扩展到100万小时弱标注音频，并额外加入400万小时伪标注音频，这些伪标签是用Whisper large-v2在未标注数据上生成的，是一种自训练或知识蒸馏形式。","weakTitle":"为什么“弱监督”很重要","weakParagraph1":"大多数早期ASR系统使用LibriSpeech（960小时）或Common Voice（每种语言几千小时）这样的数据集训练，这些数据集经过精心整理并由人类验证。Whisper选择用标签质量换规模：68万小时对比传统系统使用的几百到几千小时。论文证明这个取舍是值得的。来自互联网的数据多样性让Whisper在真实世界条件下非常稳健：背景噪声、重叠语音、口音、领域词汇，以及会让基于录音棚级朗读语音训练的模型感到困难的声学环境。","weakParagraph2":"不过，弱监督也带来了一个已知限制：幻觉。因为训练文字稿并不完美，模型有时会生成听起来合理但实际并未被说出的文本，尤其是在静音或非常安静的片段中。这是该方法固有的取舍，也是仍在持续改进的方向。"},"modelSizes":{"title":"模型尺寸","intro":"Whisper提供多个尺寸，从3900万到15.5亿参数不等。小模型更快但精度较低；大模型更准确但需要更多计算和内存。<code>.en</code>变体是英文专用模型，通常在英文上表现更好，尤其是较小尺寸。large模型没有英文专用版本。","headers":{"model":"模型","parameters":"参数量","englishOnly":"英文专用","vram":"VRAM","speed":"相对速度"},"rows":[{"model":"tiny","parameters":"39 M","englishOnly":"tiny.en","vram":"~1 GB","speed":"~10x"},{"model":"base","parameters":"74 M","englishOnly":"base.en","vram":"~1 GB","speed":"~7x"},{"model":"small","parameters":"244 M","englishOnly":"small.en","vram":"~2 GB","speed":"~4x"},{"model":"medium","parameters":"769 M","englishOnly":"medium.en","vram":"~5 GB","speed":"~2x"},{"model":"large (v1, v2, v3)","parameters":"1,550 M","englishOnly":"--","vram":"~10 GB","speed":"1x"},{"model":"turbo (large-v3-turbo)","parameters":"809 M","englishOnly":"--","vram":"~6 GB","speed":"~8x"}],"tableNote":"速度以large模型为参照。VRAM要求指原始Python实现中通过fp16精度进行GPU推理。<whisperCppLink>whisper.cpp</whisperCppLink>使用的内存明显更少（例如large模型需要约3.9 GB RAM，而不是约10 GB VRAM）。","turbo":"turbo模型（2024年10月发布）是large-v3的蒸馏版本，解码器显著更小。它保留了large-v3的大部分精度，同时速度约快8倍，因此在速度重要时是很强的选择。注意，turbo不支持翻译任务。","large":"large模型经历了三次迭代：large-v1（2022年9月）、large-v2（2022年12月）和large-v3（2023年11月）。每个版本都提高了精度，其中large-v3引入了128个梅尔频率bin（从80个提升），并使用更大的数据集训练。关于每个尺寸以及如何选择的详细拆解，请参阅我们的<guideLink>Whisper模型尺寸指南</guideLink>。"},"accuracy":{"title":"Whisper有多准确？","intro":"Whisper的精度通常用词错误率(WER)衡量，即被错误转写的词所占比例（插入、删除和替换合并计算）。WER越低越好。","englishTitle":"英文基准","englishParagraph":"在LibriSpeech test-clean上，Whisper large-v2的WER约为3.0%<ref1>[1]</ref1><ref2>[2]</ref2>。这是最常用的英文ASR基准之一，由有声书中的干净朗读语音组成。这个结果可与商业ASR系统和专门训练的模型竞争，尽管在LibriSpeech这个特定基准上，针对该数据集微调的专用模型可以达到更低WER。Whisper的强项不在于击败狭窄基准，而在于稳健性，即在广泛真实世界条件下保持稳定表现。","largeV3Title":"Large-v3的改进","largeV3Paragraph":"根据OpenAI的评估，在Common Voice 15和Fleurs评估数据集上，对于模型错误率低于60%的语言，Whisper large-v3相较large-v2显示出10-20%的错误减少<ref2>[2]</ref2><ref3>[3]</ref3>。这些改进在非英文语言上尤其明显，因为扩大的训练数据（总计500万小时）在这些语言上带来的差异最大。","compareTitle":"Whisper的对比表现（2026年2月）","compareParagraph":"自Whisper发布以来，更新的开源ASR模型已经在干净语音基准上超过它。NVIDIA的Parakeet TDT（0.6B参数）在LibriSpeech test-clean上达到1.69% WER，Canary模型达到1.6% WER，都明显低于Whisper large-v3的约2.7%。不过，由于生态成熟、语言覆盖广，以及whisper.cpp等优化运行时可用，Whisper仍是部署最广泛的开源ASR模型。","chartNote":"注：商业云API（Deepgram Nova-3、Google Chirp、AssemblyAI Universal-2）不公布LibriSpeech WER数字，因此无法在这张图中直接比较。它们的基准使用专有真实世界测试集。OpenAI还在2025年3月发布了GPT-4o-transcribe，这是一个仅API可用的模型（不开源，也不基于Whisper），据称错误率更低，但它只能在云端使用，无法本地推理。","excelsTitle":"Whisper擅长的地方","excels":[{"title":"对口音和方言的稳健性","body":"— 使用多样化互联网音频训练，Whisper比只在朗读语音上训练的系统更能处理地区口音。"},{"title":"背景噪声耐受性","body":"— 即使有音乐、其他说话人或环境噪声，模型也能保持合理精度。"},{"title":"专业词汇","body":"— 由于训练数据覆盖广，Whisper比许多通用ASR系统更能处理特定领域术语（医疗、法律、技术）。"},{"title":"多语言能力","body":"— 单一模型支持99种语言，无需按语言单独微调。"}],"strugglesTitle":"Whisper吃力的地方","struggles":[{"title":"幻觉","body":"— 最常被提到的问题。在静音或非常安静的片段中，Whisper可能生成从未说出的文本。这是弱监督训练方式的后果。"},{"title":"低资源语言","body":"— 训练数据较少的语言（68万小时中大部分是英文）错误率明显更高。"},{"title":"重复文本生成","body":"— 自回归解码器可能会“卡住”，反复生成同一句话。带特定参数的束搜索可以缓解，但不能完全消除。"},{"title":"默认并非实时","body":"— Whisper处理30秒块，因此存在固有延迟。流式方案（如whisper.cpp的实时模式）可以绕过这一点，但会增加复杂度。"}]},"whisperCpp":{"title":"whisper.cpp与本地推理","paragraph1":"原始Whisper实现需要Python、PyTorch和支持CUDA的GPU，才能获得合理性能。这使它不适合桌面应用、移动设备和边缘部署。Georgi Gerganov创建的<whisperCppLink>whisper.cpp</whisperCppLink>解决了这个问题。","paragraph2":"whisper.cpp是一个用纯C/C++完整重新实现的Whisper推理引擎，没有依赖。它读取相同的模型权重，但运行时不需要Python、不需要PyTorch，也不需要GPU（虽然有GPU会大幅受益）。关键特性包括：","features":[{"title":"零运行时内存分配","body":"— 所有内存都预先分配，使性能可预测，并适合嵌入式系统。"},{"title":"Apple Silicon优化","body":"— 在搭载M系列芯片的Mac上，编码器可以通过Core ML运行在Apple Neural Engine上，相比仅CPU执行快3倍以上。Metal GPU加速也支持完整GPU推理。"},{"title":"量化支持","body":"— 模型可量化为4位、5位或8位整数精度，在几乎不损失精度的情况下显著降低内存使用并提高速度。"},{"title":"GPU加速","body":"— 支持NVIDIA CUDA（通过cuBLAS）、Vulkan（跨厂商）、OpenVINO（Intel）以及用于CPU加速的OpenBLAS。"},{"title":"语音活动检测(VAD)","body":"— 内置VAD可跳过静音片段，同时提升速度和文字稿质量。"}],"memoryTitle":"内存要求（whisper.cpp）","memoryHeaders":{"model":"模型","disk":"磁盘大小","ram":"所需RAM"},"memoryRows":[{"model":"tiny","disk":"75 MB","ram":"~273 MB"},{"model":"base","disk":"142 MB","ram":"~388 MB"},{"model":"small","disk":"466 MB","ram":"~852 MB"},{"model":"medium","disk":"1.5 GB","ram":"~2.1 GB"},{"model":"large","disk":"2.9 GB","ram":"~3.9 GB"}],"paragraph3":"whisper.cpp可运行在macOS（Intel和Apple Silicon）、Linux、Windows、iOS、Android、FreeBSD、Raspberry Pi，甚至可通过WebAssembly在浏览器中运行。它能在普通消费级硬件上实现快于实时的转写，包括Raspberry Pi 4和iPhone 13这样的小设备。","paragraph4":"正是这种可移植性，让本地、私密的语音识别变得可行。像<openWhisprLink>OpenWhispr</openWhisprLink>这样的应用在底层使用whisper.cpp，提供完全运行在用户设备上的实时按键说话听写，音频永远不会离开机器。"},"comparison":{"title":"Whisper vs 其他ASR系统","intro":"Whisper处在更大的语音识别系统生态中。下面是它与常见替代方案的比较。","cards":[{"title":"Google Cloud Speech-to-Text","body":"Google的云端ASR服务。精度优秀，尤其是英文，支持实时流式处理和话者分离。主要区别在于：它是专有服务，需要把音频发送到Google服务器，并按处理的音频分钟计费。Whisper在通用场景下精度相近，可本地运行且免费，但缺少内置流式处理和话者分离。","tags":["仅云端","专有","按量付费"]},{"title":"Amazon Transcribe (AWS)","body":"Amazon的云端ASR，与AWS生态紧密集成。提供医疗转写、呼叫分析和自定义词汇。取舍与Google类似：依赖云端、按分钟计费、专有。对于需要离线运行、开源许可或避免供应商锁定的应用，Whisper更合适。","tags":["仅云端","专有","AWS集成"]},{"title":"Mozilla DeepSpeech","body":"Mozilla基于Baidu Deep Speech研究的开源ASR引擎。它曾是最早的高质量开源ASR模型之一，但Mozilla在2021年停止了活跃开发。Whisper在精度、多语言支持和社区势能上已经实质性取代DeepSpeech。DeepSpeech仍可作为轻量、易理解的基线，但不再推荐用于新项目。","tags":["开源","已停止维护","偏重英文"]},{"title":"Vosk","body":"一个开源离线语音识别工具包，支持20多种语言，模型小而快。Vosk非常适合资源受限环境，它的模型只有Whisper的一小部分大小，并且能在低功耗硬件上良好运行。取舍是精度：Vosk明显不如Whisper，尤其在口音语音、噪声音频和专业词汇上。","tags":["开源","轻量","精度较低"]},{"title":"Faster Whisper (CTranslate2)","body":"使用SYSTRAN的CTranslate2推理引擎重新实现的Whisper。它运行相同的Whisper模型，精度相同，但速度最高可比原始OpenAI实现快4倍，内存使用也更低。Faster Whisper基于Python（不同于whisper.cpp的C/C++），是服务器端批量转写的强选择。它支持CUDA GPU和带INT8量化的CPU推理。","tags":["开源","Python / CTranslate2","相同Whisper精度"]}]},"applications":{"title":"实际应用","intro":"Whisper结合了精度、多语言支持和开源可用性，已经成为大量应用的基础：","cards":[{"title":"桌面听写","body":"<openWhisprLink>OpenWhispr</openWhisprLink>等应用使用whisper.cpp，在macOS、Windows和Linux上提供系统级按键说话听写。音频在本地处理，不会发送到云端。"},{"title":"字幕生成","body":"Whisper的时间戳预测非常适合生成字幕和闭合字幕。<stableTsLink>stable-ts</stableTsLink>和<autoSubtitleLink>auto-subtitle</autoSubtitleLink>等工具会自动化这一工作流。"},{"title":"播客和会议转写","body":"Whisper很擅长长音频转写，尤其是配合分块处理方式。许多播客编辑器和会议记录工具使用Whisper（或Faster Whisper）批量转写录音。"},{"title":"翻译和本地化","body":"Whisper内置的翻译能力（任意语言转英文）用于跨语言内容访问、媒体本地化和实时翻译原型。"},{"title":"无障碍","body":"面向聋人和听障用户的实时字幕、面向运动障碍用户的语音界面，以及音频描述转写，都是Whisper正在被部署的活跃领域。"},{"title":"数据标注和研究","body":"研究人员使用Whisper创建伪标注数据集来训练其他模型，这与OpenAI创建large-v3训练数据的方法相同。它也被广泛用于语言学研究和语料库构建。"}]},"sources":{"title":"来源和延伸阅读","items":[{"label":"[论文]","text":"Radford, A., Kim, J. W., Xu, T., Brockman, G., McLeavey, C., & Sutskever, I. (2022). Robust Speech Recognition via Large-Scale Weak Supervision. arXiv:2212.04356.","linkText":"arxiv.org/abs/2212.04356"},{"label":"[代码]","text":"OpenAI Whisper GitHub仓库。","linkText":"github.com/openai/whisper"},{"label":"[代码]","text":"whisper.cpp — Georgi Gerganov的C/C++移植版。","linkText":"github.com/ggml-org/whisper.cpp"},{"label":"[模型卡]","text":"Hugging Face上的Whisper large-v2（LibriSpeech test-clean WER基准：约3.0%）。","linkText":"huggingface.co/openai/whisper-large-v2"},{"label":"[模型卡]","text":"Hugging Face上的Whisper large-v3（相较v2错误减少10-20%）。","linkText":"huggingface.co/openai/whisper-large-v3"},{"label":"[模型卡]","text":"Hugging Face上的Whisper large-v3-turbo（面向低延迟的蒸馏变体）。","linkText":"huggingface.co/openai/whisper-large-v3-turbo"},{"label":"[模型卡]","text":"OpenAI Whisper模型卡（训练数据组成拆解）。","linkText":"github.com/openai/whisper/blob/main/model-card.md"},{"label":"[发布说明]","text":"OpenAI large-v3发布公告和评估说明。","linkText":"github.com/openai/whisper/discussions/1762"},{"label":"[发布说明]","text":"OpenAI turbo发布讨论和速度取舍。","linkText":"github.com/openai/whisper/discussions/2363"},{"label":"[代码]","text":"Faster Whisper — SYSTRAN基于CTranslate2的重新实现。","linkText":"github.com/SYSTRAN/faster-whisper"},{"label":"[产品]","text":"OpenWhispr通过whisper.cpp使用Whisper进行本地听写工作流。","linkText":"openwhispr.com"}],"links":["https://arxiv.org/abs/2212.04356","https://github.com/openai/whisper","https://github.com/ggml-org/whisper.cpp","https://huggingface.co/openai/whisper-large-v2","https://huggingface.co/openai/whisper-large-v3","https://huggingface.co/openai/whisper-large-v3-turbo","https://github.com/openai/whisper/blob/main/model-card.md","https://github.com/openai/whisper/discussions/1762","https://github.com/openai/whisper/discussions/2363","https://github.com/SYSTRAN/faster-whisper","https://openwhispr.com"]},"cta":{"title":"用OpenWhispr在本地试用Whisper","body":"OpenWhispr使用Whisper（通过whisper.cpp）在macOS、Windows和Linux上提供本地、私密的听写。按键说话、支持99+种语言、无需云端。免费试用。","starButton":"在GitHub上给我们加星","footnote":"无需账户 · 可离线使用 · 永远开源"},"internalLinks":[{"href":"/blog/whisper-model-sizes-explained","label":"Whisper模型尺寸对比"},{"href":"/blog/local-vs-cloud-transcription","label":"本地 vs 云端转写"},{"href":"/blog","label":"所有博客文章"},{"href":"/use-cases/developers","label":"面向开发者的OpenWhispr"},{"href":"/compare","label":"比较听写工具"}]},"blogVoicePrivacy":{"metadata":{"title":"语音数据隐私：让音频留在设备上 | OpenWhispr","description":"云端语音工具会记录你的声音。OpenWhispr 的本地 AI 转录让音频留在设备上，保持私密并完全离线；免费且开源。","keywords":["本地 AI 隐私","端侧 AI","语音数据隐私","本地语音识别","私密 AI 处理","边缘 AI 语音","本地 LLM","无需云端的 AI","语音数据安全","nvidia parakeet"]},"schema":{"title":"本地 AI 的理由：为什么你的语音数据不该离开你的设备","description":"AI 已经强大到足以在你的笔记本上运行。这里讲清楚为什么本地处理对隐私、安全和掌控权更有利。"},"breadcrumb":{"home":"首页","blog":"博客","article":"本地 AI 的理由"},"layout":{"title":"本地 AI 的理由：为什么你的语音数据不该离开你的设备","subtitle":"你口袋里的硬件，已经强大到足以运行可与云端服务匹敌的语音识别。那我们为什么还在把语音数据发往服务器？","date":"2026 年 2 月 12 日","tag":"博客"},"sections":[{"id":"opening","title":"概览"},{"id":"fact-check","title":"事实核查速览"},{"id":"local-ai-revolution","title":"本地 AI 革命"},{"id":"voice-sensitivity","title":"语音数据的敏感性"},{"id":"voice-in-the-cloud","title":"云端中的语音"},{"id":"performance-gap","title":"性能差距"},{"id":"regulatory-landscape","title":"监管格局"},{"id":"cloud-makes-sense","title":"何时该用云端"},{"id":"future-of-local-ai","title":"本地 AI 的未来"},{"id":"what-you-can-do","title":"你能做什么"},{"id":"sources","title":"参考来源"}],"opening":{"lead":"本地 AI 完全在你的设备上处理数据——无需联网，没有数据离开你的机器，也不必把你的信息托付给第三方。","body1":"借助诸如","whisperLink":"通过 whisper.cpp 高效运行的 OpenAI Whisper","body2":"以及 NVIDIA Parakeet 这样的模型，消费级笔记本如今已能以匹敌或接近云端服务的准确率转录语音。本文专门论述：为什么语音数据应当留在你的设备上——以及为什么让这一切成真的技术早已就绪。","lastUpdated":"最后更新：2026 年 2 月 17 日。文中的定量与法律性陈述均经过至少两个一手来源交叉核验。"},"factCheck":{"title":"事实核查速览（双重来源）","item1":{"lead":"在消费级硬件上运行本地 ASR 是切实可行的：","body":"Whisper + whisper.cpp 与 Parakeet 都有公开的模型/运行时文档，并附有硬件指引。"},"item2":{"lead":"云服务商的日志记录行为是可配置的，而非一刀切：","body":"数据处理方式取决于厂商默认设置以及选择加入/退出的选项。","link1":"Google 数据日志记录","link2":"AWS AI 退出政策"},"item3":{"lead":"语音数据可能触发严格的合规义务：","body":"生物识别与 AI 系统风险规则已被写入欧盟法律。","link1":"GDPR 官方文本","link2":"《欧盟 AI 法案》官方文本"},"item4":{"lead":"数据泄露的财务影响实实在在：","body":"泄露成本居高不下，这让集中存储敏感音频的代价更加沉重。","link1":"IBM 数据泄露报告","link2":"FTC 语音克隆风险警示"},"item5":{"lead":"OpenWhispr 背景：","body":"OpenWhispr 以本地优先转录为设计理念，云端使用为可选项。","link2":"whisper.cpp 后端"}},"revolution":{"title":"本地 AI 革命：发生了什么变化","p1":"五年前，在笔记本上运行一个有实用价值的 AI 模型还不切实际。模型太大，硬件太慢，软件生态几乎不存在。如今这一切已发生翻天覆地的变化。","p2a":"催化剂正是","p2b":"——Georgi Gerganov 用 C/C++ 移植的 OpenAI Whisper 语音识别模型。通过用纯 C++ 重写模型推理、且零运行时依赖，Gerganov 让 Whisper 得以在从 MacBook 到 Raspberry Pi 的各种设备上运行。该项目支持的模型从 75 MB 的「tiny」版本（约需 273 MB 内存），一直到磁盘占用 2.9 GB 的完整「large-v3」模型——并可通过整数量化进一步降低内存占用。它能在 Mac（Intel 与 Apple Silicon）、Linux、Windows、Android、iOS 上运行，甚至能在 WebAssembly 中运行。","p3a":"同样的思路在整个 AI 领域全面爆发。Gerganov 的","p3b":"之于大语言模型，正如 whisper.cpp 之于语音识别——它把 LLaMA、Mistral、Qwen、DeepSeek 以及数十个其他模型带到了消费级硬件上，量化精度从 8 位一路低至 1.5 位。诸如","p3c":"和","p3d":"等工具，则把这些能力封装进了易于上手的界面。","p4a":"硬件也跟了上来。Apple 自 2020 年的 M1 起推出的 M 系列芯片，为每台 Mac 都配备了统一内存架构和专用的 Neural Engine——让本地 AI 推理快到足以实时使用。Apple 在","p4b":"上更是大举投入：默认在设备端处理 AI 任务，仅在必要时才路由到 Apple 的 Private Cloud Compute 服务器——并以强有力的保证承诺数据绝不存储、仅用于当下这一次请求。","p5":"趋势已经很清楚：模型正在同时变得更小、更好。知识蒸馏（训练小模型来模仿大模型）和量化（在不破坏准确率的前提下降低数值精度）等技术，意味着 2023 年还需要数据中心 GPU 才能跑的东西，到 2026 年已能在笔记本上运行。例如，Whisper 的「large-v3-turbo」模型在保持相当准确率的同时，速度远快于其前代。"},"sensitivity":{"title":"为什么语音数据格外敏感","p1":"并非所有数据都同样敏感。文本可以匿名化，浏览历史可以清除。但语音不同——它是一种生物识别标识符。你的声音和指纹一样，独属于你。","p2":"一段你的录音所能揭示的，远不止你说出的那些字词。语音分析领域的研究已经证明，声音特征可以反映：","identity":{"title":"身份特征","item1":"你独一无二的身份（银行和安全系统都用声音做生物识别认证）","item2":"性别、大致年龄和地域口音","item3":"母语与社会经济背景"},"health":{"title":"健康与情绪状态","item1":"情绪状态——压力、焦虑、疲劳和抑郁都可从声音模式中检测出来","item2":"神经系统状况——已有研究将声音生物标志物用于帕金森病的早期检测","item3":"呼吸系统健康——新冠（COVID-19）、哮喘等状况会影响发声特征"},"p3":"这并非空想。已有公司正在积极开发分析语音的产品，用于健康筛查、保险风险评估和招聘决策。语音数据丰富的生物识别信息，正是它价值所在——也正是它一旦脱离你的掌控就变得危险的原因。","p4a":"与密码不同，一旦声音泄露，你无法更换自己的声音。一段录音一旦落到服务器上，你就永久失去了对这份生物识别数据的独占控制。而数据泄露并非杞人忧天——IBM 的《2025 年数据泄露成本报告》发现，全球单次泄露的平均成本","breachCost1":"488 万美元","p4b":"（2024 年）","breachCost2":"444 万美元","p4c":"（2025 年）","ibmCite":"IBM，2025 年","p4d":"问题不在于泄露是否会发生，而在于何时发生。"},"cloud":{"title":"你的语音在云端会经历什么","p1":"当你使用基于云端的语音识别服务时，你的音频会被传输到远程服务器进行处理。此后会发生什么，因服务商而异——而细节至关重要。","google":{"a":"Google 的文档指出，","bold1":"默认情况下，Cloud Speech-to-Text 不会记录客户的音频数据或转录文本","b":"。不过，Google 提供了一项自愿的数据日志记录计划，用户可选择加入，以共享音频数据换取折扣价。一旦加入，Google 会用这些数据来「提升服务质量」。值得注意的是，被记录的数据","bold2":"不会在你删除项目时一并删除","c":"——你必须另行提交删除请求。","cite":"Google Cloud 文档"},"aws":{"a":"包括 Amazon Transcribe 在内的 AWS AI 服务，","bold1":"可能会使用客户内容来开发和改进 AWS 服务","b":"，除非用户通过组织级策略明确选择退出。AWS 在 2023 年迫于公众压力更新了该策略，但在许多地区，默认设置仍允许将数据用于模型改进。","cite":"AWS Transcribe 文档"},"azure":{"a":"Azure 的 Speech Service","bold1":"实时处理音频，默认不会留存客户的音频数据","b":"。不过，选择加入自定义模型训练的用户，其数据将被存储。Microsoft 的数据处理受","cite":"《数据保护附录》","c":"约束，其条款因服务层级和地区而异。"},"p2a":"撇开服务商本身不谈，还要考虑基础设施链条。你的音频可能要经过多个网络节点跳转，在另一个国家的数据中心被处理，并有可能被分包处理方或承包商访问到。即便服务商政策再严格，存储在美国境内服务器上的数据，仍可能依据","effLink":"《国家安全信函》和《外国情报监视法》第 702 条","p2b":"被政府调取，而这些都无需通知数据主体。","callout":{"lead":"需要说明的是：","body":"这些服务商通常都有强有力的安全实践，其政策也有正当理由。重点不在于云服务居心不良——而在于把数据发给任何第三方，本质上就意味着要信任它们的政策、它们的安全和它们所处的司法管辖区。本地处理则彻底消除了这种信任的必要。"}},"performance":{"title":"性能差距正在缩小","p1":"支持云端转录的传统论据很简单：云端模型准确率高得多。如今这道鸿沟已大幅收窄。","p2":"2023 年底发布的 OpenAI Whisper large-v3，在大多数语言上的词错误率已能与商用云端 API 一较高下。NVIDIA 的 Parakeet 模型则把准确率推得更远，在标准英语基准测试中取得了一些最低的 WER。针对英语语音的独立基准测试显示，这些开源模型在干净语音上的词错误率（WER）落在 3% 至 5% 区间——与 Google 和 AWS 的商用产品相当甚至更优。更新的「large-v3-turbo」版本在保持相近准确率的同时速度大幅提升，让现代硬件上的实时本地转录变得切实可行。","stat1":"Whisper 模型尺寸（从 tiny 到 large）","stat2Value":"99+ 种语言","stat2":"Whisper 模型所支持","stat3Value":"实时","stat3":"在 Apple Silicon 和现代 x86 CPU 上","p3":"硬件加速带来了重大改变。whisper.cpp 在 Mac 上支持 ARM NEON、Apple 的 Accelerate 框架和 Metal GPU，在 NVIDIA GPU 上支持 CUDA，支持用于跨平台 GPU 推理的 Vulkan，甚至还为昇腾（Ascend）NPU 和 Intel 的 OpenVINO 提供了专门的后端。与 llama.cpp 一同开发的 GGUF 模型格式实现了高效量化——在准确率损失极小的前提下，将模型的内存占用减少 50% 至 75%。","p4":"具体到听写场景——录音通常只有 5 到 30 秒——在过去三年内生产的任何机器上，本地推理几乎都是瞬时完成的。「small」版 Whisper 模型（466 MB，约 852 MB 内存）对大多数语言都能提供出色的准确率，在 8 GB 内存的机器上也能轻松运行。你不需要高端 GPU，不需要游戏 PC，只需要一台还算现代的笔记本。","p5":"Apple 很早就看准了这一趋势。其内置于 iOS 和 macOS 的端侧语音识别，随每次系统更新都大幅改进——支持数十种语言的无网联听写。Apple 这样一家拥有庞大云端基础设施的公司都在把语音识别转向设备端，这本身就足以说明这项技术的走向。"},"regulatory":{"title":"监管格局","p1":"全球各地的隐私法规都在追赶一个现实：语音数据是敏感的。对于处理语音数据的组织而言，本地处理不仅是一种隐私偏好——它正日益成为一种合规优势。","gdpr":{"title":"GDPR（欧盟）","a":"根据《通用数据保护条例》，当语音数据被用于唯一识别某人时，它被归类为","bold1":"生物识别数据","b":"——从而被纳入个人数据的「特殊类别」，受","link":"第 9 条","c":"管辖。处理生物识别数据需要明确同意，或满足一小套法律依据之一。即便不用于识别身份，语音录音也属于个人数据，须遵循 GDPR 的数据最小化原则。本地处理可彻底规避诸多 GDPR 义务——如果数据从不离开设备，就不存在数据传输、第三方处理或跨境合规问题。"},"aiAct":{"title":"《欧盟 AI 法案》","a":"《欧盟 AI 法案》于 2024 年 8 月生效，相关条款将在 2026 年前分阶段实施。该法案把为识别目的而处理生物识别数据的 AI 系统归类为","bold1":"高风险","link":"《欧盟 AI 法案》","b":"。高风险系统面临大量要求，包括合规性评估、文档义务和持续监控。在公共场所进行实时生物识别在极少数例外情形之外被彻底禁止。虽然标准的语音转文字本身未必会触发高风险归类，但任何可能从声音识别说话人的系统——哪怕是无意为之——都将在这一框架下受到审视。"},"hipaa":{"title":"HIPAA 与职业特权","a":"在医疗场景中，包含患者信息的语音录音在美国受","b":"保护。将此类录音发往云服务需要签订《业务伙伴协议》（BAA）并遵守《安全规则》。同样，律师在口述涉及客户事务的笔记时，还须顾及","bold2":"律师-当事人特权","c":"——把受特权保护的通信发往第三方服务器会带来风险。本地处理可绕开这些问题：如果音频从不离开设备，就没有需要担心的第三方数据处理方。"},"ccpa":{"title":"CCPA / CPRA（加州）","a":"《加州消费者隐私法》（经《加州隐私权法案》修订）把语音和音频数据视为受监管的个人信息，并在数据被用于生物识别时附加额外规则","link1":"加州总检察长","link2":"CPPA 常见问题","b":"。收集语音数据的企业必须提供额外披露，并尊重消费者的删除权和退出权。美国还有多个其他州——包括伊利诺伊州（BIPA）、得克萨斯州和华盛顿州——也已制定各自的生物识别隐私法，其要求和执法机制各不相同。"},"p2":"监管方向很明确：全球范围内，语音数据正受到越来越严肃的对待。对于处理语音数据的组织——无论是在医疗、法律、金融还是任何受监管行业——本地处理在结构上提供了更简洁的合规姿态。没有需要审计的数据传输，没有需要审查的分包处理方，也没有需要论证正当性的跨境数据流动。"},"cloudSense":{"title":"何时云端仍然合理","p1":"实事求是很重要。本地 AI 并非总是正确的选择，假装不是这样反而会削弱上文那些站得住脚的论点。以下是云端处理仍是更优选择的几种情形：","case1":{"title":"超大规模实时流式处理","body":"呼叫中心需要同时处理数千路并发音频流，这离不开云端基础设施。没有任何单台设备能扛得住这样的吞吐量。"},"case2":{"title":"大规模说话人分离","body":"在多人录音中识别「谁说了什么」在计算上代价高昂。Google 和 AWS 的云端 API 在这方面，目前比大多数本地方案处理得更可靠。"},"case3":{"title":"小众语言","body":"虽然 Whisper 和 Parakeet 支持多种语言，但准确率参差不齐。对于 Whisper 处理欠佳的特定语言或方言，专业云服务商可能提供更好的模型。"},"case4":{"title":"在有限硬件上处理超长录音","body":"在低端笔记本上转录一段 4 小时的会议录音可能慢到令人抓狂。云端 API 几分钟就能处理完数小时的音频。"},"p2":"理想的做法往往是混合式：默认本地，按需云端。日常听写在本地处理，敏感内容留在设备上。而当你确实需要云端级算力或专业功能时，让它成为一个有意识的决定——而非一个看不见的默认值。"},"future":{"title":"本地 AI 的未来","p1":"硬件的发展轨迹强烈利好本地 AI。现代芯片正被专门设计来高效运行神经网络：","npu1":{"title":"Apple Neural Engine","body":"Apple 的专用神经处理单元，存在于每一颗 M 系列和 A 系列芯片中，在 M4 上可达每秒 38 万亿次运算（TOPS）。Apple Intelligence 高度依赖这一硬件来进行端侧处理。"},"npu2":{"title":"Qualcomm Hexagon NPU","body":"自 2024 年起为 Windows 笔记本提供动力的 Snapdragon X Elite 系列，配备了一颗 45 TOPS 的 NPU，专为本地 AI 工作负载而设计。Microsoft 的 Copilot+ PC 计划要求设备具备 NPU 能力。"},"npu3":{"title":"Intel AI Boost NPU","body":"Intel 的 Meteor Lake Core Ultra 处理器引入了约 11 TOPS 的专用 NPU，而更新的 Lunar Lake 一代则达到 48 TOPS。Intel 明确将这些定位为支撑「AI PC」体验的能力。"},"p2a":"在模型方面，蒸馏、剪枝和高效架构方面的研究仍在不断拓展本地可运行的边界。诸如","p2b":"之类的项目证明，专门打造的小模型能以一小部分算力成本达到大模型 99% 的准确率。","p3":"这种汇聚趋势不容错辨：每一家主要芯片厂商都在加入专用 AI 硬件，每一家主要操作系统厂商都在构建端侧 AI 功能，每一家主要模型实验室都在发布更小、更高效的模型。端侧 AI 处理不是一种替代方案——它正在成为默认选项。云端处理对于重型工作负载仍将举足轻重，但对于听写这类个人计算任务，未来属于本地。"},"actions":{"title":"你今天就能做的事","intro":"你不必坐等未来。本地 AI 当下就切实可行。以下是你可以采取的具体步骤：","step1":{"title":"用本地语音识别来听写","a":"提供由 OpenAI Whisper 和 NVIDIA Parakeet 驱动的按键说话听写，并配有完整的图形界面。如果你更喜欢命令行，","b":"也可直接使用。两者都完全在你的机器上运行。"},"step2":{"title":"在本地运行 LLM 处理敏感工作","a":"诸如","b":"和","c":"等工具，让你能在本地运行有实用价值的语言模型。对于涉及个人、财务或机密信息的查询，本地模型可彻底消除数据暴露的风险。"},"step3":{"title":"审视你现有的工具","body":"检查你所使用的每一项语音服务的隐私政策。你的听写应用会把音频发往服务器吗？你的虚拟助手会存储录音吗？你的会议转录工具会用你的内容做训练吗？答案也许会让你大吃一惊。"},"step4":{"title":"优先选择提供本地处理的应用","body":"在挑选新工具时，优先选择那些提供端侧处理的产品。这种能力已经存在——去要求它。每一位选择本地处理的用户，都在向市场发出一个信号：隐私很重要。"}},"sources":{"title":"参考来源与延伸阅读","s1":"OpenAI Whisper 的 C/C++ 移植版。涵盖模型尺寸、硬件要求和支持的平台。","s2":"原始 ASR 模型发布、训练方法和评估背景。","s3Title":"NVIDIA Parakeet 模型卡","s3":"用于本地语音识别的开放 ASR 指标和基准测试细节。","s4":"C/C++ 的 LLM 推理。在数十种模型架构上支持 1.5 位至 8 位量化。","s5Title":"Google Cloud Speech-to-Text 数据日志记录","s5":"默认不记录政策及选择加入数据计划的细节。","s6Title":"AWS Transcribe 安全文档","s6":"数据处理、退出政策和加密实践。","s7Title":"《2024 年 IBM 数据泄露成本报告》","s7":"全球泄露成本、平均暴露记录数及行业分析。","s8Title":"GDPR 第 9 条","s8":"特殊类别个人数据（含生物识别数据）的处理。","s9Title":"《欧盟 AI 法案》","s9":"将处理生物识别数据的 AI 系统归类为高风险。","s10":"针对个人及敏感数据处理的加州隐私指引。","s11Title":"Azure Speech 数据隐私","s11":"Microsoft 服务层级的语音数据处理文档。","s12":"端侧处理与 Private Cloud Compute 架构。","s13Title":"用于抑郁检测的声音生物标志物","s13":"Low、Bentley、Ghosh（2020）。利用语音对精神疾病进行自动化评估。","s14Title":"EFF 关于 FISA 与《国家安全信函》","s14":"政府对美国公司所存储数据的调取。"},"cta":{"title":"你的声音，理应属于你自己","body":"OpenWhispr 默认在本地处理一切。除非你明确选择云端处理，否则没有任何音频会离开你的设备。","subtext":"开源。由 OpenAI Whisper 和 NVIDIA Parakeet 驱动。支持 macOS、Windows、Linux 和 iOS。","starButton":"在 GitHub 上给我们点 Star","footnote":"无需账户 · 离线可用 · 永久开源"},"links":{"openSource":"开源为何重要","localVsCloud":"本地转录 vs 云端转录","allArticles":"全部文章","offline":"离线使用场景","medical":"医疗听写"}},"blogDiarization":{"metadata":{"title":"说话人分离：100% 本地且私密 | OpenWhispr","description":"了解 OpenWhispr 如何用 pyannote、CAM++ 嵌入和 sherpa-onnx 在本地进行说话人分离。隐私优先、开源且免费。","keywords":["本地说话人分离","设备端说话人分离","隐私会议转录","隐私会议记录工具","本地会议笔记","离线会议转录","开源会议助手","开源会议记录工具","开源 AI 会议笔记","音频永不离开设备","隐私 AI 会议助手","自托管会议转录","说话人识别隐私","pyannote 分割","cam++ 说话人嵌入","sherpa-onnx 分离","本地声纹","谁说了什么转录","Otter.ai 隐私替代","Otter.ai 开源替代","Granola 开源替代","Fireflies 设备端替代"]},"schema":{"title":"本地说话人分离：我们如何打造 100% 隐私的会议记录工具","description":"完整技术解析 OpenWhispr 的本地说话人分离——pyannote、CAM++、sherpa-onnx。你的音频永不离开设备。"},"breadcrumb":{"home":"首页","blog":"博客","article":"本地说话人分离"},"hero":{"title":"本地说话人分离：我们如何打造 100% 隐私的会议记录工具","subtitle":"OpenWhispr 是一款音频永不离开设备的开源会议助手。每一段声纹、每一个说话人标签、每一个嵌入向量——全部在你自己的机器上处理和存储。","date":"2026 年 4 月 15 日","tag":"技术"},"sections":[{"id":"what-is-diarization","title":"什么是说话人分离"},{"id":"pipeline","title":"四阶段流程"},{"id":"vad","title":"第一阶段：语音活动检测"},{"id":"segmentation","title":"第二阶段：分割"},{"id":"embeddings","title":"第三阶段：说话人嵌入"},{"id":"clustering","title":"第四阶段：聚类"},{"id":"live","title":"实时分离"},{"id":"profiles","title":"声纹档案"},{"id":"privacy","title":"隐私"},{"id":"limitations","title":"坦诚的局限"},{"id":"why-sherpa","title":"为什么选择 Sherpa-ONNX"},{"id":"faq","title":"常见问题"}],"opening":{"lead":"本地说话人分离是指为会议录音标注“谁在何时说话”的过程，全程在用户设备上运行，不向任何外部服务器传输音频、嵌入向量或转录文本。","body":"会议转录有一个“谁说了什么”的难题。Zoom 只给你一整面文字墙。大多数 AI 记录工具——Otter.ai、Fireflies、Read、Granola——的解决办法都是把你的会议音频上传到它们的服务器，在那里运行说话人模型，再把带标签的片段发回来。你的声音，以及你同事的声音，最终都变成了别人数据库里的向量。我们不想要这样，我们的用户也不想要。OpenWhispr 是一款建立在相反假设之上的开源会议记录工具：音频永不离开设备，说话人声纹存放在你磁盘上的本地 SQLite 文件中，除非你明确选择，否则会议的任何内容都不会被上传。本文将从技术层面解析这一承诺中的分离部分——它是什么、我们选了哪些模型、它们如何协同工作，以及每个步骤中每一字节数据究竟存放在哪里。","footnote":"最后更新：2026 年 4 月 15 日。实现细节参考的是 2026 年 4 月 14 日合并进 OpenWhispr 桌面应用的说话人分离功能。下文的每一项技术声明都来自一手参考资料，或可直接在开源仓库中查验。"},"factCheck":{"title":"事实核查快照（双重来源）","items":[{"lead":"说话人分离回答的是“谁在何时说话”。","body":"它的评估基准是 NIST Rich Transcription 等社区评测以及 pyannote.audio 参考流程。","sources":[{"href":"https://www.nist.gov/itl/iad/mig/rich-transcription-evaluation","label":"NIST RT 评测"},{"href":"https://arxiv.org/abs/1911.01255","label":"pyannote.audio 论文（Bredin 等，2020）"}]},{"lead":"我们使用 pyannote-segmentation-3.0，以及来自 3D-Speaker 项目的 CAM++。","body":"两者都是开源的，都可移植为 ONNX，都直接从 sherpa-onnx 模型发布页下载。","sources":[{"href":"https://huggingface.co/pyannote/segmentation-3.0","label":"pyannote-segmentation-3.0 模型卡"},{"href":"https://arxiv.org/abs/2303.00332","label":"CAM++ / 3D-Speaker 论文（Wang 等，2023）"}]},{"lead":"所有推理都通过 sherpa-onnx 运行——一个原生 ONNX 二进制文件，无需 Python 运行时、无需 PyTorch、无需 GPU。","body":"我们在构建时锁定该二进制文件及其依赖的版本。","sources":[{"href":"https://github.com/k2-fsa/sherpa-onnx","label":"sherpa-onnx（GitHub）"},{"href":"https://github.com/k2-fsa/sherpa-onnx/releases","label":"sherpa-onnx 发布页"}]},{"lead":"声纹是 512 维的 float32 向量，作为本地 SQLite BLOB 存储，","body":"每位说话人约 2KB。它们永不离开设备。","sources":[{"href":"https://www.sqlite.org/datatype3.html","label":"SQLite 数据类型文档"},{"href":"https://onnxruntime.ai/","label":"ONNX Runtime"}]},{"lead":"分离过程中不会向设备外传输任何音频、嵌入向量或转录文本。","body":"整个功能唯一的网络活动是首次启动时的一次性模型下载，你可以亲自在开源仓库中查验。","sources":[{"href":"https://openwhispr.com/privacy","label":"OpenWhispr 隐私政策"},{"href":"https://github.com/OpenWhispr","label":"OpenWhispr（GitHub）"}]}]},"whatIs":{"title":"说话人分离究竟是什么","lead":"说话人分离是为录音标注“谁在何时说话”的过程——而不是他们说了什么。","paragraph1":"转录回答“说了什么”；分离回答“谁在说”。两者相辅相成：一款好的会议记录工具两者都需要，而让它们协同工作比单独搞定任何一个都要难。","paragraph2":"分离之所以真正困难，源于几个顽固的原因。同一通话里两个人的声音可能重叠。背景噪声会渗进来。简短的发言（“嗯”“好的”）几乎不携带说话人信息。新的说话人会在会议进行到一半时出现。而且与转录不同，分离必须做出全局决策——它无法在不与第二段对比的情况下为第七段贴标签。正是这种全局语境，使得大多数商业方案选择在服务器端针对完整录音做分离，再把带标签的转录发回来。","paragraph3":"仅有转录是不够的。一场十人参加、四十五分钟充满交叉对话的会议，如果呈现为一面无差别的文字墙，根本无法阅读。会议笔记——人们真正想要的成果——需要说话人归属，因为待办事项、决定和承诺都是挂在具体的人身上的。“Alice 会在周五前交付迁移”是一条笔记。“会在周五前交付迁移”是噪声。","calloutTitle":"分离不等于说话人识别","calloutBody":"分离说的是“这三段是说话人 A；这两段是说话人 B”。它并不知道说话人 A 是谁。说话人识别则通过把声纹与已存档案匹配，给说话人 A 安上一个名字。OpenWhispr 两者都做——分离在每场会议上运行，而一旦你为某人标注了名字，识别就开始介入。两者都保持在本地。"},"pipeline":{"title":"四阶段本地流程","lead":"OpenWhispr 的说话人分离流程分为四个阶段：语音活动检测、分割、说话人嵌入和聚类——随后还有一个合并步骤，把说话人片段与转录文本对齐。","paragraph1":"每个阶段都是一个独立的 ONNX 模型或原生二进制文件。无需 Python 运行时、无需 PyTorch、无需 CUDA。所有模型在磁盘上的总占用约为 45MB。","paragraph2":"音频采集从第一天起就是双流的。麦克风按定义就是“你”，所以我们从不浪费算力去分离你自己的声音——每一段麦克风片段都按来源标记为“你”，而非靠声音匹配。系统音频（通话中的其他所有人）才是唯一真正需要做说话人切分的东西，这一下就把工作量减半了。","paragraph3":"音频采集完成后，四个阶段依次运行。VAD 去除静音，让昂贵的嵌入阶段不会把算力浪费在空帧上。分割把连续语音切成单说话人的片段。嵌入模型把每个片段转换为一个 512 维的声纹。凝聚聚类把相似的声纹归为一组，得出最终的一组说话人 ID。最后一道工序按时间戳重叠把这些 ID 合并回转录文本，于是每个词都被打上了一个说话人标签。","paragraph4Prefix":"整套流程在一个从应用派生出的子进程内运行——没有服务器，没有监听端口，也没有任何指向 Electron 之外的 IPC。你可以通过阅读开源仓库中的","paragraph4Suffix":"来确认这一点。"},"vad":{"title":"第一阶段：语音活动检测（Silero）","lead":"语音活动检测（VAD）是在昂贵阶段运行之前，把语音和静音区分开的廉价过滤器。","paragraph1":"对静音运行说话人嵌入模型既浪费 CPU 又会产出无用的向量。VAD 正是让流程其余部分保持快速的关键。","paragraph2Prefix":"我们使用","sileroLinkLabel":"Silero VAD","paragraph2Suffix":"：一个 2MB 的 ONNX 模型，采用 MIT 许可，实际上已成为开源行业的默认选择。Silero 会针对每一帧返回当前 32 毫秒窗口包含语音的概率。在现代 CPU 上，它的开销微不足道——每个 32 毫秒片段约 0.1 毫秒，大约相当于单核性能的千分之三。","paragraph3":"实时流程在通话期间对系统音频持续运行 Silero。我们实际发布时采用的阈值是针对真实会议音频调校的，而非干净的基准测试：","list":[{"label":"窗口大小：","value":"512 个采样点（16kHz 下为 32 毫秒）"},{"label":"语音阈值：","value":"0.15——刻意设得激进，以免漏掉音量低或距离远的说话人"},{"label":"静音阈值：","value":"0.08"},{"label":"片段结束条件：","value":"连续 16 个静音窗口（约 512 毫秒）"},{"label":"嵌入所需最小片段：","value":"0.8 秒"},{"label":"实时识别节奏：","value":"一旦累积到 ≥1.6 秒的语音，每 1 秒一次"}],"calloutTitle":"坦诚的取舍","calloutBody":"0.15 这一激进的语音阈值能捕捉到远处麦克风上的轻声说话者，但偶尔也会被响亮的键盘敲击声或房间噪声误触发。0.8 秒的最小片段过滤器会在这些误报到达嵌入阶段之前剔除其中大部分，从而保护了下游的准确率。"},"segmentation":{"title":"第二阶段：分割（pyannote 3.0）","lead":"分割负责把连续语音切成各自只包含一个说话人的片段，同时检测说话人切换点和重叠区域。","paragraph1":"它是决定说话人边界在哪里的阶段。之后的所有环节都假定这些边界是正确的。","paragraph2Prefix":"我们使用","paragraph2Mid":"，这是 pyannote.audio 团队该模型的当前版本，由","paragraph2Suffix":"项目导出为 ONNX。Pyannote 是事实上的学术基准：pyannote.audio 流程在 AMI 和 CALLHOME 等标准基准上的分离错误率处于 12–15% 区间——正是商业云服务商爱引用的那组数字。","paragraph3Prefix":"为什么偏偏要 ONNX 移植版？因为在 Electron 应用里塞进 PyTorch 根本行不通——光是运行时就约 1.5GB，而且实际上要有 GPU 才够快。ONNX 导出版是一个 6.6MB 的单文件，通过","paragraph3Suffix":"在 CPU 上运行。同样的模型，重量却只有一小部分。","paragraph4Prefix":"我们通过 sherpa-onnx 的离线分离二进制文件配上几个参数来调用它。最小时长设置取自 pyannote 论文推荐的默认值：短促的语音爆发会被并入相邻的静音，而短暂的静音不会打断一个连续的说话人片段。输出是一组带临时说话人 ID 标记的起止时间对（","paragraph4Suffix":"等等）。"},"embeddings":{"title":"第三阶段：说话人嵌入（CAM++）","lead":"说话人嵌入是一串定长的数字，它代表一段声音，就像哈希值代表一个文件。","paragraph1":"同一声音的两段录音会产生在向量空间中相近的嵌入；不同的声音则产生彼此相距很远的嵌入。“相近”用余弦相似度来衡量——一个介于 -1 和 1 之间的单一数值。这是流程中其余一切的数学核心。","paragraph2Prefix":"我们的嵌入模型是来自阿里巴巴达摩院 3D-Speaker 项目的 CAM++，在","paragraph2Mid":"说话人验证基准上训练。我们实际采用的具体文件是","paragraph2Mid2":"，一个约 28MB 的 ONNX 导出版，为每个片段产出一个 512 维的 float32 向量。论文是","camppPaperLabel":"arXiv:2303.00332（Wang 等，2023）","paragraph2Suffix":"。","paragraph3Prefix":"我们选择 CAM++ 而非那个显而易见的替代——","paragraph3Suffix":"——出于三个具体原因：参数量大约只有一半、在 VoxCeleb1-O 上的等错误率更低、CPU 推理更快。NVIDIA NeMo 的 MSDD 模型没有官方 ONNX 导出，且严重依赖 PyTorch。Picovoice Falcon 带有按席位收费的商业许可。Apple 的 CoreML Speech 框架仅限 macOS。CAM++ 是唯一满足我们全部约束的选项——跨平台、ONNX、开放许可、CPU 上速度快。","paragraph4Prefix":"为了给 CAM++ 喂数据，我们在","paragraph4Mid":"中自带了一个 log-mel 滤波器组特征提取器：80 个梅尔频带、25 毫秒窗口、10 毫秒帧移，都是这类模型的标准参数。特征通过","paragraph4Suffix":"流入 ONNX 模型，在 Electron 主进程中运行。另一端出来的是一个 512 项的 Float32Array，我们把它作为 2048 字节的 BLOB 写入 SQLite。","paragraph5":"匹配步骤本身是最简单不过的事情：两个向量之间的余弦相似度。下面就是决定两个会议片段是否为同一说话人的实际代码：","calloutTitle":"为什么是 512 维？","calloutBody":"512 是当前 VoxCeleb 说话人验证排行榜上准确率与存储成本之间的最佳平衡点。256 维在区分相近但不同的声音时开始丧失辨别力；1024 维则增加了存储和算力，回报却递减。512 个 float32 × 2KB × 一千场各有十名说话人的会议，磁盘占用约 20MB。这在一个本地 SQLite 文件里绰绰有余。"},"clustering":{"title":"第四阶段：聚类——“有多少个说话人？”的难题","lead":"对每个片段做完嵌入后，你手上有 N 个向量。聚类把它们归为 K 个说话人——但难点在于你通常事先并不知道 K 是多少。","paragraph1":"一通销售电话可能有 2 个说话人。一场全员大会可能有 8 个。固定 K 值的聚类在两端都会失灵。","paragraph2":"我们使用余弦相似度阈值为 0.5 的凝聚聚类：先把每个片段各自当作一个簇，合并相似度最高的两个簇，重复进行，直到没有任何一对的相似度超过阈值为止。决定何时停止的是阈值，而非某个目标数量。这样就能从两个说话人到十个自然扩展，无需任何配置。","paragraph3Prefix":"实际的调用位于","paragraph3Suffix":"。它派生 sherpa-onnx 二进制文件，附带几个命令行参数：","paragraph4Prefix":"输出是纯文本：每个片段一行，格式为","paragraph4Mid":"。我们把它解析成一个对象数组，然后遍历转录文本，把每个转录片段匹配到时间重叠最大的分离片段。来自麦克风的片段始终被标记为","paragraph4Suffix":"——来源永远压过声音。","paragraph5Prefix":"0.5 这个阈值不是凭空猜的。我们在真实会议录音上、用自己的评测框架（","paragraph5Suffix":"）对它进行了调校。阈值过高会欠聚类——一个说话人被拆成两个。阈值过低会过聚类——两个相似的声音被合并。0.5 是我们最终收敛到的最佳平衡点。"},"live":{"title":"实时分离：边说边标注","lead":"纯批处理分离很准确，但在通话过程中是看不见的。纯实时则即时却嘈杂。OpenWhispr 两者都跑：实时标签在会议进行中出现，会后再用一次完整的批处理来优化它们。","paragraph1Prefix":"实时路径位于","paragraph1Mid":"。它以 16kHz 接入系统音频流，把每个 32 毫秒帧送进 Silero VAD，并不断累积语音帧，直到某个片段闭合（在约 512 毫秒的静音之后）或达到实时识别节奏（一旦累积到 ≥1.6 秒，每 1 秒一次）。一旦触发，它会提取一个 CAM++ 嵌入，将其与内存中当前通话嵌入及已存说话人档案的映射进行比对，并触发一个 IPC 事件——","paragraph1Suffix":"——附带最佳匹配。React 端接收到后，原地更新转录气泡。","paragraph2Prefix":"一旦有人在界面中手动设定了某个说话人的名字，这个指派就被","lockedWord":"锁定","paragraph2Mid":"。会后的批处理不得覆盖它，哪怕它自己的分析结果有异议。这一锁定行为位于","paragraph2Suffix":"，而它很重要：最糟糕的用户体验，就是让一次自动处理悄悄覆盖掉用户的修正。","paragraph3":"录制停止后，我们会对完整的 WAV 文件运行完整的离线分离。批处理拥有完整语境——它看到的是整场会议，而不只是最后的 1.6 秒——因此能产出更干净的最终标注。批处理结果会与实时标签对齐，尊重用户的每一处锁定，并用更准确的分组替换掉临时的猜测。","calloutTitle":"为什么混合模式胜过单独任何一种","calloutBody":"纯实时很快，但在短片段上很嘈杂。纯批处理很准确，却要等到会议结束才出声。混合模式让你两者兼得——通话中即时反馈，以及一份反映完整语境分析的最终转录。代价只是多一步对齐，它在后台运行一次，实际上几乎不耗成本。"},"profiles":{"title":"声纹档案：跨会议记住说话人","lead":"一旦你在某场会议里把某位说话人标注为“Alice”，OpenWhispr 就会在之后的每一场会议中自动为她标注——你无需再做任何事，而且她的声纹永远不会离开你的设备。","paragraph1":"机制很直接：每位说话人的质心嵌入都会存入你的本地 SQLite 数据库。当未来某通话出现新的嵌入时，我们用余弦相似度把它与每一个已存档案比对。相似度超过阈值的匹配会自动指派名字。这个过程完全不涉及服务器。支撑它的三张表都是普通的 SQLite：","matchingIntro":"匹配分三档，以在置信度和误报之间取得平衡：","matchingTiers":[{"label":"余弦相似度 ≥ 0.70：","body":"自动确认。标签立即出现，无需任何提示。"},{"label":"0.55 ≤ 余弦相似度 < 0.70：","body":"建议。界面会显示“这是 Alice 吗？”并附带确认 / 忽略按钮，等待你的输入。"},{"label":"余弦相似度 < 0.55：","body":"保持匿名。该片段维持为“说话人 N”，直到你手动为其命名。"}],"paragraph2Prefix":"当你确认一个匹配时，已存的质心会通过滑动平均来更新，而非直接覆盖。公式很简单：","paragraph2Suffix":"。这能应对声音漂移——一个寒冷的早晨、一支不同的麦克风、一根坏掉的线缆——而不会破坏未来的匹配。系统见过 Alice 声音的样本越多，她的档案就越稳定。","paragraph3Prefix":"新档案还会触发","retroactiveWord":"追溯重新标注","paragraph3Suffix":"。当你第一次为某位说话人命名时，一个后台任务会遍历过去每场会议已存的嵌入，找出匹配，并更新历史转录中此前未命名的说话人。被用户锁定的映射永远不会被触碰。结果是：为 Alice 命名一次，就能追溯地在她参加过的每一通话中为她标注。","paragraph4Prefix":"如果连接了 Google Calendar 且会议有参会者，说话人选择器界面会预先填入他们的姓名和邮箱。一键即可把","speakerTwo":"说话人 2","paragraph4Suffix":"映射好并永久保存——参会者与声纹档案保持关联，因此即使换了设备，正确的名字也会出现。"},"privacy":{"title":"你的数据究竟存放在哪里","lead":"音频永不离开设备。这不是一句营销口号——它是代码的机械属性，而且由于 OpenWhispr 是开源的，你可以亲自验证。","intro":"下面就是每一类数据在每个步骤究竟会发生什么。","tableHeaderWhat":"数据","tableHeaderWhere":"存放位置","tableHeaderLeaves":"是否离开设备？","tableRows":[{"what":"原始会议音频","where":"操作系统临时目录，分离完成后删除","leaves":"否——永不","leavesEmphasis":true},{"what":"分离用 ONNX 模型","where":"~/.cache/openwhispr/diarization-models/","leaves":"首次启动时下载一次，之后再不下载"},{"what":"说话人嵌入（声纹）","where":"本地 SQLite BLOB","leaves":"否——永不","leavesEmphasis":true},{"what":"说话人到姓名的映射","where":"本地 SQLite 表","leaves":"否——永不","leavesEmphasis":true},{"what":"转录文本","where":"本地 SQLite（如你开启，可选云同步）","leaves":"仅在你主动选择时"}],"paragraph1Prefix":"会议的原始 PCM 音频会被写入你操作系统临时目录中的一个文件，交给本地派生的 sherpa-onnx 二进制文件，并在分离完成时于一个","paragraph1Mid":"块中删除。OpenWhispr 的分离流程中不存在任何打开网络套接字上传音频的代码路径。由于代码是开源的，你可以在","paragraph1Mid2":"、","paragraph1Mid3":"中 grep","paragraph1Mid4":"或","paragraph1Suffix":"，亲自确认这一点。","paragraph2Prefix":"Electron 主进程与渲染进程之间的 IPC 使用内置的","paragraph2Suffix":"——它是单个进程内的内核级管道，而非网络套接字。说话人嵌入和声纹档案作为 SQLite BLOB，与你的笔记存储在同一个数据库文件中，享有相同的操作系统级文件权限。","calloutTitle":"一个例外，开诚布公","calloutBody":"这个功能唯一的网络活动，是首次启动时的一次性模型下载——从 sherpa-onnx 的 GitHub 发布页 CDN 获取三个 ONNX 文件。我们明确说明这一点，是因为“安装时一次网络调用、之后零调用”与“始终在线的云端推理”在隐私层面是截然不同的两个故事；而作为一款开源会议助手，我们宁愿主动告诉你，也不愿让你自己去发现。"},"limitations":{"title":"坦诚的局限","lead":"本地分离很好，但并不完美。下面是它确实力不从心的几种情况，以及我们针对每一种所做的应对。","items":[{"label":"说话人重叠。","body":"两个人同时讲话，对单一标签归属来说本质上是有损的。Pyannote 3.0 能检测出重叠区域，但每个片段只有一个 speaker_id，不足以表示“两个人同时在说话”。我们会把这些片段标为临时，让你在界面中加以修正。"},{"label":"冷启动准确率。","body":"与一位新同事的第一场会议只能依靠 VoxCeleb 训练出的通用特征。一旦你为他标注过一次、档案积累了几个样本，后续会议就会明显更准。"},{"label":"简短发言。","body":"“嗯”、“好的”、一声短促的笑——任何短于约 0.8 秒的内容都无法产生可靠的嵌入。我们会把这些片段从嵌入阶段剔除，转而依据周围上下文做标签传播。"},{"label":"远场或低信噪比音频。","body":"在嘈杂房间里用免提电话的人，质量会优雅但明显地下降。自动增益控制和 VAD 调校会有所帮助，但再精巧的模型也无法完全克服糟糕的物理条件。"},{"label":"仅 CPU 推理。","body":"GPU 会更快，但我们刻意不要求它。实际上，现代 CPU 的吞吐量已经足够：每个嵌入约 100 毫秒，在 M1 Mac 上对一场 45 分钟的会议做批处理分离约需 30 秒。"}],"outroPrefix":"我们在开源仓库中公开了自己的评测框架（","outroSuffix":"），这样你就能用自己的录音亲自测量上述任何一项——而不是我们精挑细选的演示片段。这比营销材料里一个光鲜的数字要诚实得多。"},"whySherpa":{"title":"我们为什么选择 Sherpa-ONNX","lead":"我们需要一个跨平台、无需 Python、CPU 上速度快且开源的分离运行时。Sherpa-ONNX 是唯一四项全中的选项。","paragraph1":"我们的要求很具体。它必须能在 macOS（Intel 和 Apple Silicon 都要）、Windows x64 和 Linux x64 上运行，且无需分别维护不同的代码路径。它不能捆绑 PyTorch——光是运行时就约 1.5GB，而且要在交互级延迟下可用基本上得有 GPU。它不能要求 CUDA。而且许可证要足够宽松，能在一款“商用但开源”的桌面应用中发布。","evaluatedIntro":"以下是我们实际评估过的选项：","evaluated":[{"label":"pyannote.audio（Python）：","body":"排除。在 Electron 应用里捆绑 PyTorch 根本行不通。"},{"label":"NVIDIA NeMo MSDD：","body":"依赖 PyTorch，无官方 ONNX 导出，以 GPU 为中心。"},{"label":"Picovoice Falcon：","body":"按活跃用户收费的商业许可，外加首次运行时需云端激活。"},{"label":"Apple CoreML + Speech 框架：","body":"仅限 macOS，没有跨平台对等性。"},{"label":"sherpa-onnx","body":"Apache 2.0 许可，原生 ONNX 二进制文件，约 35MB 的模型，派生进程读取 stdout 的接口，由 k2-fsa 积极维护。同一套工具链也支撑着其他已发布产品中的生产级 ASR。"}],"outroPrefix":"一个坦诚的提醒：sherpa-onnx 的离线说话人分离二进制文件相对较新。我们将版本锁定在","outroSuffix":"，并在升级锁定版本之前，针对每次更新运行我们完整的评测框架。对于这样一个承重的部件，这是正确的取舍。"},"faqHeading":"常见问题","faqs":[{"question":"有没有一款不会把我的音频上传到云端的隐私会议记录工具？","answer":"有。OpenWhispr 是一款免费、开源的会议记录工具，你的音频永不离开设备。转录在本地用 OpenAI Whisper 或 NVIDIA Parakeet 运行，说话人分离在本地用 pyannote 分割和 CAM++ 嵌入运行，声纹则存储在你自己机器上的本地 SQLite 文件中——而不是我们的服务器上。"},{"question":"Otter.ai、Fireflies 或 Granola 能看到我的会议吗？","answer":"能。基于云的 AI 记录工具会把你的会议音频上传到它们的服务器，以便转录和分离。说话人标签、转录文本和声音嵌入都在它们的基础设施中计算，并存储在它们的数据库里。如果你需要一个能把每一字节音频都留在设备上的替代方案，你需要的是 OpenWhispr 这样的本地优先工具。"},{"question":"什么是说话人分离？","answer":"说话人分离是为录音标注“谁在何时说话”的过程，与回答“说了什么”的转录有所区别。一份完成分离的转录会为每段发言打上说话人标签——Alice、Bob、说话人 3——而不是一整面无差别的文字墙。"},{"question":"本地说话人分离真的管用吗，还是说它是云端的降级版？","answer":"它管用。OpenWhispr 的流程——pyannote-segmentation-3.0 加上来自 3D-Speaker 项目的 CAM++ 嵌入，通过 sherpa-onnx 运行——在标准基准上的分离错误率与商业云 API 处于同一水平。主要的坦诚取舍在于对重叠说话人的处理，以及对系统从未听过的声音的冷启动准确率。"},{"question":"本地分离模型占多少磁盘空间？","answer":"总共约 45MB：pyannote 分割（6.6MB）加上 CAM++ 说话人嵌入模型（28MB）再加 Silero VAD（2MB）。这三者会在首次启动时下载一次，并缓存到 ~/.cache/openwhispr/diarization-models/。此后它们就留在磁盘上，再也不会重新下载。"},{"question":"运行 OpenWhispr 的本地分离需要 GPU 吗？","answer":"不需要。一切都通过 ONNX Runtime 在 CPU 上运行。现代 Apple Silicon 和较新的 x86 CPU 都能轻松应对——实际上，在 M1 Mac 上对一场 45 分钟的会议做批处理分离约需 30 秒，而通话过程中实时说话人标签会在一两秒内出现。"},{"question":"OpenWhispr 真的是开源的吗？","answer":"是的。分离代码、转录流程以及桌面应用的其余部分都在 GitHub 上开源。本文中的隐私声明可直接验证：在分离辅助代码里 grep fetch、axios 或 http，你在分离路径中找不到任何网络调用。"},{"question":"通话中两个人同时讲话会怎样？","answer":"Pyannote 3.0 会明确检测并标注重叠区域，但重叠期间的单一标签归属本质上是有损的。我们会把这些片段在转录中标为临时，让你通过点击说话人标签来修正。你的手动修正会被锁定，永远不会被之后的自动处理覆盖。"}],"cta":{"title":"试试这款隐私会议记录工具","body":"OpenWhispr 是一款免费、开源的会议助手，你的音频永不离开设备。本地分离、本地转录、本地声纹档案。支持 macOS、Windows 和 Linux。","starButton":"在 GitHub 上给我们点 Star","footnote":"无需账号 · 离线可用 · 永久开源"},"internalLinks":[{"href":"/blog/local-vs-cloud-transcription","label":"本地转录与云端转录对比"},{"href":"/blog/how-whisper-ai-works","label":"Whisper AI 的工作原理"},{"href":"/blog/local-ai-voice-data-privacy","label":"为何选择本地 AI"},{"href":"/blog/why-open-source-voice-tools","label":"为何选择开源语音工具"},{"href":"/blog","label":"全部博客文章"}]},"blogLocalVsCloud":{"metadata":{"title":"本地与云端语音转文字：隐私、速度和成本","description":"从隐私、准确率、速度和成本对比本地与云端语音转文字，了解为何设备端转录能让音频保持私密且免费。","keywords":["本地 vs 云端转录","本地语音转文字","云端转录隐私","离线转录","本地语音识别","隐私语音转文字","设备端转录","云端 ASR vs 本地 ASR","nvidia parakeet","whisper vs parakeet"]},"schema":{"title":"本地 vs 云端转录：隐私、速度与准确性全面对比","description":"对本地与云端语音转文字进行客观对比——隐私、延迟、准确性、成本与离线能力。"},"breadcrumb":{"home":"首页","blog":"博客","article":"本地 vs 云端转录"},"sections":{"factCheck":"事实核查速览","howCloudWorks":"云端转录","howLocalWorks":"本地转录","privacy":"隐私","accuracy":"准确性","speed":"速度与延迟","cost":"成本","offline":"离线能力","comparison":"对比表","recommendations":"推荐建议","sources":"参考来源"},"hero":{"title":"本地 vs 云端转录：隐私、速度与准确性全面对比","subtitle":"你的语音数据很有价值。它的去向，取决于你在哪里处理它。","date":"2026 年 2 月 8 日","tag":"对比"},"intro":{"lead":"本地转录会在你的设备上完整处理音频","body1":"使用 OpenAI Whisper 和 NVIDIA Parakeet 等模型，通过 whisper.cpp 等优化运行时来运行。音频从不离开你的机器。","cloudLead":"云端转录会将你的音频发送到远程服务器","body2":"由 Google、AWS、Microsoft、Deepgram 或 AssemblyAI 等服务商运营，在大型 GPU 集群上处理后返回文本转录结果。两种方案之间的关键权衡可归结为五个因素：隐私、延迟、准确性、成本和离线能力。本文将结合真实数据和真实政策逐一剖析，帮你做出明智的决定。","lastUpdated":"最后更新：2026 年 2 月 17 日。文中所有量化数据和政策表述均已与至少两个一手来源交叉核对。"},"factCheck":{"title":"事实核查速览（双重来源）","item1":{"lead":"开源本地模型已具备生产级能力：","body":"Whisper 和 Parakeet 拥有公开的模型卡和开源实现，适用于设备端推理。","source1":"OpenAI Whisper","source2":"NVIDIA Parakeet"},"item2":{"lead":"云端 STT 按用量计费：","body":"主流服务商按音频时长和模型档位计量。","source1":"Google 定价","source2":"AWS 定价"},"item3":{"lead":"各服务商的数据处理方式不同：","body":"日志记录/数据保留设置和退出选项因服务而异，须自行配置。","source1":"Google 数据日志","source2":"AWS AI 退出政策"},"item4":{"lead":"监管背景至关重要：","body":"生物识别和敏感数据相关义务因司法辖区和使用场景而异。","source1":"GDPR 官方文本","source2":"《欧盟 AI 法案》官方文本"},"item5":{"lead":"OpenWhispr 的部署模式：","body":"OpenWhispr 默认采用本地处理，并在用户需要时支持 BYOK 云端路由。","source1":"OpenWhispr","source2":"whisper.cpp 运行时"}},"howCloudWorks":{"title":"云端转录的工作原理","paragraph1":"当你使用云端语音转文字服务时，你的音频会经历一段多步骤的旅程。首先，你的设备从麦克风采集原始音频，并压缩成适合传输的格式——通常是 Opus、FLAC 或线性 PCM。随后，这段压缩后的音频会通过互联网发送到服务商的 API 端点，通常经由 HTTPS 或用于实时流式传输的 WebSocket 连接。","paragraph2":"在服务器端，服务商会让你的音频通过托管在 GPU 集群上的大型神经网络模型进行处理。这些模型通常基于数十万小时的标注语音数据训练而成，远远超出任何个人所能收集的规模。生成的转录结果随后被发送回你的设备。","providersIntro":"主流的云端转录服务商包括：","providers":{"google":"——语言覆盖广泛，提供批量和实时流式两种选项。","aws":"——亚马逊的产品，支持自动语言识别、自定义词汇和 PII 脱敏。","azure":"——与 Microsoft 生态深度集成，支持自定义语音模型和实时转录。","deepgram":"——以速度和开发者体验著称，其 Nova-3 模型在具竞争力的价格下提供出色的准确性。","assemblyai":"——专注于准确性和开发者工具，内置说话人分离和内容审核。"},"paragraph3":"云端延迟是可变的，因为它涵盖了音频采集、上传、服务器端推理和结果返回。在网络良好的环境中，它可能感觉很快；而不稳定的网络或高延迟地区则会让体验明显变慢。批量转录可能耗时数秒到数分钟不等，具体取决于文件长度和排队情况。"},"howLocalWorks":{"title":"本地转录的工作原理","paragraph1":"本地转录直接在你的设备上运行语音识别模型。整个过程不涉及网络——音频从麦克风直接进入运行在 CPU 或 GPU 上的模型，文本从另一端输出。整个过程都在你机器的内存中完成。","paragraph2a":"让高质量本地转录变得切实可行的突破，是 OpenAI 于 2022 年 9 月发布的 Whisper——一个基于 68 万小时多语言音频训练的开源模型。最近，NVIDIA 发布了其 Parakeet 系列 ASR 模型，在英语基准测试中达到了业界领先的准确性，并以开源许可证提供。不久之后，Georgi Gerganov 打造了","paragraph2b":"，一个针对 CPU 推理优化的 C/C++ 重新实现，让 Whisper 的准确性在无需独立 GPU 的情况下也能在日常硬件上发挥。","modelTable":{"intro":"Whisper 提供多种模型规格，各自在准确性与速度之间做出取舍：","headerModel":"模型","headerParameters":"参数量","headerVram":"显存","headerSpeed":"相对速度","rows":[{"model":"Tiny","params":"39M","vram":"~1 GB","speed":"~10x"},{"model":"Base","params":"74M","vram":"~1 GB","speed":"~7x"},{"model":"Small","params":"244M","vram":"~2 GB","speed":"~4x"},{"model":"Medium","params":"769M","vram":"~5 GB","speed":"~2x"},{"model":"Large-v3","params":"1550M","vram":"~10 GB","speed":"1x（基准）"},{"model":"Turbo","params":"809M","vram":"~6 GB","speed":"~8x"}]},"paragraph3":"在 Apple Silicon（M1 至 M4）上，whisper.cpp 通过 Metal 在 GPU 上完整运行推理，即便使用 medium 模型也能达到实时甚至超实时的速度。Turbo 模型——large-v3 的优化变体，准确性损失极小——在现代 MacBook 上能以高吞吐量运行，这也是它常被用于低延迟本地听写工作流的原因。","paragraph4a":"其他本地转录引擎还包括","paragraph4b":"（轻量级，准确性较低）和","paragraph4c":"（支持 ONNX 运行时的新一代 Kaldi）。不过，通过 whisper.cpp 运行的 OpenAI Whisper 和 NVIDIA Parakeet 仍是本地转录质量的黄金标准。"},"privacy":{"title":"隐私：核心权衡","paragraph1":"隐私是本地转录与云端转录之间最大的区别所在。采用本地处理时，\"我的音频数据会怎样？\"这个问题有一个简单的答案：什么都不会发生。它留在你的设备上、内存里，转录完成后即被丢弃。没有需要信任的第三方，没有需要阅读的政策，也无需担心数据泄露。","paragraph2":"云端转录则需要你信任你的服务商。以下是主流服务商关于如何处理你音频的真实说法：","google":{"body1":"Google 的语音转文字数据日志默认是关闭的。在关闭状态下，Google 表示音频数据会在内存中处理，仅用于返回转录结果，不会存储在服务器上。但如果你选择开启数据日志（或使用某些需要它的功能），你的音频和转录内容可能会被存储并用于改进 Google 的模型。Google 的","link":"数据日志文档","body2":"详细说明了这些区别。","footnote":"数据处理位置可按地区指定，这对 GDPR 合规至关重要。"},"aws":{"body1":"默认情况下，AWS 可能会使用由 Amazon Transcribe 处理的内容来开发和改进 AWS 的 AI/ML 服务。不过，你可以通过","link":"AWS AI 服务退出政策","body2":"来选择退出。在退出后，AWS 表示你的内容不会被存储或用于服务改进。音频数据在传输中和静态存储时均经过加密。","footnote":"重要提示：退出并非默认设置。你必须主动进行配置。"},"azure":{"body1":"Azure 默认不会使用客户数据来改进其基础模型。其","link":"数据隐私文档","body2":"指出，发送到语音服务的音频会被处理并立即从其服务器删除。如果你创建自定义模型，你提供的训练数据会存储在与该资源相同的地区，直到你明确删除为止。","footnote":"Azure 提供断网容器部署，可实现完全本地化的处理。"},"regulatory":{"title":"监管考量","gdprLead":"GDPR：","gdprBody":"根据欧盟法律，语音录音被视为生物识别数据。即便采用标准合同条款（SCC）或欧盟-美国数据隐私框架，将音频发送到位于美国的云端服务商仍会引发 GDPR 下的数据跨境传输问题。本地处理通过将生物识别数据保留在数据主体自己的设备上，从根本上规避了这一问题。","hipaaLead":"HIPAA：","hipaaBody":"使用云端转录的医疗机构必须确保其服务商提供商业伙伴协议（BAA）。Google、AWS 和 Azure 均提供 BAA，但配置要求十分严格。本地转录则绕开了 HIPAA 的数据处理规则，因为受保护的健康信息从不脱离涵盖实体的控制范围。"},"localAdvantage":{"title":"本地的优势","body":"采用本地转录，没有需要研读的隐私政策，没有需要信任的数据保留期限，也没有需要寻找的退出开关。你的音频在内存中处理，从不写入磁盘（除非你选择保存录音）。这是一种根本不同的信任模式——你无需信任任何人，因为数据从不离开你的机器。"}},"accuracy":{"title":"准确性：本地能有多接近？","paragraph1":"说句实话：在某些使用场景下，顶级云端服务商在原始准确性上仍占有优势。它们基于海量专有数据集训练，提供实时模型更新，还能利用上下文相关的自定义词汇。对于难度较高的音频——浓重口音、嘈杂环境、领域专业术语——像 Google 的 Chirp 2 或 Deepgram 的 Nova-3 这类云端服务往往表现出色。","paragraph2":"话虽如此，差距已经大幅缩小。基于 68 万小时多语言音频训练的 OpenAI Whisper large-v3，在大多数常见语言上都实现了具竞争力的词错率（WER）。NVIDIA 的 Parakeet 模型更进一步，在标准英语基准测试中取得了一些最低的 WER。独立基准测试一再表明，对于英语，这些开源模型的表现与商用云端服务相差仅几个百分点，在某些长尾语种上甚至超过它们。","paragraph3":"Turbo 模型——large-v3 的蒸馏变体，参数量为 8.09 亿而非 15.5 亿——保留了大部分准确性，同时运行速度快约 8 倍。对于对着像样麦克风清晰口述的实时听写而言，对大多数用户来说，Turbo 与云端 API 之间的差异微乎其微。","cloudAdvantages":{"title":"云端在准确性上的优势","item1":"自定义词汇与领域自适应","item2":"服务器端的持续模型改进","item3":"更出色的说话人分离（谁说了什么）","item4":"针对各语言调优的自动标点与格式化"},"localAdvantages":{"title":"本地在准确性上的优势","item1":"无网络传输带来的音频压缩失真","item2":"稳定、可复现的性能——不受服务器波动影响","item3":"完整无压缩的 16kHz 音频直接喂给模型","item4":"Whisper large-v3 和 Parakeet 在清晰语音听写上可媲美云端"},"bottomLine":"结论：如果你在安静环境中用像样的麦克风口述，本地运行的 Whisper Turbo、Whisper large-v3 或 Parakeet 等模型在大多数实际用途上能给出与云端服务难以区分的结果。而在嘈杂的多说话人场景、小众语言和领域专业词汇方面，云端则更胜一筹。"},"speed":{"title":"速度与延迟","paragraph1":"延迟对实时听写最为关键——你希望话音刚落，文字就立刻出现。在这方面，本地转录与云端转录有着根本不同的延迟特征。","cloudBreakdown":{"title":"云端延迟构成","item1":"音频采集 + 编码","item2Lead":"网络上传与下载","item2Rest":"（因线路质量而异）","item3Lead":"服务器端排队 + 推理","item3Rest":"（因服务商负载而异）","summary":"用户感知到的延迟可能因网络和地区而显著不同。"},"localBreakdown":{"title":"本地延迟（Apple Silicon）","item1Lead":"Tiny/base：","item1Rest":"响应最快，质量上限较低","item2Lead":"Small：","item2Rest":"速度与质量兼顾","item3Lead":"Turbo：","item3Rest":"高质量实时听写的理想之选","item4Lead":"大型模型：","item4Rest":"质量最高，算力开销更大","summary":"无网络依赖。无论连接状况如何都保持稳定。"},"paragraph2":"对于实时流式传输，在连接良好时，云端服务商能快速返回部分识别假设。但这仍需要稳定的低延迟连接。在飞机上、拥挤的公共 Wi-Fi 中，或受限的 VPN 线路之后，感知延迟可能迅速恶化。","paragraph3":"传统意义上，使用 whisper.cpp 的本地转录并非实时流式。大多数本地实现采用\"按住说话\"的模式：你说话，音频被缓冲，然后模型处理完整的语音片段。在搭载 Apple M 系列芯片上使用 Turbo 模型时，这一处理步骤足够快，对于听写来说延迟几乎可以忽略——短句通常在一秒以内完成。","footnote":"对于预先录制文件的长篇转录，云端服务能在 GPU 集群间并行处理，往往能快速完成大批量任务。本地吞吐量受限于你的机器和模型选择，因此处理超长录音时运行时间可能更慢。"},"cost":{"title":"成本：免费 vs. 按分钟付费","paragraph1":"本地转录除了你已拥有的硬件之外，不需要任何额外花费。没有 API 密钥，没有计费面板，没有按分钟收费。你下载一个模型文件（从 Tiny 的 75MB 到 Large-v3 的 3GB 不等），然后就大功告成了。永远如此。","paragraph2":"云端服务商按用量收费（按分钟或按小时，视厂商和模型而定）。套餐和折扣经常变动，因此在做预算前请先核实当前费率：","table":{"headerProvider":"服务商","headerBatch":"批量 / 预录","headerStreaming":"流式 / 实时","headerFreeTier":"免费额度","rows":[{"provider":"Google Speech-to-Text","batch":"按用量计费（按模型分档）","streaming":"按用量计费（按模型分档）","free":"试用/免费配额因账户而异"},{"provider":"AWS Transcribe","batch":"按用量计费","streaming":"按用量计费","free":"入门免费额度（限时）"},{"provider":"Azure Speech","batch":"因地区/模型而异","streaming":"因地区/模型而异","free":"有限免费配额"},{"provider":"Deepgram","batch":"按模型档位计费","streaming":"按模型档位计费","free":"基于额度的试用"},{"provider":"AssemblyAI","batch":"按模型档位计费","streaming":"按模型档位计费","free":"基于额度的试用"},{"provider":"本地（Whisper / Parakeet）","batch":"免费","streaming":"免费","free":"无限量，永久"}]},"estimate":{"title":"如何估算你的云端开销","body1a":"使用这个简单的公式：","formula":"年度音频分钟数 x 厂商费率","body1b":"如果你的团队大量使用听写，总成本会随用量和席位数线性增长。","body2":"本地推理（Whisper/Parakeet）避免了持续的 API 费用，这正是许多团队将本地作为默认方案、只在特殊情况下才路由到云端的原因。"},"footnote":"注意：云端定价通常会对说话人分离、PII 脱敏或自定义词汇等功能收取额外费用。上述基础价格仅适用于标准转录。"},"offline":{"title":"离线能力：本地的绝对胜场","paragraph1":"这一点是非此即彼的：云端转录需要联网，本地转录则不需要。对许多人来说，这是决定性因素。","scenariosIntro":"离线能力不可或缺的场景：","travel":{"title":"出行与远程办公","body":"航班、火车、乡村地区、发展中地区——任何 Wi-Fi 时有时无或根本没有的地方。本地转录在 3.5 万英尺高空的表现，与在你办公桌前完全一样。"},"airGapped":{"title":"物理隔离环境","body":"政府机构、国防承包商、安全研究实验室和金融机构通常运行在物理隔离网络上，对外网络访问被完全屏蔽。本地转录是唯一选择。"},"fieldWork":{"title":"野外作业","body":"冲突地区的记者、偏远野外站点的研究人员、乡村诊所的医护人员——这些专业人士需要在蜂窝信号可能不稳定或不可用的环境中获得可靠的转录。"},"reliability":{"title":"可靠性","body":"即便在网络良好的办公室里，云端服务也会出现宕机。AWS、Google Cloud 和 Azure 都曾发生过波及语音 API 的数小时事故。本地转录没有任何可能宕机的外部依赖。"}},"comparisonTable":{"title":"逐项对比","headerFactor":"因素","headerLocal":"本地","headerCloud":"云端","headerNotes":"说明","rows":[{"factor":"隐私","local":"full","cloud":"none","notes":"本地：数据从不离开设备。云端：音频被发送到第三方服务器。"},{"factor":"准确性（英语）","local":"partial","cloud":"full","notes":"云端模型略胜一筹；Whisper large-v3 和 NVIDIA Parakeet 大幅缩小了差距。"},{"factor":"准确性（多语言）","local":"partial","cloud":"full","notes":"云端服务商凭借定制训练的模型拥有更广的语言覆盖。"},{"factor":"延迟","local":"full","cloud":"partial","notes":"本地避免了网络往返。云端延迟取决于连接质量和地区。"},{"factor":"成本","local":"full","cloud":"none","notes":"本地没有按分钟的 API 计费。云端服务采用按用量定价。"},{"factor":"离线使用","local":"full","cloud":"none","notes":"本地随处可用。云端需要联网。"},{"factor":"配置便利性","local":"partial","cloud":"full","notes":"云端只需一次 API 调用。本地需要下载模型并考虑硬件。"},{"factor":"说话人分离","local":"none","cloud":"full","notes":"多说话人识别是云端的一大优势。"},{"factor":"可扩展性","local":"none","cloud":"full","notes":"云端可无限扩展。本地受限于你的硬件。"}]},"recommendations":{"title":"何时选择哪种方案","intro":"两种方案都没有绝对的优劣。正确的选择取决于你看重什么。","chooseLocal":{"title":"如果满足以下情况，请选择本地……","item1":"隐私不容妥协（医疗、法律、个人日记）","item2":"你需要离线或物理隔离运行","item3":"你希望转录没有任何持续成本","item4":"你的主要用途是听写（单说话人、清晰音频）","item5":"你受 GDPR、HIPAA 或类似法规约束"},"chooseCloud":{"title":"如果满足以下情况，请选择云端……","item1":"你需要为多人会议进行说话人分离","item2":"在困难音频条件下追求最高准确性","item3":"你正在构建一个需要扩展到大量用户的应用","item4":"你需要实时流式转录","item5":"你的硬件有限，无法运行大型模型"},"hybrid":{"title":"混合方案","body":"你不必只能二选一。一些应用——包括 OpenWhispr——同时支持两种模式：默认使用 Whisper 和 Parakeet 等本地模型以兼顾隐私与零成本，并可在需要云端提供的额外准确性或功能时，选择自带云端 API 密钥（OpenAI、Deepgram 等）。这种\"BYOK\"（自带密钥）模式让你兼得两者之长，而不会被任何一种方案所锁定。"}},"sourcesSection":{"title":"参考来源与延伸阅读","whisper":"——原始模型仓库，包含模型规格、基准测试和文档。","whisperCpp":"——Whisper 的 C/C++ 移植版，针对 CPU 和 Apple Silicon GPU 推理优化。","parakeetLabel":"NVIDIA Parakeet RNNT 模型卡","parakeet":"——开源 ASR 模型指标与使用指南。","googlePricingLabel":"Google Cloud Speech-to-Text 定价","googlePricing":"——Google 语音 API 的当前定价档位。","awsPricingLabel":"AWS Transcribe 定价","awsPricing":"——亚马逊批量与流式的当前按用量定价。","azurePricingLabel":"Azure Speech Services 定价","azurePricing":"——Microsoft 实时与批量语音转文字的定价。","deepgramPricingLabel":"Deepgram 定价","deepgramPricing":"——Deepgram 各模型档位的定价详情。","assemblyaiPricingLabel":"AssemblyAI 定价","assemblyaiPricing":"——AssemblyAI 用量与模型档位的定价详情。","googleLoggingLabel":"Google Speech-to-Text 数据日志","googleLogging":"——Google 关于音频数据如何处理和存储的文档。","awsOptOutLabel":"AWS AI 服务退出政策","awsOptOut":"——如何阻止 AWS 使用你的内容改进其 AI 服务。","azurePrivacyLabel":"Azure Speech 数据隐私","azurePrivacy":"——Microsoft 关于语音服务的数据隐私与安全文档。","whisperPaperLabel":"通过大规模弱监督实现稳健语音识别","whisperPaper":"——Radford 等人（2022）的原始 Whisper 论文，详述了训练方法与 WER 基准。","gdprLabel":"GDPR（欧盟官方文本）","gdpr":"——个人/生物识别数据处理的法律基准。","aiActLabel":"《欧盟 AI 法案》（欧盟官方文本）","aiAct":"——与生物识别及高风险系统相关的 AI 风险分级义务。","openwhispr":"——在单一应用中实现本地优先 + 可选 BYOK 云端工作流的范例。"},"cta":{"title":"在一个应用中体验两种方案","body":"OpenWhispr 同时支持本地模型（OpenAI Whisper 和 NVIDIA Parakeet）以兼顾隐私与零成本，并在你需要最高准确性时支持自带云端 API 密钥。开源，永久免费。","starButton":"在 GitHub 上为我们点星","footnote":"无需账户 · 可离线使用 · 永久开源"},"internalLinks":{"localAi":"本地 AI 的理由","offline":"离线转录","allBlog":"全部博客文章","allComparisons":"全部对比"}},"blogModelSizes":{"metadata":{"title":"Whisper 模型大小：完整指南 | OpenWhispr","description":"对比全部 Whisper 模型大小的参数、速度、准确率和内存需求，选择适合本地转录的 tiny、small、large-v3 或 turbo。","keywords":["Whisper 模型尺寸","Whisper tiny 与 large 对比","Whisper 模型对比","该用哪个 Whisper 模型","Whisper large v3","Whisper 模型准确率","Whisper 模型速度","Whisper 模型参数","whisper.cpp 模型尺寸"]},"schema":{"title":"Whisper 模型尺寸详解：Tiny、Base、Small、Medium 与 Large 对比","description":"一份实用指南，帮你选择合适的 Whisper 模型。对比参数量、速度、准确率和硬件要求。"},"breadcrumb":{"home":"首页","blog":"博客","article":"Whisper 模型尺寸详解"},"layout":{"title":"Whisper 模型尺寸详解：Tiny、Base、Small、Medium 与 Large 对比","subtitle":"全面对比每个 Whisper 模型。参数量、速度、准确率，以及你究竟该用哪一个。","date":"2026 年 2 月 5 日","tag":"博客"},"sectionsToc":{"factCheck":"事实核查速览","comparisonTable":"模型对比","tiny":"Tiny（39M）","base":"Base（74M）","small":"Small（244M）","medium":"Medium（769M）","large":"Large（1.55B）","turbo":"Turbo（809M）","englishVsMultilingual":"英语专用版与多语言版","hardware":"硬件要求","recommendations":"选型建议","openwhispr":"OpenWhispr","sources":"参考来源"},"cards":{"resources":"资源占用","speed":"速度","accuracy":"准确率","bestFor":"最适合："},"intro":{"lead":"OpenAI 的 Whisper 提供 9 个模型版本，分为 5 个尺寸类别：","body1":"tiny（3900 万参数）、base（7400 万）、small（2.44 亿）、medium（7.69 亿）和 large（15.5 亿）。每个尺寸都有多语言版本和英语专用版本（large 和 turbo 除外，二者仅有多语言版）。此外还有一个 turbo 模型（8.09 亿）——它是 large-v3 的蒸馏版本，准确率几乎不相上下，但速度大幅提升。模型越大越准确，但速度越慢，占用的内存也越多。","body2":"大多数本地转录应用使用的是 whisper.cpp 量化（GGML）格式，其模型文件大小从 75 MB（tiny）到 2.9 GB（large）不等。运行时内存占用从 tiny 的约 273 MB 到 large 的 3.9 GB 不等。合适的模型取决于你的硬件、语言以及你对准确率的要求。","lastUpdated":"最近更新：2026 年 2 月 17 日。参数、内存和基准测试数据均已与至少两个权威来源交叉核对。"},"factCheck":{"title":"事实核查速览（双重来源）","item1":{"lead":"官方模型系列与参数量：","body":"tiny/base/small/medium/large 以及 turbo 均由 OpenAI 官方记录。","link1":"OpenAI Whisper README","link2":"Whisper 模型卡"},"item2":{"lead":"GGML/GGUF 磁盘与运行时内存背景：","body":"本地部署占用数据来自 whisper.cpp 的转换/运行时文档。","link1":"whisper.cpp","link2":"whisper.cpp 内存说明"},"item3":{"lead":"基准测试注意事项：","body":"WER 因数据集和语言不同而差异巨大；LibriSpeech 的数据并非通用的质量评分。","link1":"Whisper 论文","link2":"large-v2 模型卡"},"item4":{"lead":"Turbo 模型定位：","body":"turbo 是一个侧重速度的版本，其准确率上的取舍在发布说明/模型卡中均有记录。","link1":"turbo 模型卡","link2":"turbo 发布讨论"},"item5":{"lead":"OpenWhispr 使用场景：","body":"OpenWhispr 提供多个 Whisper 尺寸，让用户能在设备本地调节速度与准确率。","link1":"OpenWhispr","link2":"whisper.cpp 后端"}},"table":{"title":"完整模型对比","intro":"本表涵盖了每一个官方 Whisper 模型。「相对速度」以 large 为基准（1 倍）。VRAM 数据针对 GPU 上的 PyTorch 推理。GGML 列显示的是 whisper.cpp 的文件大小和内存占用，而大多数桌面应用使用的正是 whisper.cpp。","headers":{"model":"模型","params":"参数量","englishOnly":"英语专用","vram":"VRAM（GPU）","ggmlDisk":"GGML 磁盘","ram":"内存（whisper.cpp）","speed":"速度","englishWer":"英语 WER","multilingualWer":"多语言 WER"},"notAvailable":"不适用","notesLead":"说明：","notesBody":"WER（词错误率）数据是来自 Whisper 论文、HuggingFace 模型卡（英语用 LibriSpeech 测试集）以及 OpenAI 公开基准测试的近似平均值。确切的 WER 会因数据集、语言、音频质量和评测方法的不同而显著变化。数值越低越好。速度以 large 为基准（1 倍 = 最慢）。"},"tiny":{"title":"Tiny（3900 万参数）","para1":"tiny 模型是 Whisper 中最小、最快的版本。仅有 3900 万参数，GGML 文件大小为 75 MB，几乎可以在任何硬件上轻松运行——包括 Raspberry Pi、老旧笔记本电脑和低端移动设备。相对于 large 模型，它的转录速度约为实时速度的 10 倍，因此在延迟比完美准确率更重要时，它是最佳选择。","resources":{"l1":"GGML：磁盘占用 75 MiB","l2":"内存：运行时约 273 MB","l3":"VRAM：约 1 GB（GPU 模式）"},"speed":{"l1":"比 large 快约 10 倍","l2":"在 Apple Silicon 上近乎瞬时","l3":"在大多数 CPU 上可实时运行"},"accuracy":{"l1":"英语 WER：约 7.6%","l2":"多语言 WER：约 12%","l3":"难以应对口音/噪声"},"para2":"在 tiny.en 版本（英语专用）上，LibriSpeech test-clean 的 WER 约为 5.6%，在 test-other（嘈杂/带口音音频）上升至约 14.9%。多语言版 tiny 模型在英语上表现稍逊，但支持全部 99 种 Whisper 语言。","para3":"tiny 非常适合快速记笔记、低风险口述、实时字幕原型，以及算力有限的嵌入式或边缘设备。它是许多基于 whisper.cpp 工具的默认起步模型，因为它几秒钟即可下载完成并即刻运行。","bestFor":"追求极致转录速度、低资源硬件、Raspberry Pi、实时应用"},"base":{"title":"Base（7400 万参数）","para1":"base 模型的参数量是 tiny 的两倍（7400 万对 3900 万），但仍然非常轻量。GGML 文件 142 MiB，运行时内存约 388 MB，依然能在几乎任何现代设备上轻松运行。相比 tiny，准确率的提升十分明显——尤其是在嘈杂音频和带口音的语音上。","resources":{"l1":"GGML：磁盘占用 142 MiB","l2":"内存：运行时约 388 MB","l3":"VRAM：约 1 GB（GPU 模式）"},"speed":{"l1":"比 large 快约 7 倍","l2":"在 CPU 上依然很快","l3":"在 Apple Silicon 上低于一秒"},"accuracy":{"l1":"英语 WER：约 5.0%","l2":"多语言 WER：约 10%","l3":"在嘈杂音频上优于 tiny"},"para2":"base.en 版本在 LibriSpeech test-clean 上的 WER 约为 4.3%，在 test-other 上约为 12.8%。相比 tiny.en（分别为 5.6% 和 14.9%），这是一次有意义的提升，尤其是在高难度音频上。","para3":"对于 tiny 准确率尚不够用的低功耗设备，base 是个不错的选择。它在实时转录流水线中也很受欢迎——你想要速度与质量的平衡，又不希望内存占用超过 500 MB 时，它正合适。","bestFor":"需要比 tiny 更高准确率的低功耗设备、实时流式转录、入门级硬件"},"small":{"title":"Small（2.44 亿参数）","para1":"从 small 开始，Whisper 才真正让人感觉准确。凭借 2.44 亿参数，它相比 base 实现了 3.3 倍的飞跃，准确率显著提升——尤其是在非英语语言和嘈杂录音上。GGML 文件为 466 MiB，运行时内存占用约 852 MB。它的运行速度约为 large 的 4 倍。","resources":{"l1":"GGML：磁盘占用 466 MiB","l2":"内存：运行时约 852 MB","l3":"VRAM：约 2 GB（GPU 模式）"},"speed":{"l1":"比 large 快约 4 倍","l2":"在现代笔记本上很快","l3":"在 Apple Silicon 上可实时运行"},"accuracy":{"l1":"英语 WER：约 3.4%","l2":"多语言 WER：约 7%","l3":"能很好地处理口音"},"para2":"small.en 版本在 LibriSpeech test-clean 上的 WER 约为 3.0%——对于一个不到 500 MB 的模型来说，这是相当出色的成绩。在许多仅需英语的场景中，small.en 能以一小部分资源成本提供接近 medium 的准确率。","para3":"small 常被推荐为大多数用户的默认模型。它在任何现代笔记本上都运行良好（包括配备 8 GB 内存的 MacBook Air M1），在各种语言上都有不错的准确率，且转录速度足够支撑实时口述工作流。","recommendationLead":"我们的建议：","recommendationBody":"如果你不确定从哪个模型入手，就从 small 开始。对于绝大多数硬件和使用场景，它在速度、准确率和资源占用之间取得了最佳平衡。","bestFor":"使用现代笔记本的大多数用户、默认推荐模型、良好的多语言支持"},"medium":{"title":"Medium（7.69 亿参数）","para1":"从 medium 开始，边际收益逐渐递减——但相比 small，准确率的提升仍然有意义，尤其是在多语言转录、专业词汇和高难度音频条件下。凭借 7.69 亿参数，它需要 1.5 GiB 磁盘空间（GGML），运行时内存约 2.1 GB。","resources":{"l1":"GGML：磁盘占用 1.5 GiB","l2":"内存：运行时约 2.1 GB","l3":"VRAM：约 5 GB（GPU 模式）"},"speed":{"l1":"比 large 快约 2 倍","l2":"在 8 GB 及以上机器上从容运行","l3":"需要不错的硬件"},"accuracy":{"l1":"英语 WER：约 2.9%","l2":"多语言 WER：约 5%","l3":"在专业内容上表现强劲"},"para2":"medium.en 模型在 LibriSpeech test-clean 上的 WER 约为 3.0%，在 test-other 上约为 7.5%。然而在更广泛的基准测试中——尤其是涉及多样口音、背景噪声和特定领域词汇的测试——medium 始终优于 small。","para3":"对于看重准确率、但又没有硬件（或耐心）去跑 large 的专业转录工作流来说，medium 是个有力的选择。它在配备 8 GB 及以上内存的机器上运行良好，并能从 GPU 加速中获得显著收益。","bestFor":"专业用途、多语言转录、专业内容、配备 8 GB 及以上内存的机器"},"large":{"title":"Large（15.5 亿参数）","para1":{"part1":"large 模型是最准确的 Whisper 版本，拥有 15.5 亿参数。它有三个版本：","part2":"（最初的发布版），","part3":"（训练轮数增加 2.5 倍并加入正则化），以及","part4":"（在更多数据上训练，并采用 128 个梅尔频率分箱而非 80 个）。large-v3 是目前追求最高准确率的推荐版本。"},"resources":{"l1":"GGML：磁盘占用 2.9 GiB","l2":"内存：运行时约 3.9 GB","l3":"VRAM：约 10 GB（GPU 模式）"},"speed":{"l1":"1 倍（基准——最慢）","l2":"强烈推荐使用 GPU","l3":"CPU：可能慢于实时速度"},"accuracy":{"l1":"英语 WER：约 2.4%（v3）","l2":"多语言 WER：约 3.5%（v3）","l3":"在所有语言上都最佳"},"versionsTitle":"large-v2 与 large-v3 对比","v2":{"l1":"训练轮数比 large-v1 多 2.5 倍","l2":"加入正则化以获得更好的泛化能力","l3":"LibriSpeech test-clean WER：约 3.0%","l4":"在某些音频类型上更稳定"},"v3":{"l1":"128 个梅尔分箱（对比 80 个）——更精细的频率分辨率","l2":"在 100 万小时标注音频 + 400 万小时伪标注音频上训练","l3":"跨语言相比 large-v2 降低 10-20% 的错误率","l4":"新增粤语支持"},"whichLead":"该用哪个 large 模型：","whichBody":"对于新项目，请使用 large-v3。在各语言的平均表现上，它都严格优于 large-v2。不过，一些用户反映，在音频片段非常安静或静音的特定边缘情况下，large-v2 产生的幻觉（生成音频中并不存在的文本）更少。如果你在使用 large-v3 时遇到幻觉问题，可以尝试退回 large-v2 作为备选。","bestFor":"追求最高准确率、专业/医疗/法律转录、低资源语言、批量处理"},"turbo":{"title":"Turbo（8.09 亿参数）","para1":"turbo 模型是 large-v3 的蒸馏版本，它在保留大部分准确率的同时大幅缩短了推理时间。OpenAI 通过将解码器从 32 层精简至仅 4 层实现了这一点——参数量从 15.5 亿减少到 8.09 亿。而负责繁重工作的编码器与 large-v3 保持完全一致。","resources":{"l1":"GGML：磁盘占用约 1.6 GiB","l2":"内存：运行时约 2.3 GB","l3":"VRAM：约 6 GB（GPU 模式）"},"speed":{"l1":"比 large 快约 8 倍","l2":"准确率接近 large，速度接近 tiny","l3":"配合 torch.compile 在 GPU 上表现出色"},"accuracy":{"l1":"英语 WER：约 2.5%","l2":"多语言 WER：约 3.7%","l3":"相比 large-v3 质量略有损失"},"para2":"当你想要接近 large-v3 的准确率、却无法承受完整 large 模型的延迟时，turbo 是最佳选择。在 GPU 上配合 torch.compile 等优化，turbo 相比标准推理可实现最高 4.5 倍的速度提升，因而极为快速。","limitationLead":"重要限制：","limitationBody":"turbo 模型未针对翻译任务进行训练。如果你需要把一种语言的语音翻译成英语，请改用 medium 或 large-v3。","bestFor":"高准确率实时转录、对延迟敏感的应用、large-v3 太慢的场景"},"englishVsMultilingual":{"title":"英语专用版与多语言版模型","para1":{"part1":"对于 tiny、base、small 和 medium 尺寸，Whisper 提供两个版本：多语言模型和英语专用（","part2":"）模型。large 和 turbo 模型仅有多语言版——没有英语专用版本。"},"enTitle":"何时使用 .en（英语专用）","en":{"l1":"你只转录英语音频","l2":"在 tiny/base 尺寸级别上英语准确率更高","l3":"稍快一些——无需语言检测开销","l4":"文件大小与多语言版相同"},"multiTitle":"何时使用多语言版","multi":{"l1":"你需要转录非英语音频","l2":"你的音频包含多种混合语言","l3":"你需要语音翻译（译为英语）","l4":"使用 large 或 turbo 时必须用它（不存在 .en 版本）"},"para2":".en 版与多语言版模型之间的性能差距在较小尺寸上最为明显。对于 tiny 和 base，.en 版本在英语上明显更好。等到了 small 和 medium，差距就大幅缩小了。在 large 级别只有多语言模型——而它在英语上的表现同样出色。","ruleLead":"经验法则：","ruleBody":"如果你只用英语，并且在 tiny 或 base 中选择，那就选 .en 版本。对于 small 及以上，多语言版用于英语效果很好，还能让你灵活应对其他语言。"},"hardware":{"title":"硬件要求：哪些模型能在哪里运行","intro":"你需要的硬件在很大程度上取决于模型尺寸，以及你使用的是 whisper.cpp（CPU/Metal/CUDA）还是 PyTorch 实现（侧重 GPU）。以下是常见硬件类别上的预期表现。","m1Air":{"title":"MacBook Air M1（8 GB 内存）","r1Label":"tiny/base：","r1":"瞬时完成。快于实时速度。","r2Label":"small：","r2":"很快。借助 Metal 可达实时或更优。","r3Label":"medium：","r3":"可用。在 CPU 上可能略慢于实时速度。Metal 能带来显著帮助。","r4Label":"large：","r4":"可行但内存吃紧。预计会慢于实时速度。","r5Label":"turbo：","r5":"很合适。借助 Metal 加速可接近实时速度。"},"mbpPro":{"title":"MacBook Pro M2/M3（16-36 GB 内存）","r1Label":"tiny/base/small：","r1":"瞬时完成。全部远在硬件上限之内。","r2Label":"medium：","r2":"很快。借助 Metal 从容运行。","r3Label":"large：","r3":"良好。借助 Metal 可达实时或接近实时。","r4Label":"turbo：","r4":"出色。快速实时转录。"},"midRange":{"title":"中端 Windows/Linux 笔记本（8 GB 内存，集成显卡）","r1Label":"tiny/base：","r1":"很快。仅用 CPU 即可。","r2Label":"small：","r2":"良好。在大多数现代 CPU 上从容运行。","r3Label":"medium：","r3":"可行。在 CPU 上可能比实时速度慢 2-3 倍。","r4Label":"large：","r4":"有挑战。仅用 CPU 会很慢。","r5Label":"turbo：","r5":"可行。若有 Vulkan 可从中受益。"},"nvidia":{"title":"配备 NVIDIA 显卡的台式机（RTX 3060 及以上，8 GB 及以上 VRAM）","r1Label":"所有模型：","r1":"很快。GPU 能从容应对一切。","r2Label":"large：","r2":"需要 10 GB VRAM（RTX 3060 12GB 或更高）。","r3Label":"turbo：","r3":"配合 CUDA 是最快的选择。明显快于 large。","r4":"在 whisper.cpp 中使用 CUDA 后端可获得最佳性能。"},"backendsTitle":"whisper.cpp 加速后端","backendsIntro":"whisper.cpp——大多数桌面应用使用的 C/C++ 移植版本——支持多种硬件加速后端，可大幅提升性能：","metal":{"title":"Metal（Apple Silicon）","body":"在 M1/M2/M3/M4 Mac 上进行完整的 GPU 推理。整个模型在 GPU 上运行，零内存拷贝。这是 macOS 用户最快的后端，也是 OpenWhispr 在 Mac 上所采用的方案。"},"coreml":{"title":"Core ML（Apple）","body":"使用 Apple 的神经网络引擎进行推理。在较新芯片上，某些模型尺寸的速度可能快于 Metal，且更省电。需要先将模型转换为 Core ML 格式。"},"cuda":{"title":"CUDA（NVIDIA）","body":"为 NVIDIA 显卡提供 GPU 加速。对于拥有独立 NVIDIA 显卡的用户，这是最快的后端。需要 CUDA 工具包。是台式机上运行 large 和 turbo 模型的最佳选择。"},"vulkan":{"title":"Vulkan（跨厂商 GPU）","body":"可与 Intel、AMD 和 NVIDIA 显卡配合使用。对于非 NVIDIA 系统是不错的备选方案。在 Linux 和 Windows 上均可用。性能因 GPU 厂商和驱动而异。"},"speedTitle":"近似转录速度","speedIntro":"针对一段 60 秒音频的实时系数（RTF）。RTF < 1.0 表示快于实时速度。这些只是粗略估计——实际性能取决于音频内容、模型量化和系统负载。","speedTable":{"model":"模型","m1":"M1 MacBook Air","m3":"M3 Pro MacBook","intel":"Intel i7（CPU）","rtx":"RTX 3060（CUDA）"},"readingLead":"如何读懂此表：","readingBody":"0.1x 表示转录 60 秒音频约需 6 秒。1.0x = 实时速度。高于 1.0x 的数值表示慢于实时速度。所有数据均使用默认设置的 whisper.cpp，并在适用时采用 Metal（Mac）或 CUDA（NVIDIA）后端。"},"recommendations":{"title":"你该使用哪个模型？","intro":"以下是基于常见场景的具体建议。拿不定主意时，就从 small 开始，再根据使用体验向上或向下调整。","fastest":{"q":"「我想要尽可能快」","part1":"使用","part2":"或","part3":"。在现代硬件上以毫秒级完成转录。准确率较粗糙，但用于快速记笔记和低风险口述足够了。"},"laptop":{"q":"「我想在笔记本上获得不错的准确率」","part1":"使用","part2":"。对于现代笔记本，它在速度和准确率之间取得了最佳平衡。如果你的机器有 16 GB 及以上内存，","part3":"也是个绝佳选择。","tag":"small 或 medium"},"accuracy":{"q":"「我想要最高的准确率」","part1":"使用","part2":"。它是整体上最准确的 Whisper 模型。如果速度也重要，","part3":"能以 8 倍速度提供 95% 的准确率。","tag":"large-v3 或 turbo"},"englishOnly":{"q":"「我只用英语」","part1":"使用你所选尺寸的","part2":"版本，以获得最佳英语准确率。对于 tiny 和 base，.en 模型明显更好。对于 small 及以上，差异很小。","tag":"small.en 或 medium.en"},"multiLang":{"q":"「我使用多种语言」","part1":"使用多语言版本。","part2":"或","part3":"可获得均衡的表现。","part4":"可获得最佳的多语言准确率。","tag":"small、medium 或 large-v3"},"limited":{"q":"「我的硬件非常有限」","part1":"使用","part2":"或","part3":"。两者都能在 1 GB 内存下运行，甚至可在 Raspberry Pi 和其他单板计算机上工作。base 相比 tiny 准确率有明显提升，而额外成本极小。","tag":"tiny 或 base"}},"openwhispr":{"title":"OpenWhispr 如何处理模型选择","intro":"是一款开源的桌面口述应用，底层使用 whisper.cpp。它让你直接在设置中下载并切换任意 Whisper 模型——无需命令行。模型只需下载一次，即可本地存储。你可以随时更换模型，为你的硬件平衡速度与准确率。","c1":{"title":"下载任意模型","body":"在设置中从 tiny 到 large-v3 任选。OpenWhispr 会自动下载 GGML 版本。"},"c2":{"title":"即时切换","body":"随时更换模型。无需重启。试用不同尺寸，找到最适合你硬件的那一个。"},"c3":{"title":"针对硬件优化","body":"在 Apple Silicon 上使用 Metal，在 NVIDIA 上使用 CUDA，在其他平台上使用优化过的 CPU 推理。全程本地，全程私密。"},"suggestionLead":"我们的建议：","suggestionPart1":"从","suggestionPart2":"开始。如果觉得慢，就降到 base 或 tiny。如果你想要更高的准确率且硬件能够胜任，就升到 medium 或 large-v3。OpenWhispr 让尝试变得轻松。"},"sources":{"title":"参考来源","s1":{"link":"Radford 等人《Robust Speech Recognition via Large-Scale Weak Supervision》（2022）","desc":"——附有完整基准测试结果的 Whisper 原始论文。"},"s2":{"link":"OpenAI Whisper GitHub 仓库","desc":"——官方模型表、参数量、VRAM 要求和相对速度数据。"},"s3":{"link":"OpenAI Whisper 模型卡","desc":"——官方训练数据构成与评测注意事项。"},"s4":{"link":"whisper.cpp GitHub 仓库","desc":"——GGML 模型尺寸、内存占用和硬件加速详情。"},"s5":{"link":"Whisper large-v2 模型卡（HuggingFace）","desc":"——众多 WER 对比所采用的 LibriSpeech 基准背景。"},"s6":{"link":"Whisper large-v3 模型卡（HuggingFace）","desc":"——训练细节、相对 large-v2 的架构变化和多语言改进。"},"s7":{"link":"Whisper large-v3-turbo 模型卡（HuggingFace）","desc":"——Turbo 架构细节、解码器精简的具体做法和性能取舍。"},"s8":{"link":"OpenAI Whisper Turbo 发布公告（GitHub Discussion #2363）","desc":"——官方 turbo 发布说明和跨语言性能对比。"},"s9":{"link":"OpenWhispr","desc":"——选择模型尺寸的实用本地部署背景。"}},"cta":{"title":"在 OpenWhispr 中试用不同的 Whisper 模型","body":"开源、本地优先的口述工具。下载任意 Whisper 模型，一键切换尺寸，为你的硬件找到最完美的平衡。","starButton":"在 GitHub 上给我们加星","footnote":"无需账户 · 离线可用 · 永久开源"},"internalLinks":{"howWhisperWorks":"Whisper AI 的工作原理","linux":"Linux 上的语音口述","allPosts":"全部博客文章","developers":"面向开发者的 OpenWhispr","localVsCloud":"本地转录与云端转录对比"}},"blogOpenSource":{"metadata":{"title":"为什么开源对语音工具至关重要 | OpenWhispr","description":"语音工具处理着最为私密的数据之一——你真实的声音。这就是为什么开源是唯一能赢得真正信任的方式。","keywords":["开源语音识别","开源听写","开源语音转文字","为什么开源很重要","开源语音隐私","开源 AI 工具","透明的语音软件","开源 whisper","nvidia parakeet"]},"breadcrumb":{"home":"首页","blog":"博客","article":"为什么开源很重要"},"schema":{"title":"为什么开源对语音工具至关重要","description":"语音工具处理着最为私密的数据。这就是为什么开源是唯一能赢得真正信任的方式。"},"layout":{"title":"为什么开源对语音工具至关重要","subtitle":"你的声音是生物特征数据。处理它的软件理应可供审计。","date":"2026年2月10日","tag":"博客"},"sections":[{"id":"fact-check-snapshot","title":"事实核查速览"},{"id":"voice-data-is-different","title":"语音数据与众不同"},{"id":"the-trust-problem","title":"信任难题"},{"id":"what-open-source-gives-you","title":"开源带给你的东西"},{"id":"the-open-model-effect","title":"开放模型效应"},{"id":"open-source-polish","title":"开源同样精致"},{"id":"business-model","title":"商业模式"},{"id":"what-to-look-for","title":"该关注什么"},{"id":"sources","title":"参考来源"}],"opening":{"lead":"开源语音工具让你能够准确核实你的音频究竟经历了什么。","body":"使用专有语音软件时，你是在信任一家公司的隐私政策。而使用开源软件时，你可以阅读代码、自行构建，并确认你的语音数据从未离开过你的设备。这一区别对语音工具的意义，远超几乎任何其他类别的软件，因为语音数据独一无二地私密——它是生物特征，难以匿名化，而且一旦被录制，就无法当作没听过。","lastUpdated":"最后更新：2026年2月17日。涉及事件和政策的论断均已对照至少两个一手来源进行交叉核实。"},"factCheck":{"title":"事实核查速览（双来源）","voiceData":{"lead":"语音数据可识别身份且风险极高：","body":"法律与政策框架日益将生物特征处理视为敏感事项。","source1":"GDPR 官方文本","source2":"欧盟《AI 法案》官方文本"},"assistants":{"lead":"主流语音助手都曾陷入语音数据争议：","body":"Amazon、Google 和 Apple 的相关事件已有监管机构和/或主流新闻媒体记录在案。","source1":"FTC 诉 Amazon 案","source2":"路透社关于 Siri 和解的报道"},"cloning":{"lead":"语音克隆滥用是现实存在的消费者风险：","body":"美国监管机构与网络安全指南均提示了冒充风险。","source1":"FTC 警示","source2":"IBM 风险背景资料"},"openModel":{"lead":"开放模型生态切实存在：","body":"Whisper 及其下游运行时让可审计的本地部署成为可能。","source1":"OpenAI Whisper","source2":"whisper.cpp"},"positioning":{"lead":"OpenWhispr 的定位：","body":"OpenWhispr 在开放 ASR 之上构建，是一套开放、本地优先的语音工作流。","source1":"OpenWhispr","source2":"Whisper 上游"}},"voiceData":{"title":"语音数据与众不同","paragraph1":"文本就是文本。你可以剥离元数据、为人名匿名化，并将其汇总为统计数据。而语音不具备这些特性。","paragraph2":"你的声音是一种生物特征标识。就像指纹一样，它独属于你——银行和政府机构使用的声纹认证系统正是依赖这一事实。但同样的独特性，也使语音录音本身就具有身份识别性。一份泄露的文本转录数据库是一起隐私事件，而一份泄露的语音录音数据库则是一起生物特征数据泄露。","reveal":{"title":"语音录音可能暴露的信息","items":["身份——声纹生物特征可唯一识别说话者","健康状况——声学生物标记可提示帕金森病、抑郁症、呼吸系统疾病","情绪状态——压力、愤怒、疲惫都可从声音模式中被检测出来","人口统计特征——年龄、性别、口音、母语、地域来源"]},"anonymize":{"title":"为何语音更难匿名化","items":["你可以从转录文本中抹去文字，但你无法在不破坏录音的前提下从录音中抹去声音","语音克隆技术意味着泄露的音频可被用来生成你声音的深度伪造","欧盟《AI 法案》将生物特征识别系统——包括语音——归类为高风险 AI，须遵守严格的合规要求"]},"paragraph3a":"语音克隆已从研究奇观变成一项商品化服务。只需几秒钟的音频，现代语音合成模型就能生成以假乱真的人声复刻。这并非假想的风险——","ftcLink":"FTC 已就基于 AI 的语音克隆诈骗发出警告","paragraph3b":"自 2023 年起便是如此。当处理你声音的软件是一个黑箱时，你根本无从知晓录音是否被存储、传输，或被用于模型训练。"},"trustProblem":{"title":"专有语音工具的信任难题","intro":"支持开源的理由并非纸上谈兵。每一款主流语音助手，都曾被发现以用户意料之外的方式处理音频。","alexa":{"title":"Amazon Alexa","body1":"2019年4月，Bloomberg","reportedLink":"报道","body2":"称 Amazon 雇用了全球数千名员工，收听在客户家中和办公室采集到的 Alexa 语音录音。这些员工对音频片段进行转录、标注和审核，以改进语音识别系统。一些审核人员表示曾听到令他们不安的录音，包括疑似性侵的内容。Amazon 在回应中承认了这一做法，但强调它仅适用于“极小的样本”。该公司后来增加了退出选项，但这些录音早已根据大多数用户从未仔细阅读的隐私政策被采集。2023年，Amazon","settlementLink":"支付了 2500 万美元","body3":"以了结 FTC 的指控——该指控称其无限期保留儿童的 Alexa 语音录音，违反了儿童隐私规定。"},"google":{"title":"Google Assistant","body1":"2019年7月，比利时广播机构 VRT NWS","obtainedLink":"获取了超过 1000 段 Google Assistant 录音","body2":"录音来自一名承包商。这些录音包含卧室对话、商务通话以及与儿童的互动——其中许多是在无人说出“OK Google”时因误触发而被采集的。Google 确认人工审核人员收听了约 0.2% 的全部语音录音，并在泄露事件后于欧洲暂停了这一做法。汉堡数据保护机构随后展开的调查，促成了在整个欧盟范围内临时禁止该做法。"},"siri":{"title":"Apple Siri","body1":"2019年7月，一名举报人向 The Guardian 透露，Apple 的承包商","heardLink":"经常听到机密的医疗信息、毒品交易和性接触内容","body2":"——这些都发生在他们为 Siri 录音评级的过程中。Apple 此前从未在其隐私文档中披露过这一人工审核项目。该公司致歉，在全球范围内暂停了该项目，并将其改为需用户主动选择加入——但那些已被审核的录音，是在未经知情同意的情况下采集的。Apple 后来就 Siri 隐私侵权问题以","lawsuitLink":"9500 万美元","body3":"和解了一项集体诉讼，时间为 2025 年 1 月。"},"pattern":"这一模式始终如一：一家公司口口声声“我们高度重视你的隐私”，而实际的数据处理做法却只能通过泄露、举报或监管行动才被揭露。这些可不是偷工减料的小公司。它们是 Apple、Google 和 Amazon——这个星球上技术最尖端、资源最雄厚的三家机构。","quote":"“我们不会存储你的数据”是一个需要信任的说法。而开源是一个只需阅读理解能力的说法。"},"whatOpenSource":{"title":"开源真正带给你的东西","intro":"开源不是一个营销标签。它是一组具体的属性，会改变你与所用软件之间的信任模型。","properties":[{"title":"可审计性","body":"任何人都可以阅读源代码，准确核实音频数据是如何被处理的。这款应用会把录音发送到服务器吗？它会把音频存到磁盘上吗？它会回传遥测数据吗？你不必去信任一份隐私政策——你可以自己查。"},{"title":"可复现性","body":"你可以从源代码构建应用，并将其与发布的二进制文件进行比对。这弥合了“我们公开了代码”与“你下载的应用确实运行的就是这份代码”之间的鸿沟。可复现构建是软件信任的黄金标准。"},{"title":"社区审查","body":"安全研究人员、隐私倡导者和经验丰富的开发者都可以独立审查代码。漏洞被公开地发现并修复。这并非纸面上的好处——正是这一点让 Linux、OpenSSL 以及整个互联网基础设施得以建立在开源之上。"},{"title":"分叉权利","body":"如果维护者做出了你不认同的决定——比如增加遥测、移除功能、卖给收购方——社区可以分叉该项目并独立继续开发。这不仅是一份保险，更是一种结构性的激励，促使维护者以用户的利益为重。"},{"title":"无锁定","body":"你的工作流不会被一家公司的商业决策所绑架。如果一款专有听写工具被收购、转型或关停，你为学习和配置它而投入的心血便会付诸东流。而开源工具只要还有人愿意运行它，就会一直存在。"},{"title":"长久存续","body":"公司会消失，开源项目却会长存。Apache HTTP Server 自 1995 年起运行至今。PostgreSQL 自 1996 年。GCC 自 1987 年。Dragon NaturallySpeaking 曾数十年稳居听写产品的领先地位——然后 Nuance 被 Microsoft 收购，这款独立产品也基本停止了开发。开源不存在这种崩溃模式。"}]},"openModel":{"title":"开放模型效应：从 Whisper 到 Parakeet","paragraph1a":"2022年9月，OpenAI","releasedLink":"发布了 Whisper","paragraph1b":"——一个在 68 万小时多语言音频上训练的通用语音识别模型——并以 MIT 许可证将其开源。这是语音工具领域的一个分水岭时刻。","paragraph2":"在 Whisper 之前，高质量的自动语音识别（ASR）既昂贵又封闭。Google Cloud Speech-to-Text、Amazon Transcribe 和 Microsoft Azure Speech Services 都按音频分钟数收费，并要求将录音发送到它们的服务器。最好的离线方案——CMU Sphinx、Kaldi、VOSK——虽能用，但准确率明显逊于云端 API。","paragraph3":"OpenAI Whisper 一夜之间改变了这一格局。第一次，一个准确率比肩甚至超越商业 API 的模型，可供任何人免费下载并在本地运行，且没有任何数据离开他们的机器。NVIDIA 也走了类似的路线，推出了其 Parakeet 系列 ASR 模型——在英语识别上达到业界领先的准确率，并以开放许可证发布。趋势已经很清楚：最优秀的语音识别模型正越来越多地走向开源。","ecosystem":{"title":"随之而来的生态系统","items":[{"name":"whisper.cpp","desc":"——Georgi Gerganov 的 C/C++ 移植版，针对 CPU 推理进行了优化。让 Whisper 能在从笔记本到树莓派的各种设备上运行。GitHub 星标超过 46,000。"},{"name":"faster-whisper","desc":"——基于 CTranslate2 的重新实现，在保持相同准确率的同时，推理速度最高可达原版的 4 倍。"},{"name":"WhisperX","desc":"——为 Whisper 增加了强制对齐和说话人分离能力，实现词级时间戳和多说话人转录。"},{"name":"构建于开放模型之上的应用","desc":"——如今已有数十款桌面、移动和网页应用以 OpenAI Whisper 或 NVIDIA Parakeet 作为其语音识别引擎，包括 OpenWhispr、MacWhisper、Buzz 和 Vibe。"}]},"paragraph4":"OpenAI Whisper 的发布，是开源如何缔造生态系统最清晰的例证之一。一个开放模型催生了优化运行时、全新能力，以及数十款若模型仍困在 API 付费墙之后便永无诞生之日的产品。NVIDIA Parakeet 延续了这一模式，证明多家机构都看到了将高质量 ASR 开源的价值。如今，任何开发者都能构建一款拥有业界领先准确率的语音工具——无需按分钟付费，无需把音频发给第三方，也无需征得任何人许可。"},"polish":{"title":"开源并不意味着不够精致","intro":"有一种根深蒂固的成见，认为开源意味着粗糙，意味着你要用精致去换取自由。这在 2005 年或许还说得过去，但在 2026 年已不再成立。","apps":[{"name":"Firefox","desc":"一款被数亿人使用的主流浏览器"},{"name":"VS Code","desc":"全世界最受欢迎的代码编辑器"},{"name":"Signal","desc":"端到端加密、用户体验精致的即时通讯应用"},{"name":"Blender","desc":"斩获奥斯卡奖的 3D 动画与视觉特效工具"},{"name":"VLC","desc":"什么都能播，哪儿都能跑，没有广告"},{"name":"Linux","desc":"驱动着全世界绝大多数服务器和智能手机"}],"paragraph1":"开源软件的质量已大幅提升，因为激励结构发生了变化。如今有公司在开源内核之上建立业务（Red Hat、Elastic、GitLab）。资金充裕的团队全职维护项目。社区贡献帮助捕捉缺陷、添加小团队根本无暇顾及的功能。","paragraph2":"客观地说说权衡之处：开源语音工具有时团队规模更小，发布周期也比资金雄厚的专有竞品更慢。文档质量参差不齐。但这些都是现实层面的挑战，而非固有的局限。而且这一差距一直在迅速缩小——尤其在语音工具领域，像 OpenAI Whisper 和 NVIDIA Parakeet 这样的开放模型，让开源项目得以获得与任何商业产品相同的基础模型质量。"},"businessModel":{"title":"开源语音工具如何维持自身运转","intro":"一个合理的问题：如果软件是免费的，又有谁能拿到报酬来开发它呢？","paragraph1a":"答案是，“开源”和“免费”并非同义词。最可持续的开源项目会将核心引擎（免费、开放、可审计）与那些足以支撑付费版的便利功能区分开来。这被称为","openCoreModel":"开放核心模式","paragraph1b":"，它之所以行得通，是因为它让激励保持一致：公司靠把产品做得更好来赚取收入，而不是靠锁定用户或变现他们的数据。","openCore":{"title":"为何开放核心模式对注重隐私的工具尤为奏效","points":["核心产品可本地离线工作——免费版无需任何数据采集","云端功能（如托管的转录 API）提供了值得为之付费的实实在在的便利","社区贡献让产品对所有人都变得更好，付费客户也在其中","无力付费的用户依然能从中受益，并贡献缺陷报告、翻译和代码"]},"paragraph2":"这正是 OpenWhispr 所采用的模式：开源桌面应用使用 OpenAI Whisper 和 NVIDIA Parakeet 等模型在本地完成转录，完全免费。可选的 Pro 套餐为那些希望获得更快结果、或不想在自有硬件上运行模型的用户，增加了云端转录和 AI 文本清理功能。重要的是，选择权在你手中——本地、私密、功能完整的版本永远免费。"},"whatToLookFor":{"title":"挑选开源语音工具时该关注什么","intro":"并非所有“开源”的说法都等价。以下是一份用于评估语音工具的实用清单。","checklist":[{"question":"完整的源代码是否公开？","detail":"有些项目把一个库开源，却将应用本身保持专有。要确认你实际使用的产品——桌面应用、移动应用——其源代码已经公开发布。"},{"question":"你能从源代码构建吗？","detail":"无法编译的公开源代码，在实质意义上算不上开放。要留意是否有构建说明、CI 流水线，以及社区关于成功构建的反馈。"},{"question":"采用什么许可证？（MIT、Apache、GPL、AGPL？）","detail":"宽松许可证（MIT、Apache 2.0）提供最大的灵活性。Copyleft 许可证（GPL、AGPL）则确保衍生作品保持开放。两者都正当合理——只需清楚自己拿到的是什么。"},{"question":"它能离线且本地运行吗？","detail":"需要联网才能工作的开源代码，仍然会把你的数据发送到服务器。对语音工具而言，支持离线的本地处理是隐私的底线。"},{"question":"它是否在被积极维护？","detail":"查看提交历史、问题追踪器和发布节奏。一个被弃置的开源项目，可能存在未修补的安全漏洞。"},{"question":"是否存在社区？（Issues、PR、讨论）","detail":"一个健康的开源项目，贡献者不止一人。要寻找社区参与的迹象——来自外部贡献者的拉取请求、问题讨论，以及响应及时的维护者。"}]},"sources":{"title":"参考来源与延伸阅读","bloomberg":"Bloomberg：《有人在 Alexa 上监听你吗？一支全球团队在审核音频》（2019年4月）","vrt":"VRT NWS：《Google 员工在偷听，连佛兰德人的客厅也不放过》（2019年7月）","guardian":"The Guardian：《Apple 承包商“经常听到机密细节”——来自 Siri 录音》（2019年7月）","reuters":"路透社：《Apple 同意支付 9500 万美元，了结 Siri 隐私诉讼》（2025年1月）","ftcAmazon":"FTC：Amazon 因保留儿童 Alexa 录音被控违反儿童隐私法（2023年5月）","ftcCloning":"FTC：诈骗者利用 AI 语音克隆来强化家庭紧急情况骗局（2023年3月）","openaiWhisper":"OpenAI：Whisper 发布介绍（2022年9月）","whisperGithub":"OpenAI Whisper GitHub 仓库（模型、许可证、文档）","radford":"Radford 等人（2022）：《通过大规模弱监督实现稳健语音识别》","gdpr":"GDPR 官方文本（EU 2016/679）","aiAct":"欧盟《AI 法案》官方文本（EU 2024/1689）","whisperCpp":"whisper.cpp——OpenAI Whisper 的 C/C++ 移植版","fasterWhisper":"faster-whisper——基于 CTranslate2 的 Whisper 实现","openwhispr":"OpenWhispr——开放、本地优先的听写产品背景资料"},"cta":{"title":"OpenWhispr 是开源的","body":"我们以开源方式打造 OpenWhispr，因为我们相信语音软件理应透明。不必听我们的一面之词——自己查查代码吧。","starButton":"在 GitHub 上为我们点星","footnote":"无需账户 · 可离线使用 · MIT 许可 · 永远开源"},"internalLinks":{"localAi":"支持本地 AI 的理由","localVsCloud":"本地转录 vs 云端转录","allPosts":"全部博客文章","compare":"对比 OpenWhispr","useCases":"使用场景"}},"blogGoogleDocsVoiceTyping":{"metadata":{"title":"Google Docs Voice Typing：完整指南| OpenWhispr","description":"如何在 Google Docs 中打开语音输入、每个语音命令、它停止工作的原因，以及如何在任何应用程序（甚至离线）中听写。","keywords":["谷歌文档语音输入","语音输入谷歌文档","语音转文字谷歌文档","谷歌文档听写","打开语音输入 Google 文档","谷歌文档语音命令","谷歌文档语音输入不起作用","谷歌文档语音输入语言"]},"schema":{"title":"Google Docs Voice Typing：完整指南","description":"如何在 Google Docs 中打开语音输入、每个语音命令、它停止工作的原因，以及如何在任何应用程序（甚至离线）中听写。"},"breadcrumb":{"home":"首页","blog":"博客","article":"Google Docs Voice Typing"},"layout":{"title":"Google Docs Voice Typing：完整指南","subtitle":"如何在 Google Docs 中打开语音输入、主要语音命令、为什么它有时会停止工作以及如何在计算机上的其他任何地方进行口述。","date":"2026 年 6 月 26 日","tag":"指导"},"sections":[{"id":"quick","title":"快速解答"},{"id":"enable","title":"如何打开它"},{"id":"commands","title":"语音命令"},{"id":"languages","title":"语言"},{"id":"mobile","title":"在手机和平板电脑上"},{"id":"troubleshooting","title":"不工作？"},{"id":"limits","title":"局限性"},{"id":"alternative","title":"随时随地语音输入"},{"id":"faq","title":"常问问题"}],"intro":{"lead":"要在 Google Docs 中打开语音输入，请在 Chrome 中打开文档，转到“工具”>“语音输入”，或者在 Windows 上按 Ctrl+Shift+S 或在 Mac 上按 Cmd+Shift+S，然后单击麦克风并开始讲话。","body":"它是免费的、内置的，对于快速初稿非常有用。代价是它只能在桌面上的 Chrome 中运行，需要互联网连接，并且只能在 Google Docs 中输入。","summary":"下面您将找到设置步骤、重要的命令类别、支持的语言注释、故障排除修复、隐私限制以及需要在文档之外进行听写时的实用选项。"},"enable":{"title":"如何在Google Docs中开启语音输入","intro":"语音输入是 Google Docs 桌面版的一部分，专为 Google Chrome 设计。这是完整的过程：","steps":["在Chrome中打开docs.google.com处的文档。","在菜单栏中，单击工具 > 语音输入。快捷方式：Windows 上为 Ctrl+Shift+S，Mac 上为 Cmd+Shift+S。","页面左侧会出现一个麦克风框。单击它即可开始。","第一次，您的浏览器会请求麦克风权限。单击“允许”。","说话自然。当您说话时，您的话会出现在光标处。","当您想停止时，再次单击麦克风。"],"tipLead":"提示：","tipBody":"语音输入也适用于 Google Slides 的发言者备注字段，具有相同的工具菜单和麦克风控制。"},"commands":{"title":"语音命令：标点符号、格式设置和编辑","intro":"您可以说出标点符号、换行符和编辑命令。重要的一点是：格式化和编辑命令仅在文档语言设置为英语时才有效。","groups":[{"title":"标点","items":["\"period\"","\"comma\"","\"question mark\"","\"exclamation point\"","\"new line\" / \"new paragraph\""]},{"title":"格式化","items":["\"bold\" / \"italics\"","\"underline\"","\"apply heading 1\"","\"create bulleted list\"","\"align center\""]},{"title":"编辑和移动","items":["\"select [word]\"","\"delete last word\"","\"go to end of paragraph\"","\"next paragraph\"","\"stop listening\""]}]},"languages":{"title":"支持哪些语言？","paragraph1":"语音输入可以处理大约一百种语言和方言的听写。从麦克风盒的下拉菜单中选择您的。基本标点符号广泛适用于各种语言。","paragraph2":"语音格式化和编辑命令是例外。粗体、选择段落或应用标题等命令仅适用于英语，因此非英语听写通常意味着使用键盘或鼠标进行格式化。"},"mobile":{"title":"在手机和平板电脑上进行语音输入","intro":"Google Docs 移动应用程序不包含桌面语音输入工具。在手机和平板电脑上，您可以像在任何应用程序中一样进行口述：点击键盘上的麦克风。","items":["Android：当光标位于文档中时，点击 Gboard 上的麦克风或您喜欢的键盘。","iPhone和iPad：点击iOS键盘上的听写键。"],"outro":"这是您的设备听写，而不是 Google Docs 桌面功能，因此准确性、命令和行为将与 Chrome 版本不匹配。"},"troubleshooting":{"title":"语音输入不起作用？试试这个","items":[{"problem":"语音输入按钮缺失或呈灰色","fix":"在桌面上使用Google Chrome。移动应用程序不支持语音输入，并且不正式支持其他浏览器。"},{"problem":"它不会拾取我的声音","fix":"检查docs.google.com在Chrome中是否有麦克风权限，并确保其他应用没有持有麦克风。"},{"problem":"它总是自行停止","fix":"语音输入可能会在暂停后或单击远离选项卡时停止。再次单击麦克风并保持“文档”选项卡处于焦点状态。"},{"problem":"什么也没发生","fix":"您可能处于离线状态。语音输入基于云，需要互联网连接。重新连接并重新加载文档。"}]},"limits":{"title":"Google Docs Voice Typing不能做什么","intro":"这是一个真正有用的免费功能，但它是为一项工作而构建的：在 Google 文档中进行口述。如果您经常口述，这些限制很重要：","items":[{"title":"仅限Chrome。","body":"Safari、Firefox 或 Edge 中没有可靠的支持。"},{"title":"仅限在线。","body":"音频在 Google 云中处理，因此没有连接就意味着没有听写。"},{"title":"仅文档。","body":"它不会输入到 Gmail、Slack、浏览器栏、代码编辑器或任何其他应用程序中。"},{"title":"云隐私。","body":"您的声音会发送到 Google 进行转录，这对于敏感材料来说并不理想。"}]},"alternative":{"title":"想要每个应用程序都支持语音输入吗？使用OpenWhispr","introBefore":"如果您只能在文档中进行口述，那么语音输入即可完成这项工作。但是，当您想要对电子邮件、Slack 消息、代码注释、表单或笔记应用程序进行对话时，","introAfter":"涵盖了 Google Docs 没有涵盖的内容：","points":["在系统范围内的任何应用程序或文本字段中输入一键通热键","使用 OpenAI Whisper 和 NVIDIA Parakeet 在您的设备上 100% 运行，完全离线","您的音频永远不会离开您的计算机","适用于任何浏览器和桌面应用程序，而不仅仅是 Chrome","开源，所以任何人都可以检查它的功能","智能体模式可以清理并重新格式化您口述的内容"],"outro":"您可以将其视为整个计算机的 Google Docs 语音输入，无需依赖云。","viewOnGitHub":"在GitHub上查看"},"faqTitle":"常见问题解答","faqs":[{"question":"如何在Google Docs中开启语音输入？","answer":"在 Chrome 中打开文档，然后转到“工具”>“语音输入”，或者在 Windows 上按 Ctrl+Shift+S 或在 Mac 上按 Cmd+Shift+S。单击麦克风框，出现提示时允许使用麦克风，然后开始讲话。"},{"question":"为什么我的 Google Docs 语音输入不起作用？","answer":"通常的原因是使用 Chrome 以外的浏览器、缺少麦克风权限、另一个应用程序持有麦克风、暂停或未聚焦的“文档”选项卡或没有互联网连接。"},{"question":"Google Docs语音输入可以离线使用吗？","answer":"不需要。Google Docs 语音输入基于云，需要互联网连接。如果需要离线听写，可以使用OpenWhispr等本地工具。"},{"question":"Google Docs语音输入支持哪些语言？","answer":"它支持大约一百种语言和方言的听写，但语音格式和编辑命令仅在文档语言为英语时才有效。"},{"question":"Google Docs语音输入可以在iPhone或Android上使用吗？","answer":"不像桌面语音输入工具。在移动应用程序中，请改用手机键盘麦克风，例如 Android 上的 Gboard 或 iOS 听写键。"},{"question":"Google Docs语音输入私密吗？","answer":"音频会发送到 Google 进行转录，因此很方便，但不限于本地。对于机密笔记、法律工作、患者信息或未发布的业务内容，本地听写更适合。"},{"question":"我可以在 Google Docs 之外使用语音输入吗？","answer":"Google 的内置语音输入功能仅适用于 Google Docs 中，而演讲者注释则适用于 Google Slides 中。 OpenWhispr 提供了一个系统范围的热键，可以命令任何应用程序。"}],"links":{"items":[{"href":"/use-cases","label":"所有用例"},{"href":"/compare/apple-dictation","label":"VS Apple Dictation"},{"href":"/pricing","label":"定价"}]}},"useCaseMedical":{"breadcrumb":{"home":"Home","useCases":"Use Cases","medical":"For Clinicians"},"metadata":{"title":"Free Medical Dictation Software For Clinicians - OpenWhispr","description":"OpenWhispr is free, open-source medical dictation software. Dictate the visit, get the SOAP note drafted, and keep patient audio on your machine.","keywords":["medical dictation software","clinical dictation software","physician dictation","HIPAA compliant dictation software","dictation software for doctors","SOAP note dictation","offline medical transcription"]},"schema":{"name":"OpenWhispr","applicationCategory":"Medical dictation software","operatingSystem":"macOS, Windows, Linux","description":"Local-first, open-source medical dictation software. Patient audio is transcribed on the clinician's own device and never leaves it. Free for unlimited local dictation.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Medical dictation software for doctors & clinicians","title":"Dictate the visit once. Get the note drafted.","body":"OpenWhispr drafts the SOAP note, the referral letter or the discharge summary from what you dictate, on your own computer. Free for every clinician, and it keeps working on a locked-down hospital network.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Clinician walking through a sunlit hospital corridor"}},"trustBar":{"label":"Dictate into:","names":["Epic","Oracle Health","athenahealth","eClinicalWorks","NextGen","Meditech"]},"pain":{"label":"Cloud or local. Choose for the visit.","title":"A full clinic should not end with an hour of notes.","subtitle":"Dictate the findings, plan and follow-up while each visit is fresh, then review and sign the draft instead of rebuilding the day at home. Choose Cloud for speed or local processing when patient audio needs to stay on your machine.","cards":[{"title":"The whole encounter, uploaded.","body":"An ambient scribe can send the whole encounter to the cloud, including things the patient asked you not to put in the chart. OpenWhispr processes what you dictate on your device."},{"title":"Pajama-time charting.","body":"A full clinic, then an hour of notes at the kitchen table. OpenWhispr drafts the note from your dictation while the visit is fresh, so you review and sign instead of type."},{"title":"One more vendor for compliance to vet.","body":"Each cloud scribe adds a BAA, a risk assessment and another data flow to map. In local mode, no transcription vendor sits in the data path."},{"title":"\"Trust us\" is not a compliance posture.","body":"OpenWhispr's code is public on GitHub. Give your security team the repository, then choose the cloud or local path that fits your review."}]},"features":{"label":"What you get","title":"Faster charting. Same confidentiality.","subtitle":"Medical dictation software often trades privacy for speed or bills per provider every month. OpenWhispr keeps the speed without sending patient audio away or charging per clinician for local mode.","items":[{"image":{"label":"Clinician speaking beside a laptop in a medical practice"},"title":"Your mic, or the telehealth call, transcribed on-device","body":"Dictate the visit into your mic, or record a telehealth consult straight from your speakers on macOS 14.2 or later, Windows and most Linux desktops. It's transcribed locally and never uploaded. You decide what's kept and when it's deleted."},{"image":{"label":"Clinician reviewing a printed document at a hospital workstation"},"title":"SOAP notes and letters, drafted for you","body":"Save your own SOAP, referral and discharge prompts as one-click actions, then run whichever one the visit needs. You review every word before it's signed, then paste it into the chart."},{"image":{"label":"Doctor speaking while working on a laptop in an examination room"},"title":"Dictate into the chart field you're already in","body":"Press a hotkey and dictate wherever the cursor already is: the HPI box, a reply to a portal message, a refill note in your inbox. Clean, punctuated text, processed locally."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Patient audio sent to us","value":"None","footnote":"in local mode, transcription never leaves your device"},{"label":"Cost per clinician","value":"$$0","footnote":"unlimited local dictation, the whole practice, forever"},{"label":"Local dictation is","value":"Free","footnote":"unlimited on every plan"}]},"howItWorks":{"label":"How it works","title":"The note drafted before the next patient is roomed.","subtitle":"Most medical dictation is a per-provider cloud subscription or an outsourced transcription service. OpenWhispr lives on your machine, with no account or credit card required.","image":{"label":"Doctor reaching for a closed laptop in an examination room"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Pick a local model during setup."},{"title":"Dictate the visit.","body":"Speak your findings, plan and follow-up, and it's transcribed on your device."},{"title":"Review, sign, delete.","body":"Edit the draft note, paste it into the chart, and delete the audio. You control all of it."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons clinicians choose OpenWhispr","subtitle":"Choosing medical dictation software is a patient-confidentiality decision before it's anything else. Here are the facts, and your compliance officer can verify every one in the repo or at trust.openwhispr.com.","items":[{"title":"Audio never leaves your device","body":"Dictation, and note drafts with a local model, are processed on your machine. The open-source code shows there's no upload step."},{"title":"Open source, auditable by anyone","body":"Your IT or compliance team can read every line on GitHub. No faith required."},{"title":"Retention is yours, not a vendor's","body":"Audio and transcripts are files on your machine, not rows in someone else's retention schedule. Delete the audio the moment the note is signed."},{"title":"Optional Corti clinical transcription","body":"For specialist terminology, Corti Transcribe is available as a clinical-grade medical transcription provider. Choose a local model for note drafting that stays on-device."},{"title":"No bot joins the telehealth call","body":"Nothing appears in the participant list, so you never have to explain a third party to the patient before you can begin."},{"title":"Works fully offline","body":"Ward round, rural clinic, a locked-down hospital network. No signal required, so it never stalls mid-dictation."},{"title":"Dictate into any EHR","body":"A system-wide hotkey drops clean text wherever your cursor is: Epic, athenahealth, a referral letter, an email."},{"title":"Free for the whole practice","body":"Unlimited local dictation at $0, every clinician in the group, forever. No per-provider licence to renew, and you can test it on your own clinic list before anyone signs anything."},{"title":"SOC 2 Type 2 audit complete. HIPAA attestation. ISO 27001-aligned.","body":"Your security review can start at trust.openwhispr.com. Enterprise includes custom data processing agreements and compliance reviews."},{"title":"100+ languages","body":"Dictate the history in the language the visit actually happened in, transcribed locally."}]},"faq":{"label":"Before you download","title":"The questions clinicians ask first.","subtitle":"Straight answers on privacy, compliance and accuracy. If yours isn't here, ask us in Discord.","items":[{"question":"Does patient audio ever leave my device?","answer":"Not in local mode. Pick a local model during setup and transcription runs on your machine; the open-source code shows there is no upload step. Cloud transcription remains optional."},{"question":"Is it HIPAA compliant?","answer":"Compliance depends on your workflow, not the tool alone. OpenWhispr's SOC 2 Type 2 audit is complete, with the final report available under NDA once issued; it has a HIPAA attestation and an ISO 27001-aligned information-security programme. For PHI, local-only and bring-your-own-key modes route content around OpenWhispr Cloud. Cloud voice audio is processed in real time and not persisted after transcription. Enterprise includes custom data processing agreements and compliance reviews."},{"question":"Can it handle medical terminology?","answer":"Yes. It runs Whisper-class models trained on a wide range of accents and 100+ languages. The custom dictionary feeds the words that trip everything else up to both the transcriber and the AI cleanup: metoprolol succinate ER, cholecystectomy, the abbreviations your specialty writes and nobody else reads."},{"question":"How is this different from an ambient AI scribe?","answer":"An ambient scribe listens to the whole encounter, sends it to a vendor's cloud, and bills per provider every month. Medical dictation software is the other shape: you say what belongs in the note, a local model drafts it from your own prompt, and it stays on your machine at $0. For a telehealth consult it captures your speakers directly, so nothing joins the call and nothing is uploaded."},{"question":"Does it draft the note, or just transcribe?","answer":"Both. Run a local model (Llama, Qwen, Gemma) or connect your own AI key, then save your own SOAP, referral and discharge prompts as one-click actions and run whichever one the visit needs. A local model keeps the PHI on your machine; your own key means the model and data path are ones you already trust."},{"question":"Is it actually free?","answer":"Yes. Unlimited local dictation, 100+ languages, all three desktop platforms: $0 forever, for every clinician. Pro ($8/month) adds more cloud transcription and mobile dictation if you want them."}]},"reviews":{"label":"What users say","title":"Built by people who don't want to be recorded either.","subtitle":"From our Discord, GitHub and inboxes, the same instinct that matters in the exam room.","score":{"value":"0","label":"Patient recordings on a server","body":"In local mode, audio stays on your device and is never stored on a server."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for doctors & clinicians","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Clinicians who want free, private, offline dictation and are fine with a quick local setup"}]},{"parts":[{"text":"Clinicians who need unlimited cloud transcription, cross-device sync and mobile access"}]},{"parts":[{"text":"Practices and teams that need unlimited meeting recordings, speaker labels and Agent mode"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Free offline dictation, with patient audio staying on your device"}]},{"parts":[{"text":"Unlimited cloud transcription without managing local models or keys"}]},{"parts":[{"text":"Unlimited meeting recordings, speaker labels and workflow features for a team"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some conversations stay in the exam room. We built for those.","paragraphs":["OpenWhispr started with a simple objection: what a patient tells you in the exam room shouldn't have to pass through a stranger's server to become a note.","Our code is public on GitHub. In local mode, transcription runs on the device with no internet required.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Sign the last note before you leave the building.","body":"Download OpenWhispr. Dictate the visit, draft the note, then paste it into the chart while it is still fresh. Choose local models when processing needs to stay on your computer.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free unlimited local dictation on Mac, Windows & Linux."}}},"authGate":{"providers":{"google":"Google","microsoft":"Microsoft","apple":"Apple"},"continueWithProvider":"继续 {provider}","or":"或","fields":{"email":"电子邮件地址","password":"密码","fullName":"全名","passwordHint":"至少 8 个字符"},"actions":{"checking":"正在检查...","continueWithEmail":"继续使用电子邮件","back":"返回","signingIn":"正在登录...","signIn":"登录","createAccount":"创建帐户","tryAgain":"再试一次","sending":"正在发送...","sendResetLink":"发送重置链接","creatingAccount":"正在创建帐户...","resendVerificationEmail":"重新发送验证邮件","useDifferentEmail":"使用不同的电子邮件","useSso":"使用单点登录 (SSO)"},"status":{"completingSignIn":"正在完成登录...","redirecting":"正在重定向..."},"errors":{"generic":"出了点问题。请再试一次。","authenticationTimedOut":"身份验证超时。请再试一次。","sessionExpired":"会话已过期。请尝试重新注册。","waitingForVerification":"仍在等待验证。检查您的收件箱或重新发送电子邮件。","unableToCheckEmail":"无法检查此电子邮件。请再试一次。","network":"网络错误。请检查您的连接。","networkRetry":"网络错误。请检查您的连接并重试。","incorrectPassword":"密码不正确。请再试一次。","accountExists":"使用此电子邮件的帐户已存在。请登录。","createAccountFailed":"创建帐户失败。请再试一次。","resendFailed":"重发失败。请再试一次。"},"signin":{"title":"欢迎回来","forgotPassword":"忘记密码？","newHere":"新来的？"},"forgot":{"title":"重置您的密码","description":"输入您的电子邮件，我们将向您发送重置链接","emailSentTitle":"检查您的电子邮件","emailSentBody":"我们将密码重置链接发送至","didNotReceive":"没有收到电子邮件？检查您的垃圾邮件文件夹或重试。"},"signup":{"title":"创建您的帐户","haveAccount":"有帐户吗？"},"verifyEmail":{"title":"检查您的电子邮件","message":"我们将验证链接发送至"},"legal":{"line":"继续即表示您同意我们的<termsLink>服务条款</termsLink>和<privacyLink>隐私政策</privacyLink>。"}},"getStarted":{"title":"开始使用","subtitle":"下载 OpenWhispr，免费且开源。","processing":"开始下载...","skip":"无需帐户即可继续","verificationMessage":"请验证邮箱以开始下载。链接已发送至","metadata":{"title":"开始使用 | OpenWhispr","description":"创建 OpenWhispr 账户，开始使用私密的 AI 语音听写。"}},"checkout":{"title":"开始使用 OpenWhispr","processing":"正在重定向至结帐...","period":{"month":"/月","year":"/年"},"plan":{"proMonthly":"Pro · 每月","proAnnual":"Pro·年度","businessMonthly":"Business · 每月","businessAnnual":"Business·年度"},"badge":{"twoMonthsFree":"2个月免费"},"metadata":{"title":"结账 | OpenWhispr","description":"安全地完成 OpenWhispr 订阅购买。"}},"checkoutSuccess":{"appRedirect":{"title":"正在重定向至 OpenWhispr...","body":"您可以关闭此选项卡","fallback":"如果应用程序未打开，请点按此处"},"success":{"title":"欢迎来到Pro！","body":"您的订阅现已激活"},"download":{"autoNotice":"您的下载应该会自动开始...","fallback":"如果未启动，请单击此处","open":"下载OpenWhispr","all":"查看所有下载"},"linuxOptions":[{"label":"Linux AppImage (x86_64)","meta":"AppImage"},{"label":"Linux DEB (amd64)","meta":"Debian 软件包"},{"label":"Linux RPM (x86_64)","meta":"RPM封装"},{"label":"Linux tar.gz (x64)","meta":"便携式压缩包"}],"metadata":{"title":"结账成功 | OpenWhispr","description":"OpenWhispr 结账已完成，你的订阅现已可用。"}},"upgrade":{"expiredTitle":"链接已过期","redirecting":"正在重定向至结帐...","errors":{"missingToken":"缺少升级令牌。请检查您的电子邮件链接。","invalidLink":"该链接不再有效。您可以从应用程序内升级。","generic":"出了点问题。请重试或从应用程序内升级。"},"metadata":{"title":"升级 | OpenWhispr","description":"升级你的 OpenWhispr 套餐，解锁更多语音听写功能。"}},"authCallback":{"failedTitle":"登录失败","calendarFailedTitle":{"gcal":"无法连接 Google 日历","mcal":"无法连接 Microsoft 日历"},"redirecting":"重定向至OpenWhispr","closeTab":"您可以关闭此选项卡","desktopFallback":"如果OpenWhispr没有打开请点击这里","mobileFallback":"如果应用程序未打开，请点按此处","backToSignIn":"返回登录","desktopNextStep":"请返回 OpenWhispr 应用重新登录。","errors":{"access_denied":"未获得权限，所以您未登录。如果您拒绝了请求，请重新登录并选择允许。如果需要您所在组织的 IT 管理员先批准 OpenWhispr，请让他们允许后重新登录。","account_not_linked":"此电子邮件地址已有 OpenWhispr 帐户，且使用其他登录方式。请用您第一次登录的方式登录，例如用密码。","state_mismatch":"此登录链接已过期或已被使用，所以您未登录。请重新开始登录，并在 5 分钟内完成。","state_not_found":"此登录链接已过期或已被使用，所以您未登录。请重新开始登录，并在 5 分钟内完成。","please_restart_the_process":"此登录链接已过期或已被使用，所以您未登录。请重新开始登录，并在 5 分钟内完成。","invalid_state":"此登录链接已过期或已被使用，所以您未登录。请重新开始登录，并在 5 分钟内完成。","invalid_code":"登录已开始但未能完成，所以您未登录。请重新登录。如果再次失败，请发邮件至 support@openwhispr.com。","unable_to_get_user_info":"我们无法从您的 Google、Microsoft 或 Apple 帐户读取您的姓名和电子邮件地址，所以您未登录。请重新登录。","noSession":"无法确认您的登录，所以 OpenWhispr 应用仍未登录。请返回应用重新登录。","default":"出了点问题，您未登录。请重新登录。如果一直失败，请发邮件至 support@openwhispr.com。","sso_required":"贵公司要求使用自己的登录方式，即 SSO。请返回 OpenWhispr 应用，用工作邮箱登录。应用会使用贵公司的 SSO。","account_mismatch":"您登录的帐户与您输入的帐户不同。请用输入的帐户登录，或继续使用已登录的帐户。","consent_required":"您所在组织的 IT 管理员尚未批准 OpenWhispr。请让他们允许后再登录。现在重试无效。","admin_policy_enforced":"您所在组织的 IT 管理员尚未批准 OpenWhispr。请让他们允许后再登录。现在重试无效。","user_cancelled_authorize":"您在完成前关闭了登录窗口，所以您未登录。请重新开始登录。","server_error":"您用于登录的服务那边出了问题，登录未能完成。请等几分钟后重新登录。","slack_connect_invalid":"此 Slack 连接链接不完整或无效，因此 Slack 尚未连接。请回到 OpenWhispr 应用重新连接 Slack。","linear_connect_invalid":"此 Linear 连接链接不完整或无效，因此 Linear 尚未连接。请回到 OpenWhispr 应用重新连接 Linear。"},"errorTitles":{"cancelled":"登录已取消","permissionNotGranted":"未获得权限","approvalNeeded":"需要批准","linkExpired":"登录链接已过期","companySignIn":"需要公司登录","wrongAccount":"帐户不符","slackConnectFailed":"无法连接 Slack","linearConnectFailed":"无法连接 Linear"},"calendarErrors":{"access_denied":"访问被拒绝。请再试一次并授予日历权限。","no_email":"无法读取此日历账户的电子邮件地址。请再试一次。","token_exchange_failed":"无法完成与您的日历提供商的连接。请再试一次。","server_error":"连接时出错。请检查网络连接后再试一次。","admin_policy_enforced":"需要您所在组织的管理员批准 OpenWhispr 访问日历。","consent_required":"需要您所在组织的管理员批准 OpenWhispr 访问日历。","default":"无法连接您的日历。请再试一次。"},"metadata":{"desktopCallback":{"title":"正在重定向… | OpenWhispr","description":"正在完成 OpenWhispr 桌面版登录并将你重定向到应用。"},"mobileCallback":{"title":"正在重定向… | OpenWhispr","description":"正在完成 OpenWhispr 移动版登录并将你重定向到应用。"},"error":{"title":"登录错误 | OpenWhispr","description":"无法完成 OpenWhispr 登录。请查看错误并重试。","connectTitle":"{heading} | OpenWhispr"}},"mismatch":{"switchAccount":"使用其他帐户登录","continueAnyway":"继续使用已登录的帐户"}},"notFound":{"title":"哎呀！找不到页面","subtitle":"看起来此页在转录过程中丢失了！","quip":"别担心，我们思考的速度比打字的速度快......","description":"您要查找的页面可能已被移动、删除或暂时不可用。","home":"带我回家","back":"返回","autoRedirectLabel":"自动重定向：","autoRedirectBody":"{seconds} 秒带你回家..."},"verifyEmail":{"loading":"正在验证您的电子邮件...","errors":{"noToken":"未找到验证令牌。请检查您的电子邮件链接。","failed":"验证失败。该链接可能已过期。","network":"网络错误。请检查您的连接并重试。"},"resend":{"signInFirst":"请先登录，然后请求新链接。","failed":"无法重新发送。请再试一次。","sent":"验证邮件已发送！检查您的收件箱。","network":"网络错误。请再试一次。"},"success":{"title":"邮箱已验证！","desktop":"如果您没有自动重定向，请点击下面。","browser":"您可以关闭此选项卡并返回到您的注册以继续。","openApp":"打开OpenWhispr"},"actions":{"sending":"正在发送...","requestNewLink":"请求新链接"},"metadata":{"title":"验证邮箱 | OpenWhispr","description":"验证你的电子邮箱地址，以完成 OpenWhispr 账户设置。"}},"resetPassword":{"loading":"正在加载...","errors":{"passwordMismatch":"密码不匹配。","passwordTooShort":"密码必须至少为 8 个字符。","network":"网络错误。请检查您的连接并重试。"},"invalid":{"title":"重置链接无效","description":"此密码重置链接无效或已过期。","cta":"转到开始使用"},"success":{"title":"密码已更新","description":"您现在可以使用新密码登录。","cta":"继续登录"},"form":{"title":"创建新密码","description":"在下面输入您的新密码","newPassword":"新密码","confirmPassword":"确认新密码","passwordHint":"至少 8 个字符","resetting":"正在重置...","submit":"重置密码"},"metadata":{"title":"重置密码 | OpenWhispr","description":"为你的 OpenWhispr 账户设置新密码。"}},"consent":{"banner":{"ariaLabel":"Cookie 同意","title":"我们重视您的隐私","eu":"我们使用 cookie 来衡量使用情况和我们的营销。在您允许之前，分析和营销 cookie 会保持关闭状态。","us":"我们和我们的合作伙伴使用 cookie 进行分析和广告。您可以随时选择退出。","row":"我们使用 cookie 来了解使用情况并改进 OpenWhispr。","gpc":"我们检测到 Global Privacy Control 信号并已选择让您退出广告 cookie。","policyLine":"请查看我们的<cookieLink>Cookie 政策</cookieLink>和<privacyLink>隐私政策</privacyLink>。","rejectAll":"全部拒绝","customize":"定制","acceptAll":"全部接受","privacyChoices":"您的隐私选择","cookieSettings":"Cookie 设置","gotIt":"明白了","updated":"自您上次访问以来，我们更新了 Cookie 政策。您当前的选择将持续有效，直到您进行更改。"},"settings":{"title":"Cookie 偏好设置","description":"选择 OpenWhispr 可以使用哪些 cookie。您可以随时从页脚更改此设置。","necessary":{"title":"绝对必要","description":"登录和安全所需。始终活跃。","ariaLabel":"绝对必要的cookie（始终开启）"},"analytics":{"title":"分析","description":"帮助我们理解用法（Google Analytics 4）。","ariaLabel":"分析 cookies"},"marketing":{"title":"营销","description":"衡量我们的广告和营销活动（Google Ads、Meta）以及创作者推荐（Dub）的效果。","gpcDescription":"关闭 — 您的浏览器的 Global Privacy Control 信号会选择您退出。","ariaLabel":"营销cookie"},"rejectAll":"全部拒绝","acceptAll":"全部接受","save":"保存首选项"},"trigger":{"privacyChoices":"您的隐私选择","cookieSettings":"Cookie 设置"}},"referral":{"someone":"有人","loading":"正在加载...","invitedBy":"{name} 邀请","invited":"你被邀请了","title":"解锁 1 个月免费 OpenWhispr Pro","claimPrompt":"注册即可领取 Pro 的免费一个月","continueWithGoogle":"继续Google","or":"或","processing":"正在设置您的帐户...","alreadyHaveAccount":"已经有帐户？","downloadOpenWhispr":"下载OpenWhispr","fields":{"email":"电子邮件地址","password":"密码","fullName":"全名","passwordMin":"密码（最少 8 个字符）"},"actions":{"checking":"正在检查...","continueWithEmail":"继续使用电子邮件","signingIn":"正在登录...","signInClaim":"登录并领取优惠","useDifferentEmail":"使用不同的电子邮件","creatingAccount":"正在创建帐户...","createClaim":"创建帐户并领取优惠","useSso":"使用单点登录 (SSO)"},"signin":{"as":"登录身份"},"signup":{"for":"创建帐户"},"legal":{"line":"继续即表示您同意我们的<termsLink>条款</termsLink>和<privacyLink>隐私政策</privacyLink>。"},"errors":{"network":"网络错误。请再试一次。","incorrectPassword":"密码不正确。请再试一次。","accountExists":"帐户已存在。请登录。","createAccountFailed":"创建帐户失败。","signInTimedOut":"登录超时，请重试。","signInFailed":"登录失败。","ssoEmailRequired":"请在上方输入你的工作邮箱以使用单点登录。","ssoFailed":"单点登录失败。"},"spectrogram":{"friend":"一个朋友","signalId":"信号ID","scanAriaLabel":"扫描音频签名","scan":"扫描","stop":"停止","replay":"重播","unlocked":"解锁","freeMonth":"1 个月免费 Pro","signature":"{name}的音频签名","scanning":"正在扫描..."},"continueWithProvider":"使用 {provider} 继续","providers":{"google":"Google","microsoft":"Microsoft","apple":"Apple"},"metadata":{"title":"推荐邀请 | OpenWhispr","description":"通过推荐链接加入 OpenWhispr，开始使用私密的 AI 语音听写。"}},"demo":{"ariaLabel":"产品演示","tabpanelFallback":"演示","tabs":[{"label":"语音转文本"},{"label":"会议记录"},{"label":"AI 对话"},{"label":"音频上传"}],"panelLabels":["语音转文本演示","会议记录演示","人工智能聊天演示","音频上传演示","自定义词典演示","集成演示","历史演示"],"sidebar":{"search":"搜索","settings":"设置","userName":"Alex Chen","userEmail":"alex@company.com","nav":{"home":"首页","notes":"注释","upload":"上传","dictionary":"词典","integrations":"集成"}},"speech":{"dictationText":"嘿，团队，我刚刚审查完 Q2 路线图，我认为我们应该在新集成之前优先考虑语音管道。","withoutOpenWhispr":"没有 OpenWhispr","slowWpm":"~40 WPM","messageApp":"Slack — #product","placeholder":"输入消息...","withOpenWhispr":"使用 OpenWhispr","fastWpm":"~150 WPM","faster":"速度提高 3 倍"},"history":{"transcriptions":{"title":"转录","items":[{"time":"上午 9:15","text":"季度指标看起来很有希望。收入环比增长 12%，用户保留率显着提高。"},{"time":"下午 4:30","text":"让我们安排下周的设计评审。我希望整个团队在发货前看到新组件。"}]},"upcomingMeetings":{"title":"即将召开的会议","items":[{"time":"上午 11:00 - 上午 11:30","title":"设计评审"},{"time":"下午 2:00 - 下午 2:45","title":"冲刺计划"},{"time":"下午 4:00 - 下午 4:30","title":"1:1 与 Sarah"}]}},"notes":{"rawNotes":"- 3月份发布语音管道\n- 与设计团队讨论新用户界面\n- 预算问题：询问财务","meetingTitle":"冲刺计划","title":"产品同步","recording":"录音","you":"你","date":"2026 年 3 月 11 日","foldersLabel":"文件夹","notesLabel":"笔记","viewTabs":{"transcript":"转录稿","notes":"笔记","enhanced":"已增强"},"sections":[{"heading":"决定","items":["语音流程是最高优先级：在其他工作流之前，于 3 月底前发布","UI 重新设计推迟到语音流程落地之后，避免分散注意力","批准的词典功能：医学、法律和技术术语的自定义单词"]},{"heading":"行动项目","items":["Sarah：在冲刺结束时前端迁移到新的组件库","Alex：API 重构 + 延迟基准（目标 sub-200ms，当前 ~280ms）","Gabriel：跟进 Q2 预算的财务情况，需要在周五之前批准人数","新编辑器 UI 的设计评审 - 下周二下午 2 点，整个团队"]},{"heading":"开放性问题","items":["笔记上的实时协作 — CRDT 与 OT 方法？还没有时间表","公开开源发布之前需要第三方依赖审核","如何处理移动设备的离线同步——本地优先还是基于队列？"]}],"done":"完成","primaryAction":"套用模板","enhanceNotes":"增强笔记","actions":[{"name":"同步模板","description":"使用团队模板格式化"},{"name":"会议纪要","description":"总结决策和行动"},{"name":"电子邮件草稿","description":"后续电子邮件草稿"}],"folders":[{"name":"个人","count":5,"isActive":true},{"name":"会议","count":3,"isActive":false},{"name":"工作","count":2,"isActive":false}],"noteList":[{"title":"产品同步","preview":"每周同步的行动项目...","time":"上午 10:30"},{"title":"语音管道设计","preview":"v2 的架构说明...","time":"上午 9:15"},{"title":"每周站立会","preview":"冲刺进度和阻碍...","time":"昨天"}]},"aiChat":{"title":"人工智能聊天","listening":"听...","askAnything":"询问有关您笔记的任何问题","speakQuestion":"说出你的问题","voicePoweredQa":"语音驱动的数据问答","userMessage":"总结一下我们今天会议讨论的内容","response":"根据今天的讨论：\n\n• 语音管道 — 3 月底发货\n• UI 重新设计推迟到后期流程\n• 跟进Q2 预算的财务情况","releaseToFinish":"释放 {hotkey} 完成","holdToSpeak":"按住{hotkey}说话"},"upload":{"noteTitle":"设计系统迁移规划","transcript":"该团队同意继续推进新的设计系统。 Sarah 将领导前端迁移，Alex 将在冲刺结束时处理 API 重构。","now":"现在","preview":"团队一致致力于迁移到新的设计系统...","dateMeta":"2026 年 3 月 12 日 · 847 字","structuredIntro":"团队一致同意迁移到新的基于组件的设计系统。前端迁移以 Sarah 领导组件审核开始本次冲刺。 Alex 将并行重构 API 层。","sections":[{"heading":"关键决策","items":["在所有产品表面采用新的设计标记","前端迁移逐个组件处理","API 重构范围为身份验证和数据获取层"]},{"heading":"行动项目","items":["Sarah — 组件审核和迁移计划（周五截止）","Alex — 范围 API 重构；星期三之前起草 RFC","Jamie — 更新 CI 管道以获得新的令牌格式","团队 — 在下一次站立之前回顾 Figma 库 v2"]},{"heading":"时间轴","items":["Sprint 14：审核 + 迁移计划 + API RFC","Sprint 15：组件迁移、设计令牌","Sprint 16：集成测试，弃用旧版本"]}],"dropPrompt":"拖放音频文件，或点击浏览","formats":"MP3、WAV、M4A、WebM、OGG、FLAC、AAC","fileName":"meeting-recording.mp3","fileMeta":"12:04 持续时间 · 18.3 MB","folder":"个人","transcribe":"转录","transcribing":"本地转录...","complete":"转录完成","openNote":"打开笔记","newUpload":"新建上传"},"dictionary":{"title":"自定义词典","clearAll":"全部清除","placeholder":"添加单词或短语...","hint":"单词作为提示发送，以提高转录准确性。用逗号分隔多个单词。"},"integrations":{"title":"集成","subtitle":"连接您最喜欢的工具","calendarDescription":"自动加入会议并实时转录","connected":"已连接","comingSoon":"更多集成即将推出"}},"blogComponents":{"author":{"defaultRole":"工程","updated":"更新了 {date}"},"layout":{"defaultTag":"博客","tableOfContents":"目录","onThisPage":"在此页面上"},"diarizationPipeline":{"title":"本地说话人分离管道","steps":[{"label":"音频捕捉","sub":"麦克风+系统，16kHz"},{"label":"VAD","sub":"Silero, 2MB"},{"label":"细分","sub":"pyannote 3.0"},{"label":"嵌入","sub":"CAM++, 512-dim"},{"label":"聚类","sub":"Agglomerative, 0.5"},{"label":"标签","sub":"匹配 SQLite 中的配置文件"}],"whyLocal":"为什么它保持本地化：","bullets":["每个阶段都通过单个 sherpa-onnx 二进制文件在您的设备上运行 - 没有 Python，没有云。","ONNX 模型中的 ~45MB 下载一次，然后缓存在 ~/.cache/openwhispr/diarization-models/. 中","嵌入是 512-dim float32 向量，在本地磁盘上存储为 SQLite BLOBs 。"]},"linuxSpectrum":{"title":"Linux 听写景观 (2026)","left":"精致的用户体验","right":"DIY控制"},"windowsSpectrum":{"title":"Windows 听写景观 (2026)","left":"人工智能驱动","right":"专业化"},"localVsCloud":{"title":"本地与云：关键权衡","local":{"title":"本地（OpenWhispr)","points":[{"label":"隐私","detail":"音频保留在设备上"},{"label":"延迟","detail":"无网络往返"},{"label":"成本","detail":"没有按分钟 API 计费"},{"label":"离线","detail":"无需互联网即可工作"},{"label":"控制","detail":"确定性局部行为"}]},"cloud":{"title":"云 APIs","points":[{"label":"隐私","detail":"策略+配置依赖"},{"label":"延迟","detail":"网络和区域敏感"},{"label":"成本","detail":"基于使用情况的计费"},{"label":"离线","detail":"不可用"},{"label":"规模","detail":"轻松应对大批量工作负载"}]},"recommendation":"最佳实用模式：本地优先，仅在需要时将边缘情况路由到云。"},"openSourceTrust":{"title":"为什么开源可以提高语音数据信任","layers":[{"title":"可审计性","detail":"直接在代码中检查数据流"},{"title":"再现性","detail":"从源代码构建并验证二进制文件"},{"title":"分叉权利","detail":"如果维护者改变方向，用户仍能保持控制权"},{"title":"无硬锁定","detail":"迁移堆栈而不丢失工作流程"}]},"voicePrivacyFlow":{"title":"语音数据路径：本地与云风险面","microphone":"麦克风","localInference":"局部推理","localText":"本地文本","zeroExposure":"零暴露","cloudApi":"云API","remoteText":"远程文本","networkExposure":"网络曝光","defaultPath":"OpenWhispr 默认：本地路径优先，仅在显式配置时可选云。"},"whisperArchitecture":{"title":"Whisper 如何处理音频","steps":["音频输入","Log-Mel 特性","编码器","解码器","文字"],"whyItMatters":"为什么它对 OpenWhispr: 很重要","bullets":["通过 whisper.cpp 在本地运行 — 原始音频保留在设备上。","模型大小控制速度/准确性的权衡。","相同的管道支持 macOS、Windows 和 Linux 之间的听写。"]},"werChart":{"title":"词错误率：Whisper 与开源竞争对手","subtitle":"LibriSpeech test-clean（英语朗读演讲）——越低越好","werLabel":"WER","turboNote":"速度提高 8 倍","otherOpenSource":"其他开源","sources":"来源：Hugging Face 模型卡、NVIDIA 模型卡、Open ASR Leaderboard (Feb 2026)。仅限纯英语朗读语音 - 现实世界的准确性因音频质量、口音和领域而异。商业 API 被省略，因为它们不发布 LibriSpeech 基准。"},"modelTradeoff":{"title":"Whisper 模型大小：速度与准确性","subtitle":"气泡大小反映了相对参数数量","ariaLabel":"Whisper 模型速度与精度权衡的散点图","fasterAxis":"更快 →","accuracyAxis":"更准确 →","recommendation":"建议：从小处开始，根据硬件进行调整。"},"openAsrWerChart":{"averageWer":"平均 WER","title":"英语准确率：平均词错误率","caption":"Hugging Face Open ASR Leaderboard 基准测试中多个数据集的平均值（越低越好）。Nemotron 以流式配置展示。","sources":"来源：Hugging Face 上的 NVIDIA 模型卡和 Open ASR Leaderboard，2026 年 7 月。WER = 转写错误的词所占百分比。"},"streamingAsrFlow":{"title":"OpenWhispr 中的实时流式处理路径"}},"compareFathom":{"metadata":{"title":"Fathom 替代方案——本地、私密、开源","description":"在寻找 Fathom 替代方案？OpenWhispr 在本地处理语音转文字和会议笔记，无需机器人、云端或账户。免费且开源。","keywords":["Fathom 替代方案","Fathom AI 替代方案","私密 Fathom 替代方案","Fathom 开源替代方案","本地会议笔记","无机器人会议笔记"]},"breadcrumb":{"home":"首页","comparisons":"对比","fathom":"对比 Fathom"},"hero":{"title":"私密、开源的 Fathom 替代方案","description":"OpenWhispr 是为希望在自己的设备上 100% 运行<voice>语音转文字</voice>和<meetings>会议笔记</meetings>的用户打造的 Fathom 替代方案。没有机器人加入通话，音频绝不离开设备，整个应用均为开源。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"速览","fathomIntro":"是一款精致的云端会议笔记工具，以慷慨的免费方案闻名。它的机器人会加入 Zoom、Teams 和 Meet 通话，录音并转录，几乎无需设置即可生成清晰的摘要、重点和待办事项。它是最容易上手的自动会议回顾工具之一。","openWhisprIntro":"专注于 Fathom 不提供的能力：让音频留在自己的设备上。这款语音转文字与设备端会议笔记工具基于 OpenAI Whisper 和 NVIDIA Parakeet。没有机器人加入，不需要账户，也不会上传任何内容。按下快捷键即可在任意应用中语音输入，或在本地录制并转录会议。","chooseFathom":"如果你想要最轻松的自动会议摘要、强大的免费方案，并且不介意云端处理，请选择 Fathom。","chooseOpenWhispr":"如果你想要私密、本地的语音转文字和会议笔记，并且绝不上传音频，请选择 OpenWhispr。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"隐私与数据控制","openWhispr":{"status":"full","details":"100% 本地——会议音频留在设备上"},"competitor":{"status":"none","details":"云端处理——录音存储在服务器上"}},{"feature":"开源","openWhispr":{"status":"full","details":"MIT 许可，可在 GitHub 审查"},"competitor":{"status":"none","details":"专有闭源"}},{"feature":"离线运行","openWhispr":{"status":"full","details":"本地模型无需互联网"},"competitor":{"status":"none","details":"需要互联网连接"}},{"feature":"会议捕获","openWhispr":{"status":"partial","details":"本地录制系统音频，无机器人"},"competitor":{"status":"full","details":"机器人自动加入 Zoom、Teams 和 Meet"}},{"feature":"通用听写","openWhispr":{"status":"full","details":"在任意应用中进行系统级语音输入"},"competitor":{"status":"none","details":"仅限会议，并非听写工具"}},{"feature":"转录与语言","openWhispr":{"status":"full","details":"Whisper + Parakeet，支持 100+ 种语言"},"competitor":{"status":"full","details":"云端转录与 AI 摘要"}},{"feature":"免费方案","openWhispr":{"status":"full","details":"本地听写无限使用，永久免费"},"competitor":{"status":"full","details":"面向个人的慷慨免费方案"}},{"feature":"团队与 CRM 功能","openWhispr":{"status":"partial","details":"API/MCP 访问；Business 基础团队功能"},"competitor":{"status":"full","details":"付费方案提供 CRM 同步与团队功能"}}],"whyOpenWhispr":{"title":"为什么选 OpenWhispr？","items":[{"title":"默认本地运行：","body":"转录在你的设备上运行。参与者列表中没有机器人，不上传音频，也不存入第三方服务器。"},{"title":"随处听写：","body":"系统级快捷键可将语音输入任何应用——邮件、文档、编辑器或 AI 聊天。Fathom 仅用于会议。"},{"title":"开源：","body":"采用 MIT 许可且可审查，你可以准确验证音频的处理方式。"}]},"whyFathom":{"title":"为什么选 Fathom？","items":[{"title":"轻松生成会议回顾：","body":"机器人自动加入并总结通话，会议结束时重点和待办事项已准备就绪。"},{"title":"慷慨的免费方案：","body":"个人用户云端会议笔记工具中功能最强的免费方案之一。"},{"title":"精致且配置简单：","body":"清晰的摘要和 CRM 同步，多数团队几乎无需配置即可采用。"}]},"faq":{"title":"常见问题"},"faqs":[{"question":"最佳 Fathom 替代方案是什么？","answer":"这取决于你的需求。如果你需要私密、开源的日常语音转文字和设备端会议笔记，OpenWhispr 是可靠选择：它使用 OpenAI Whisper 和 NVIDIA Parakeet 在本地 100% 处理语音，通过按键说话快捷键在整个系统中工作，绝不将音频发送到云端。如果你主要想要最轻松的自动会议摘要和慷慨的免费方案，Fathom 本身非常出色。"},{"question":"有免费的 Fathom 替代方案吗？","answer":"有。OpenWhispr 免费且开源。在 macOS、Windows 和 Linux 上，可离线免费使用 100+ 种语言的无限本地听写。可选 Pro（$8/月）增加无限云端转录、更多会议录制时长、移动端和设备同步，但核心本地语音转文字永远不需要它。"},{"question":"有本地或私密的 Fathom 替代方案吗？","answer":"OpenWhispr 正是为此而建。核心转录在你的设备上进行，音频绝不离开设备，并可完全离线工作。应用开源，因此可准确审查数据处理方式。通话中没有机器人，也不会上传内容，这对人力资源、法律或涉及客户数据的敏感会议尤为重要。"},{"question":"OpenWhispr 会像 Fathom 一样让机器人加入会议吗？","answer":"不会，这正是关键。Fathom 会让机器人出现在通话中，并将音频上传到云端。OpenWhispr 直接在设备上捕获会议的系统音频，因此参与者列表中没有机器人，也不会上传任何内容。你仍应告知他人正在录音，但对话完全留在自己的设备上。"},{"question":"OpenWhispr 与 Fathom 有何不同？","answer":"Fathom 是云端会议笔记工具，专注于轻松的自动摘要和强大的免费方案。OpenWhispr 是隐私优先的语音转文字与设备端会议笔记工具：按下快捷键即可在任意应用中说话，或在本地录制和转录会议。两者解决不同问题；很多人用云端工具处理日常通话，用 OpenWhispr 进行私密听写和敏感会议。"}],"cta":{"title":"绝不上传音频的 Fathom 替代方案","description":"私密、开源的语音转文字和设备端会议笔记。在 macOS、Windows 和 Linux 上免费使用。","viewOnGitHub":"在 GitHub 上查看"}},"featureDictationSoftware":{"metadata":{"title":"面向专业人士的开源语音输入软件 - OpenWhispr","description":"OpenWhispr 是免费的开源语音输入软件，可在任何应用中将语音转换为文字。支持离线使用和本地运行，让语音数据留在你的设备上。","keywords":["听写软件","语音听写软件","免费听写软件","语音转文字软件","语音输入","听写应用","离线听写","本地听写软件"]},"breadcrumb":{"home":"首页","dictationSoftware":"听写软件"},"structuredData":{"description":"开源听写软件。在设备上私密运行，或使用 OpenWhispr Cloud 获得最高准确度，在任何应用中将语音转成文字，支持 100+ 种语言。","offerDescription":"免费开始。本地听写不限量，可选 Pro 和 Business 方案。","featureList":["在任何应用中使用全系统按键听写","本地离线语音转文字，以及高准确度的 OpenWhispr Cloud","100+ 种语言，内置翻译","OpenAI Whisper 和 NVIDIA Parakeet 模型","用于姓名、术语和缩写的自定义词典","内置 AI 写作助手，可整理和润色听写内容","在 macOS、Windows 和 Linux 上开源"]},"hero":{"badge":"免费开始 · 开源","title":"适用于所有应用的免费听写软件","description":"按下快捷键并说话，带标点的清晰文字就会出现在邮件、文档、Slack、编辑器或 AI 聊天中。可在设备上私密运行，也可启用 OpenWhispr Cloud 获得最高准确度。无论哪种方式，音频都不会被存储。","github":"在 GitHub 上查看","platforms":"在 macOS、Windows 和 Linux 上免费开始 · 无需银行卡或账户"},"trustBar":{"title":"在你常用的每个应用中听写"},"painPoints":{"eyebrow":"用户为何切换","title":"大多数听写软件都会让你有所取舍。","description":"用速度换隐私、用隐私换准确度，或让整个工作流受制于一家公司的云。OpenWhispr 让你无需选择。","items":[{"title":"你的想法比手指更快。","body":"说话速度远快于打字。OpenWhispr 以思考的速度把语音变成清晰文字，让邮件、提示词和草稿不再受键盘限制。"},{"title":"你看不到多数语音工具如何处理你的声音。","body":"多数听写软件是闭源且仅限云端。OpenWhispr 开源；无论本地还是 OpenWhispr Cloud，音频只会被转写，不会被存储、出售或用于训练，代码可查。"},{"title":"你的配置不该被某个应用或操作系统锁定。","body":"封闭工具会把工作流锁在其平台。OpenWhispr 可在 Mac、Windows 和 Linux 的所有应用中运行，并支持你自己的 AI 密钥。"},{"title":"系统自带听写还不够。","body":"Apple Dictation 和 Windows Speech 准确度不足、依赖网络且绑定单一系统。OpenWhispr 在三个平台提供 Whisper 级识别和自定义词典。"}]},"features":{"eyebrow":"你将获得","title":"一个快捷键，连接你的声音与所有应用。","description":"光标在哪里，就在哪里听写：写邮件、向 AI 发出提示词、记录想法，无需复制粘贴。","items":[{"title":"在任何应用中听写","body":"按下快捷键、说话、松开。带标点的文字会出现在邮件、文档、Slack 或代码编辑器的光标处。"},{"title":"与 AI 对话，不必打字","body":"在 ChatGPT、Claude 或 Cursor 中口述包含丰富上下文的长提示词，以最快方式提供 AI 所需细节。"},{"title":"准确识别你的用词","body":"通过自定义词典教会 OpenWhispr 识别姓名、术语和缩写，然后选择本地引擎或 OpenWhispr Cloud。"}]},"stats":{"items":[{"value":"100+","label":"听写语言","foot":"并内置翻译"},{"value":"0","label":"存储的录音","foot":"本地音频不离开设备；云端音频转写后即丢弃"},{"value":"5分钟","label":"设置时间","foot":"从下载到说出第一句话"}]},"howItWorks":{"eyebrow":"使用方法","title":"三步即可与电脑对话。","description":"无需云账户，无需注册。软件安装在你的电脑上，约五分钟即可运行。","items":[{"title":"免费下载。","body":"支持 Mac、Windows 或 Linux。选择本地模型或 OpenWhispr Cloud，无需账户或银行卡。"},{"title":"设置快捷键。","body":"一个快捷键，可按住录音或按键切换，并在全系统所有应用中生效。"},{"title":"按下。说话。完成。","body":"支持 100+ 种语言，清晰且带标点的文字会出现在光标处。"}]},"reasons":{"eyebrow":"事实无需说服","title":"选择 OpenWhispr 听写的 10 个理由","description":"切换工具需要信任。这里的每项说明都可在开源仓库中验证。","items":[{"title":"适用于所有应用","body":"全系统快捷键。只要能放置光标，就能听写。"},{"title":"隐私优先且可审计","body":"开源且不保留数据：本地音频留在设备，云端音频转写后丢弃。"},{"title":"本地或云端，由你决定","body":"在设备上运行以获得完整隐私，或用 OpenWhispr Cloud 获得准确度和速度。"},{"title":"Whisper 和 Parakeet 级准确度","body":"领先的开源语音转文字模型，并提供适合普通硬件的云端选项。"},{"title":"属于你的自定义词典","body":"添加姓名、术语和缩写，让转写第一次就正确。"},{"title":"100+ 种语言和翻译","body":"用一种语言说话，以另一种语言输出，可指定或自动检测。"},{"title":"使用自己的 AI 密钥","body":"使用 OpenAI、Anthropic、Gemini 或 Groq 密钥进行无加价的云端转写。"},{"title":"边说边整理","body":"内置 AI 写作助手可在本地或云端整理和润色听写内容。"},{"title":"Mac、Windows 和 Linux","body":"少数可在三个平台原生运行且功能一致的听写应用之一。"},{"title":"减轻双手负担","body":"如果打字缓慢、疲劳或疼痛 — RSI、疲劳、长时间工作 — 就让声音代劳。"}]},"faq":{"eyebrow":"下载之前","title":"关于 OpenWhispr 听写软件的常见问题","items":[{"question":"OpenWhispr 能在所有应用中使用吗？","answer":"可以。快捷键在全系统生效，包括邮件、文档、Slack、浏览器、代码编辑器和 AI 聊天。能打字的地方就能听写。"},{"question":"语音转文字有多准确，包括不同口音吗？","answer":"它使用 Whisper 级模型和 NVIDIA Parakeet，针对多种口音和 100+ 种语言训练。可选择更大的本地模型、添加术语或启用 OpenWhispr Cloud。"},{"question":"本地听写与云端听写有何不同？","answer":"本地模式在设备上运行，完全私密、离线且不限量。OpenWhispr Cloud 无需下载模型，提供高准确度和速度，音频转写后不存储。免费版每周 2,000 个云端单词，Pro 不限量。"},{"question":"它免费吗？哪些功能收费？","answer":"应用免费且开源：100+ 种语言的本地听写不限量，含词典、每周 2,000 个 OpenWhispr Cloud 单词和 5 小时会议录音。Pro ($8/月) 增加无限云端转写、20 小时、移动端和同步。"},{"question":"我的声音会离开设备吗？","answer":"本地模式绝不会，开源代码可以证明。使用 OpenWhispr Cloud 或 API 密钥时，音频仅为转写而发送，随后丢弃。我们不会存储、出售数据或用其训练。"},{"question":"能在 Mac、Windows 和 PC 上使用吗？","answer":"可以。OpenWhispr 是少数原生支持 macOS、Windows 和 Linux 的开源听写应用之一，三个平台使用相同引擎和功能。"},{"question":"它能学习我的词汇和术语吗？","answer":"可以。自定义词典让 OpenWhispr 学会姓名、技术术语和缩写，首次即可正确转写；所有本地或云端方案均包含。"},{"question":"这是 AI 听写吗？能格式化和清理文字吗？","answer":"可以。内置 AI 写作助手可用本地或云端模型添加标点、重新格式化和改写；智能体模式还能自动处理转写内容。"},{"question":"它与 Dragon 或系统自带听写有何区别？","answer":"Dragon 功能强但昂贵且以 Windows 为主；Apple 和 Windows 听写依赖网络和单一系统。OpenWhispr 支持三个平台，开源，并可选择本地或云端。"},{"question":"运行需要什么？支持离线吗？","answer":"一台较新的 Mac、Windows 或 Linux 电脑即可。小型本地模型可在普通硬件上运行，安装后完全离线。OpenWhispr Cloud 无需下载模型。"}]},"testimonials":{"eyebrow":"用户评价","title":"告别打字的人们。"},"pricing":{"eyebrow":"价格","title":"免费开始。升级即可获得云端速度和更多功能。","description":"本地听写永久免费且不限量。Pro 和 Business 增加无限 OpenWhispr Cloud、会议时长、移动端和团队功能。"},"openSource":{"eyebrow":"公开构建","title":"以你希望的方式处理你的声音。","description":"OpenWhispr 源于一个简单质疑：多数听写软件在闭源代码背后收听一切。因此我们公开构建，每个版本和每行代码都在 GitHub 上，已有 5,000+ 名工程师点赞、复刻和审查。无论本地还是云端，我们都不存储、出售或用你的数据训练，代码可以证明。","github":"在 GitHub 上点赞"},"finalCta":{"title":"你的下一句话不需要键盘。","description":"下载 OpenWhispr，在电脑上的任何应用中听写 — 支持 100+ 种语言，隐私优先，五分钟即可开始。","footer":"在 Mac、Windows 和 Linux 上免费开始 · 无需银行卡或账户"},"demo":{"dictationText":"团队好，我刚刚审阅完 Q2 路线图，我认为应该在新集成之前优先处理语音管线。","without":"不使用 OpenWhispr","typingSpeed":"~40 WPM","windowTitle":"Slack — #product","messagePlaceholder":"输入消息…","with":"使用 OpenWhispr","dictationSpeed":"~150 WPM","faster":"快 3x","transcribing":"正在转写…","listening":"正在聆听…","hotkeyPrompt":"按下快捷键并说话…","indicator":"按住快捷键，说话后松开 — 文字会出现在当前应用中。"}},"compareFireflies":{"metadata":{"title":"Fireflies.ai 替代方案 — 私密、本地、开源","description":"私密、开源的 Fireflies.ai 替代方案。语音转文字和会议笔记 100% 在本地运行，无机器人加入通话，也不上传音频。","keywords":["fireflies 替代方案","fireflies.ai 替代方案","私密 fireflies 替代方案","开源 fireflies 替代方案","本地会议笔记","无机器人会议笔记"]},"breadcrumb":{"home":"首页","comparisons":"对比","fireflies":"对比 Fireflies.ai"},"hero":{"title":"私密、开源的 Fireflies.ai 替代方案","description":"OpenWhispr 是为希望在自己的设备上 100% 运行<voice>语音转文字</voice>和<meetings>会议笔记</meetings>的用户打造的 Fireflies.ai 替代方案。没有机器人加入通话，音频绝不离开你的电脑，整个应用均为开源。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"快速了解","firefliesIntro":"是一款云端会议助手。它的机器人会加入 Zoom、Teams 和 Meet 通话，录音并转写内容，生成摘要和待办事项，并为销售和营收团队提供对话智能与 CRM 集成。它确实很擅长这项工作。","openWhisprIntro":"围绕另一项优先需求打造：隐私和所有权。它是一款语音转文字和设备端会议笔记工具，使用 OpenAI Whisper 和 NVIDIA Parakeet，完全在你的电脑上运行。没有机器人加入，无需账户，也不会上传任何内容。按下快捷键，即可在任何应用中输入你说的话；也可以在本地录制会议音频并在设备上转写。","chooseFireflies":"如果你需要可自动加入通话并提供团队分析和 CRM 同步的机器人，请选择 Fireflies.ai。","chooseOpenWhispr":"如果你需要私密、本地且绝不上传音频的语音转文字和会议笔记，请选择 OpenWhispr。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"隐私与数据控制","openWhispr":{"status":"full","details":"100% 本地 — 会议音频保留在你的设备上"},"competitor":{"status":"none","details":"基于云端 — 音频和转写稿存储在服务器上"}},{"feature":"开源","openWhispr":{"status":"full","details":"采用 MIT 许可证，可在 GitHub 上审查"},"competitor":{"status":"none","details":"专有闭源软件"}},{"feature":"离线工作","openWhispr":{"status":"full","details":"本地模型无需联网即可运行"},"competitor":{"status":"none","details":"需要互联网连接"}},{"feature":"会议采集","openWhispr":{"status":"partial","details":"在本地录制系统音频 — 无机器人加入"},"competitor":{"status":"full","details":"机器人自动加入 Zoom、Teams 和 Meet 通话"}},{"feature":"通用听写","openWhispr":{"status":"full","details":"在任意应用中进行系统级语音输入"},"competitor":{"status":"none","details":"专注于会议，并非听写工具"}},{"feature":"转写与语言","openWhispr":{"status":"full","details":"Whisper + Parakeet，100+ 种语言"},"competitor":{"status":"full","details":"支持 AI 搜索的云端转写"}},{"feature":"CRM 与团队智能","openWhispr":{"status":"partial","details":"API/MCP 访问；Business 提供基础团队功能"},"competitor":{"status":"full","details":"Salesforce/HubSpot 同步与对话分析"}},{"feature":"免费方案","openWhispr":{"status":"full","details":"无限本地听写，永久免费"},"competitor":{"status":"partial","details":"免费方案有限制 — 请查看其网站"}}],"whyOpenWhispr":{"title":"为什么选择 OpenWhispr？","items":[{"title":"默认本地运行：","body":"转写在你的电脑上运行。参与者列表中没有机器人，不上传音频，也不会在第三方服务器上存储任何内容。"},{"title":"随处听写：","body":"一个系统级快捷键即可将语音输入任何应用，包括电子邮件、文档、编辑器或 AI 聊天。Fireflies 仅用于会议。"},{"title":"开源：","body":"采用 MIT 许可证且可供审查，你可以准确验证音频的处理方式。"}]},"whyFireflies":{"title":"为什么选择 Fireflies.ai？","items":[{"title":"自动加入通话：","body":"机器人会自行加入预定通话并自动采集，无需你动手操作。"},{"title":"CRM 与营收智能：","body":"为销售团队提供深入的 Salesforce/HubSpot 同步和对话分析。"},{"title":"共享团队工作区：","body":"开箱即用，在整个组织中提供可搜索、可协作的转写稿。"}]},"faqs":[{"question":"最佳 Fireflies.ai 替代方案是什么？","answer":"这取决于你的需求。如果你想要一款私密、开源的工具，用于日常语音转文字和设备端会议笔记，OpenWhispr 是很好的选择：它使用 OpenAI Whisper 和 NVIDIA Parakeet 在本地 100% 运行语音识别，通过一键说话快捷键在整个系统中工作，并且绝不会将音频发送到云端。如果你特别需要能自动加入通话并提供团队分析和 CRM 集成的机器人，Fireflies.ai 本身仍然非常出色。"},{"question":"有免费的 Fireflies.ai 替代方案吗？","answer":"有。OpenWhispr 免费且开源。在 macOS、Windows 和 Linux 上，你可以免费离线使用支持 100+ 种语言的无限本地听写。可选的 Pro 方案（$8/月）增加无限云端转写、更多会议录制时长、移动端和设备同步，但使用核心的本地语音转文字功能完全不需要它。"},{"question":"有本地或私密的 Fireflies.ai 替代方案吗？","answer":"OpenWhispr 正是为此打造。核心转写在你自己的电脑上进行，音频绝不离开设备，并且完全支持离线工作。由于应用开源，你可以准确审查数据的处理方式。通话中没有机器人，也不会上传任何内容，这对于人力资源、法律或涉及客户数据的敏感会议尤其重要。"},{"question":"OpenWhispr 会像 Fireflies 一样通过机器人加入会议吗？","answer":"不会，这正是关键所在。Fireflies 会派出一个出现在通话中的机器人，并将音频上传至其云端。OpenWhispr 则直接在你的设备上采集会议的系统音频，因此参与者列表中没有机器人，也不会上传任何内容。你仍需告知他人正在录音，但对话会完全保留在你的电脑上。"},{"question":"OpenWhispr 与 Fireflies.ai 有何不同？","answer":"Fireflies.ai 是一款云端会议助手，专注于为销售和营收团队自动加入通话、提供对话智能和 CRM 同步。OpenWhispr 则是一款隐私优先的语音转文字工具，并提供设备端会议笔记：按下快捷键即可在任意应用中说话，或在本地录制并转写会议。两者解决不同的问题：许多人会保留机器人工具用于协作式销售通话，同时使用 OpenWhispr 处理私密的日常听写和敏感会议。"}],"faq":{"title":"常见问题"},"cta":{"title":"绝不上传音频的 Fireflies.ai 替代方案","description":"私密、开源的语音转文字和设备端会议笔记。可在 macOS、Windows 和 Linux 上免费使用。","viewOnGitHub":"在 GitHub 上查看"}},"unsubscribe":{"metadata":{"title":"取消订阅 — OpenWhispr","description":"管理您的 OpenWhispr 营销邮件订阅。","keywords":["取消订阅 OpenWhispr","电子邮件偏好设置","营销邮件"]},"breadcrumb":{"home":"首页","unsubscribe":"取消订阅"},"invalid":{"title":"此链接无效","description":"取消订阅链接缺失或已损坏。请尝试使用较新邮件中的链接，或直接联系我们将您从邮件列表中移除。","contact":"请发送邮件至 <email>support@openwhispr.com</email>，我们会为您处理。"},"success":{"title":"您已取消订阅","description":"<email>{emailAddress}</email> 将不再收到 OpenWhispr 的营销邮件。此更改已立即生效。","serviceMessages":"您仍会收到收据、安全警报和密码重置等服务邮件，因为这些邮件与您的账户相关。"},"confirm":{"title":"取消订阅营销邮件？","description":"<email>{emailAddress}</email> 将不再收到 OpenWhispr 的产品动态和优惠。与您账户相关的服务邮件不受影响。","button":"取消订阅"}},"legalDpa":{"metadata":{"title":"数据处理附录 | OpenWhispr","description":"OpenWhispr数据处理附录（Gizmo Labs Inc.作为数据处理者），无需签署即生效，包含不使用数据训练AI的承诺、SCC及UK Addendum传输条款和产品次级处理者附件。"},"breadcrumb":{"home":"首页","dpa":"数据处理附录"},"authoritativeNotice":"本译文仅为方便阅读而提供。如与英文版本存在任何不一致，应以英文版本为准。","header":{"title":"数据处理附录","subtitle":"Gizmo Labs Inc.（OpenWhispr）如何代表其客户处理个人数据。客户内容绝不会用于训练AI模型（第3条）。","effective":"生效日期：2026年7月17日 · 版本1.2（2026年8月21日更新）"},"preamble":{"title":"关于本附录","body":"本数据处理附录（\"DPA\"）构成OpenWhispr服务条款的一部分并通过引用纳入其中；如客户（\"客户\"、\"数据控制者\"）与Gizmo Labs Inc.之间另有主服务协议或企业协议（\"协议\"），则构成该协议的一部分并通过引用纳入其中。凡Gizmo Labs Inc.（一家在Delaware注册成立、注册地址为221 W 9th Street, Wilmington, Delaware 19801的公司，称\"OpenWhispr\"、\"数据处理者\"）在提供OpenWhispr服务（\"服务\"）过程中代表客户处理个人数据，本DPA即适用。本DPA无需签署即生效；如客户采购流程要求已签署文本，可使用所附签署栏。如本DPA与协议在数据保护事项上冲突，以本DPA为准；如本DPA与SCC（或UK Addendum）就受限传输发生冲突，在冲突范围内以SCC（或UK Addendum）为准。本DPA项下通知应发送至双方的账户联系人，发送给OpenWhispr的通知还应发送至support@openwhispr.com。"},"definitions":{"title":"1. 定义","body":"本文件未定义的首字母大写术语具有数据保护法律赋予的含义。\"数据保护法律\"指适用于协议项下个人数据处理的所有法律，包括欧盟GDPR（条例(EU) 2016/679）、英国GDPR及Data Protection Act 2018、瑞士FADP和适用的美国州隐私法（包括CCPA/CPRA）。\"SCC\"指欧盟委员会实施决定(EU) 2021/914中的Standard Contractual Clauses。\"UK Addendum\"指ICO的International Data Transfer Addendum to the EU SCCs。\"受限传输\"指向未获充分性决定的第三国传输个人数据。\"次级处理者\"指OpenWhispr为代表数据控制者处理个人数据而聘用的第三方。"},"scope":{"title":"2. 范围和角色","body":"OpenWhispr仅作为数据处理者代表客户处理个人数据。处理的事项、期限、性质和目的、个人数据类型及数据主体类别载于附件1。为免疑义：当最终用户仅在本地运行OpenWhispr或使用自己的模型API密钥（\"bring your own key\"）时，音频和转录/推理内容从用户设备直接发送至用户选择的提供商，或在设备上处理；这些内容流由用户指示，所涉第三方提供商依用户指示行事，而非作为OpenWhispr的次级处理者。无论采用何种模式，账户数据、云同步（如启用）、使用和账单数据仍由OpenWhispr及其次级处理者按照本DPA处理。OpenWhispr还可为运营、保护和改进服务而处理经汇总及去标识化、已不再构成个人数据的数据（例如服务性能指标），且不会尝试重新识别。"},"noAiTraining":{"title":"3. 不使用客户数据训练AI","body":"OpenWhispr不会使用客户个人数据（包括语音音频、转录文本、笔记、提示词及与AI代理的对话）训练、重新训练或改进任何人工智能或机器学习模型，无论该模型属于OpenWhispr还是第三方，并且不允许其次级处理者这样做。附件3所列模型提供商仅为提供服务而处理此类内容，OpenWhispr账户的训练功能已禁用或选择退出。本限制在协议终止后继续有效。"},"processorObligations":{"title":"4. 数据处理者义务","intro":"OpenWhispr将：","items":["仅依客户的书面指示（协议、本DPA及客户对服务的配置构成此类指示）处理个人数据，包括进行传输，但OpenWhispr所受法律要求处理的除外；在该情况下，除法律禁止外，OpenWhispr将在处理前将该法律要求告知客户；","确保获准处理个人数据的人员受保密义务约束；","实施附件2所列技术和组织安全措施，并定期审查；","考虑处理性质，就数据主体请求、安全、泄露通知、数据保护影响评估和事先咨询向客户提供协助（第6–9条）；","在服务终止时根据客户选择删除或返还全部个人数据，并在30日内删除现有副本，但OpenWhispr所受法律要求保留的除外；","提供合理必要的信息以证明合规，并按照第10条允许并协助审计；以及","如OpenWhispr认为某项指示违反数据保护法律，立即通知客户。"]},"subprocessing":{"title":"5. 次级处理","body":"客户一般性书面授权OpenWhispr聘用次级处理者。现行名单载于附件3，并发布于OpenWhispr信任中心<trustLink>trust.openwhispr.com</trustLink>（Subprocessors部分）。OpenWhispr向每位次级处理者施加与本DPA同等的数据保护义务（包括传输义务），并继续对其履行负责。OpenWhispr至少提前30日向客户书面通知拟新增或替换的次级处理者（向客户账户所有者发送电子邮件并更新公开名单）；客户可在此期间基于合理且有书面依据的数据保护理由提出异议。如双方不能解决异议，客户可终止受影响的服务，并获得已预付费用中与所终止服务相应的按比例退款。"},"security":{"title":"6. 安全","body":"OpenWhispr维持附件2中的措施，并在考虑技术发展水平和个人数据性质后确保其与风险相适应。通过云端转录路径处理的语音音频会被实时处理，转录后不会由OpenWhispr保存。"},"breach":{"title":"7. 个人数据泄露","body":"OpenWhispr在获悉影响客户个人数据的个人数据泄露后，将不无故拖延地、并在任何情况下于72小时内通知客户，提供客户为履行自身通知义务而合理需要的信息，并配合补救。"},"dataSubjectRequests":{"title":"8. 数据主体请求","body":"考虑处理性质，OpenWhispr将通过适当的技术和组织措施协助客户回应数据主体行使权利（访问、更正、删除、限制、可携带、反对）的请求。最终用户可在产品内自行导出和删除；OpenWhispr直接收到且能识别客户组织的请求将不无故拖延地转交客户。"},"dpias":{"title":"9. 数据保护影响评估和事先咨询","body":"在OpenWhispr掌握所需信息的范围内，OpenWhispr就数据保护影响评估及与监管机构的事先咨询提供合理协助。"},"audit":{"title":"10. 审计","body":"应请求，OpenWhispr将在保密前提下通过信任中心提供现行合规文件（SOC 2 Type 2报告，涵盖2026年5月1日至2026年7月31日期间，可依NDA提供；通过其合规平台出具的GDPR和HIPAA证明，2026；ISO/IEC 27001:2022认证，证书编号KCISM202607012）。如数据保护法律赋予客户强制性审计权且上述报告不能满足该权利，OpenWhispr允许客户或独立审计员在任何12个月内进行不超过一次审计，并在发生经确认、影响客户个人数据的泄露后额外进行审计；每次均须至少提前30日通知，并受保密义务和OpenWhispr安全政策约束。客户自行承担审计费用。对于受限传输，本条不影响SCC项下的审计权。"},"customerObligations":{"title":"11. 客户义务和保证","body":"客户保证，其指示及向OpenWhispr提供个人数据符合数据保护法律，包括具有合法依据、向数据主体提供必要通知，并且在最终用户可能口述或处理特殊类别个人数据（例如健康信息）时，确保具备适当法律依据并签订任何必要协议（HIPAA见第14条）。"},"transfers":{"title":"12. 国际传输","intro":"本条适用于从EEA、英国或瑞士向这些地区以外国家传输个人数据，包括任何受限传输。OpenWhispr及其次级处理者主要在美国处理个人数据（各次级处理者地点见附件3）。每项传输应采用以下第一个适用机制：","mechanisms":["充分性。 如目的地享有充分性决定，传输以该决定为依据。","Standard Contractual Clauses。 对于来自EEA的任何受限传输，双方签订SCC模块二（数据控制者 → 数据处理者），客户为数据输出方，OpenWhispr为数据输入方；该SCC通过引用纳入本DPA，并按以下方式填写：不采用第7条（加入）；第9条选项2（一般授权，提前30日通知）；不采用第11条可选救济机构；第13条/附件I.C的主管监管机构：Irish Data Protection Commission；第17条准据法：爱尔兰法律；第18条法院：爱尔兰法院；本DPA附件1–3构成SCC附件I–III。向次级处理者的后续传输适用模块二SCC第8.8及9(b)条（传递相同义务）。","英国和瑞士传输。 对于英国个人数据，双方签订并以附件信息填写UK Addendum to the EU SCCs；对于瑞士个人数据，SCC按FADP修订后适用。"],"futureFramework":"如OpenWhispr将来依据EU–US Data Privacy Framework完成自我认证，传输还可依赖该认证。","supplementaryMeasures":"如传输影响评估确定需要补充措施，双方将予以实施，OpenWhispr将支持客户的评估。OpenWhispr同意SCC规定的义务，包括在无法遵守时通知客户。"},"governmentRequests":{"title":"13. 政府访问请求","body":"如OpenWhispr收到公共机关就客户个人数据提出的具有法律约束力的请求，将在法律允许范围内：及时通知客户；审查请求合法性并在有合理依据时提出质疑；仅披露法律要求的最少数据。OpenWhispr保存此类请求记录供客户审查。"},"hipaa":{"title":"14. HIPAA","body":"本DPA并非Business Associate Agreement。如客户属于HIPAA规定的Covered Entity或Business Associate，除非双方另行签订Business Associate Agreement，否则最终用户不得通过云服务处理受保护的健康信息（protected health information，PHI）。OpenWhispr可应发送至support@openwhispr.com的请求，向Business及以上套餐的客户提供BAA。签订后，该BAA在冲突范围内取代本DPA并规管PHI。"},"usStatePrivacy":{"title":"15. 美国州隐私法","body":"在CCPA/CPRA或其他美国州隐私法适用的范围内，OpenWhispr作为\"service provider\"/\"processor\"行事：不出售或共享个人数据；除提供服务或法律允许外，不保留、使用或披露个人数据；并确认其理解且将遵守这些限制。"},"liability":{"title":"16. 责任和期限","body":"数据保护事项的责任按协议规定并受其中任何上限约束。只要OpenWhispr仍代表客户处理个人数据，本DPA即持续有效。"},"signatures":{"title":"签署（可选——供需要已签署文本的客户使用）","nameRole":"姓名和职务","signature":"签名","date":"日期","customer":"代表客户（数据控制者）","processor":"代表Gizmo Labs Inc.（数据处理者）","executive":"Gabriel Stein, CEO"},"annex1":{"title":"附件1 — 处理详情","parties":"A. 双方（SCC附件I.A）。 数据输出方：客户（数据控制者）——名称、地址和联系人以协议或订单为准。数据输入方：Gizmo Labs Inc.（数据处理者），221 W 9th Street, Wilmington, Delaware 19801, USA，联系人：support@openwhispr.com；活动：提供下述服务。","descriptionTitle":"B. 处理说明（SCC附件I.B）。","field":"字段","detail":"详情","rows":[{"field":"事项","detail":"提供OpenWhispr语音听写和AI笔记服务（桌面应用程序及API）"},{"field":"期限","detail":"协议期限，加上第4(e)条规定的终止后删除/返还期"},{"field":"性质和目的","detail":"实时语音转文字；AI辅助文本生成、格式化和推理；笔记存储、搜索和共享；账户管理；账单；交易邮件"},{"field":"个人数据类别","detail":"账户和身份数据（姓名、电子邮件、认证标识符）；语音音频（在云路径中实时处理，OpenWhispr不保存）；转录文本和笔记（可能含最终用户口述的任何个人数据）；AI功能对话历史；使用和设备元数据；账单数据（由支付次级处理者处理）"},{"field":"特殊类别数据","detail":"可能偶然包含：口述内容是用户主导的自由文本，可能含健康或其他特殊类别数据。保障措施：传输中及静态加密、不保存音频、仅本地及BYOK模式、PHI须有BAA（第14条）"},{"field":"数据主体类别","detail":"客户的最终用户（雇员、承包商）；口述内容或笔记中提及的人员"},{"field":"频率","detail":"在服务期限内持续进行"}]},"annex2":{"title":"附件2 — 技术和组织措施","intro":"OpenWhispr安全计划概要（当前详情：OpenWhispr信任中心 — <trustLink>trust.openwhispr.com</trustLink>，Security Controls部分）：","bullets":["个人数据在传输中（TLS 1.2+）及静态时加密。","云端语音音频实时处理，转录后不由OpenWhispr保存（转录/推理次级处理者依其协议临时处理音频）；仅本地及bring-your-own-key模式完全绕过OpenWhispr云。","按最小权限原则进行访问控制，生产系统启用MFA及日志记录/监控；每月审查会话和登录监控。","漏洞和补丁管理；安全开发实践。","具备韧性和备份并记录恢复目标（主要数据存储RTO ≤ 4h / RPO ≤ 1h），每年测试应急计划。","人员保密义务和安全培训。","次级处理者治理：合同义务传递（第5条）、公开名单，与主要内容存储次级处理者（数据库；云端转录/LLM）签订BAA；其余HIPAA路径次级处理者的覆盖正在推进。","独立审查：SOC 2 Type 2报告已就2026年5月1日至2026年7月31日期间出具，涵盖安全性、保密性与可用性，可依NDA提供。合规证明：GDPR和HIPAA（2026，通过OpenWhispr合规平台）。信息安全管理体系已通过ISO/IEC 27001:2022认证（证书编号KCISM202607012，由KVQA Certification Services颁发，经加拿大标准委员会认可；初次注册2026年7月27日，有效期至2027年7月27日）。"]},"annex3":{"title":"附件3 — 次级处理者","intro":"截至2026-07-17获授权的次级处理者（实时名单：OpenWhispr信任中心，<trustLink>trust.openwhispr.com</trustLink>，Subprocessors部分）。法人实体名称及处理地点已于2026-07-17根据各供应商条款/DPA核实。\"US (primary)\"指供应商条款将美国列为主要处理地点，同时允许其自身次级处理者在其他国家处理。\"第9(b)条义务传递\"指在受限传输适用时，依据模块二SCC第8.8及9(b)条对次级处理者施加相同义务的合同。","headers":{"subprocessor":"次级处理者","service":"服务","dataCategories":"数据类别","country":"国家","safeguard":"后续传输保障"},"rows":[{"subprocessor":"Neon, LLC (an affiliate of Databricks, Inc.)","service":"主要数据库和认证存储","dataCategories":"账户数据、笔记、转录文本、对话历史","country":"US（AWS US区域，项目创建时固定）","safeguard":"第9(b)条义务传递"},{"subprocessor":"Vercel Inc.","service":"应用托管和交付","dataCategories":"通过服务传输的所有数据","country":"US（主要）","safeguard":"第9(b)条义务传递"},{"subprocessor":"OpenAI OpCo, LLC","service":"云端语音转文字、语言模型、嵌入（已退出训练）","dataCategories":"语音音频（临时）、提示词、文本","country":"US（主要）","safeguard":"第9(b)条义务传递"},{"subprocessor":"Parasail, Inc.","service":"云端AI推理（默认推理；零数据保留推理）","dataCategories":"提示词、文本（传输中）","country":"US（主要）","safeguard":"第9(b)条义务传递"},{"subprocessor":"Groq, Inc.","service":"云端AI推理和转录故障转移","dataCategories":"语音音频（临时）、提示词、文本（传输中）","country":"US（主要）","safeguard":"第9(b)条义务传递"},{"subprocessor":"Embedchain, Inc. d/b/a Mem0","service":"代理功能的AI记忆","dataCategories":"与用户ID关联的对话内容","country":"US","safeguard":"第9(b)条义务传递"},{"subprocessor":"Exa Labs, Inc.","service":"代理功能的网页搜索","dataCategories":"搜索查询文本","country":"US (AWS)","safeguard":"第9(b)条义务传递"},{"subprocessor":"Stripe, LLC","service":"支付处理","dataCategories":"账单和支付数据","country":"US（主要）","safeguard":"第9(b)条义务传递"},{"subprocessor":"Plus Five Five, Inc. (d/b/a Resend)","service":"交易邮件","dataCategories":"电子邮件地址、消息内容（包括共享笔记标题）","country":"US（AWS us-east-1）","safeguard":"第9(b)条义务传递"},{"subprocessor":"Slack Technologies, LLC (Salesforce)","service":"内部运营和支持通信","dataCategories":"支持相关数据（政策：不含PHI）","country":"US（默认数据驻留地）","safeguard":"第9(b)条义务传递"}],"footnote":"本附件列出在产品服务中处理客户个人数据的次级处理者。Google（OAuth登录及可选Calendar集成）不是次级处理者：登录由Google作为独立数据控制者提供，日历数据仅缓存在用户设备上。OpenWhispr更广泛的供应商透明度名单（包括RB2B、lemlist和GitHub等营销网站及运营供应商；对于这些供应商，Gizmo Labs作为数据控制者，或其不接收客户个人数据）发布于OpenWhispr信任中心。"}},"blogBestNoteTakers":{"metadata":{"title":"2026 年最佳 AI 会议笔记工具（免费且注重隐私）| OpenWhispr","description":"对比 2026 年最佳 AI 会议笔记工具：Fathom、Otter、Granola、Fireflies，以及免费、私密、开源并在本地运行的 OpenWhispr。","keywords":["最佳 ai 会议笔记工具","ai 会议笔记工具推荐","最佳免费 ai 会议笔记","开源 ai 会议笔记工具","注重隐私的 ai 会议笔记","ai 会议笔记工具对比","私密 ai 会议笔记工具","最佳 ai 会议记录"]},"breadcrumb":{"home":"首页","blog":"博客","article":"最佳 AI 会议笔记工具"},"schema":{"title":"2026 年最佳 AI 会议笔记工具（含免费和开源选择）","description":"对比 2026 年最佳 AI 会议笔记工具：Fathom、Otter、Granola、Fireflies，以及免费、私密、开源并在本地运行的 OpenWhispr。"},"faqs":[{"question":"哪款 AI 会议笔记工具最好？","answer":"没有一款工具适合所有人，关键取决于你的需求。想以最少设置自动生成会议摘要，Fathom 和 Otter 都很出色。需要 CRM 同步和对话分析的销售团队可以选择 Fireflies。如果最看重隐私，希望音频留在自己的设备上，OpenWhispr 更为突出：它在本地转写、代码开源，而且不会向通话中派出机器人。"},{"question":"哪款免费 AI 会议笔记工具最好？","answer":"Fathom 为个人用户提供了云端产品中相当慷慨的免费方案。如果还希望免费工具能保护数据隐私，OpenWhispr 免费且开源，并提供不限量的本地转写：没有每月上限、无需银行卡，也不会上传任何内容。"},{"question":"有开源的 AI 会议笔记工具吗？","answer":"有。OpenWhispr 采用 MIT 许可证，完全开源。语音转文字通过 OpenAI Whisper 和 NVIDIA Parakeet 在你的设备上运行，因此你可以准确审查音频的处理方式。Meetily 也是一款值得了解的开源、本地优先产品。"},{"question":"哪款 AI 会议笔记工具最能保护隐私？","answer":"最私密的方案是在设备上转写，而且绝不上传音频。OpenWhispr 默认就是如此：没有机器人加入通话，录音留在你的电脑上。因此，它很适合人力资源、法律事务或涉及客户数据等敏感会议。相比之下，大多数云端笔记工具会把会议上传并存储在自己的服务器上。"},{"question":"AI 会议笔记工具必须派机器人加入会议吗？","answer":"大多数需要。Otter、Fireflies 和 Fathom 会派一个机器人加入，并显示在参会者列表中。也有少数例外：Granola 无需机器人即可采集电脑音频；OpenWhispr 也是如此，并且整个转写过程都留在设备上。"},{"question":"哪款 AI 会议笔记工具可以离线使用？","answer":"Otter、Fireflies 和 Fathom 等云端工具需要联网，因为转写在它们的服务器上完成。OpenWhispr 可以完全离线运行：本地模型无需网络即可在你的电脑上转写，适合乘飞机、进入安全设施，或任何不希望音频离开设备的场景。"}],"sections":[{"id":"how-to-choose","title":"如何选择"},{"id":"comparison","title":"快速对比"},{"id":"picks","title":"推荐工具"},{"id":"faq","title":"常见问题"}],"hero":{"title":"2026 年最佳 AI 会议笔记工具（含免费和开源选择）","subtitle":"自动生成云端会议摘要，还是在设备上保留私密笔记？这份公平、坦诚的对比会说明各款主流 AI 会议笔记工具真正擅长什么，又有哪些不足。","date":"2026 年 7 月 6 日","tag":"对比"},"opening":{"lead":"没有唯一一款最好的 AI 会议笔记工具；正确选择取决于你更重视自动化、价格还是隐私。","body":"想全自动获得云端摘要，Fathom 和 Otter 很出色；销售团队可以考虑 Fireflies；希望音频留在自己设备上，则可选择 OpenWhispr。下面会坦诚分析每款工具，包括其他产品表现更好的地方。","updated":"最后更新：2026 年 7 月 6 日。OpenWhispr 也在这份榜单中；我们尽力如实说明其他工具胜出的方面。"},"howToChoose":{"title":"如何选择 AI 会议笔记工具","intro":"几乎所有 AI 会议笔记工具都会录制通话、转写内容并生成摘要。真正的差异可以归结为三个问题：","criteria":[{"lead":"音频会被发送到哪里？","body":"云端工具会把会议上传并存储在自己的服务器上，本地工具则把数据留在设备内。对于人力资源、法律、医疗或客户对话，这往往是决定性因素。"},{"lead":"机器人会加入通话吗？","body":"大多数工具会在参会者列表中加入一个可见的机器人。少数工具无需机器人即可采集音频，干扰更小，也不会引发大家追问陌生参会者是谁的尴尬。"},{"lead":"免费方案到底包含什么？","body":"“免费”可能只是每月几场会议，也可能真正不限量。决定使用前，请先查看上限。"}]},"comparison":{"title":"快速对比","headers":{"tool":"工具","bestFor":"最适合","runs":"运行位置","bot":"机器人？","freeTier":"免费方案"}},"picks":{"title":"逐一了解最佳 AI 会议笔记工具","keepInMind":"需要注意：","openWhisprLink":"了解 OpenWhispr","compareLink":"OpenWhispr 与 {name} 对比","tools":[{"name":"OpenWhispr","bestFor":"隐私与开源","runs":"本地","bot":"无机器人","freeTier":"本地不限量，免费","blurb":"先说明：这是我们的产品，所以下面只讲真实情况。OpenWhispr 使用 OpenAI Whisper 和 NVIDIA Parakeet 在你的设备上转写，采用开源 MIT 许可证，并且无需向通话中派机器人即可采集会议音频。它也能作为全系统听写工具，让你在任何应用中用语音输入。","pros":["音频绝不离开设备；不会上传任何内容","在 macOS、Windows 和 Linux 上开源且可审查","本地转写不限量，永久免费"],"caveat":"它不会派机器人自动加入日程中的通话，你需要自己开始录音。要获得专用云端笔记工具那样精致的摘要，也需要多做一些设置。","href":"/ai-meeting-note-taker"},{"name":"Fathom","bestFor":"最佳免费云端工具","runs":"云端","bot":"机器人","freeTier":"慷慨的免费方案","blurb":"Fathom 是获取自动会议回顾最省事的方式之一，在云端笔记工具中也拥有相当慷慨的免费方案。它的机器人会加入并录制通话，几乎无需设置就能生成清晰的摘要、重点和待办事项。","pros":["面向个人的优秀免费方案","轻松生成摘要和重点","良好支持 Zoom、Teams 和 Meet"],"caveat":"它在云端运行：机器人会加入通话，会议也会上传至 Fathom 的服务器。","href":"/compare/fathom"},{"name":"Otter.ai","bestFor":"实时协作","runs":"云端","bot":"机器人","freeTier":"300 分钟/月（以往）","blurb":"Otter 是成熟的会议转写平台，具备强大的实时转写、共享和集成功能。它适合希望共同保存、搜索通话记录并协作标记重点的团队。","pros":["实时转写","团队共享和工作区功能","广泛的集成"],"caveat":"它通过机器人在云端运行，免费方案有上限，重度用户很快就会用完。","href":"/compare/otter"},{"name":"Fireflies.ai","bestFor":"销售与 CRM","runs":"云端","bot":"机器人","freeTier":"有上限的免费方案","blurb":"Fireflies 面向营收团队：机器人加入并转写通话，再提供对话分析，并与 Salesforce 和 HubSpot 深度同步。如果会议笔记需要进入销售管道，它是很有力的选择。","pros":["对话智能与分析","深度 CRM 集成","可搜索的团队转写资料库"],"caveat":"它通过机器人在云端运行；最实用的功能位于按席位付费的方案中。","href":"/compare/fireflies"},{"name":"Granola","bestFor":"无机器人的云端记事本","runs":"云端","bot":"无机器人","freeTier":"免费方案","blurb":"Granola 的做法比大多数产品更自然：无需在参会者列表中加入机器人，就能采集电脑音频，并配有简洁的记事本和日历同步。它很受希望获得整洁 AI 会议笔记的 Mac 用户欢迎。","pros":["没有机器人加入通话","简洁、快速的笔记体验","日历同步"],"caveat":"处理仍在云端进行，因此即使看不到机器人，会议音频也会离开你的设备。","href":"/compare/granola"},{"name":"Notta","bestFor":"多语言会议","runs":"云端","bot":"机器人","freeTier":"有上限的免费方案","blurb":"Notta 是云端转写和会议工具，支持的语言尤其广泛，还能上传文件和录音。会议涉及多种语言时，它是不错的选择。","pros":["广泛的语言支持","转写实时通话或上传的文件","导出简单"],"caveat":"它在云端运行；更长的录音和更高的准确率需要付费方案。","href":null},{"name":"tl;dv","bestFor":"录制与片段分享","runs":"云端","bot":"机器人","freeTier":"免费录制方案","blurb":"tl;dv 专注于录制和分享会议片段，包括带时间戳的短片、重点和交接内容，并提供实用的免费方案。需要向团队分享会议片段时很方便。","pros":["剪辑并分享关键时刻","免费录制方案","支持 Zoom、Meet 和 Teams"],"caveat":"它通过机器人在云端运行；录制内容保存在 tl;dv 的服务器上。","href":null}]},"faq":{"title":"常见问题"},"cta":{"title":"想让 AI 会议笔记永不离开设备？","body":"OpenWhispr 是免费、开源的 AI 会议笔记工具，直接在你的电脑上录音和转写：无需机器人，也不上传云端。macOS、Windows 和 Linux 均可免费使用。","github":"在 GitHub 上查看"}},"blogZoomRecording":{"metadata":{"title":"如何录制 Zoom 会议（无论是否为主持人）| OpenWhispr","description":"了解主持人或参会者如何在本地、云端或无机器人的情况下录制 Zoom 会议，并在自己的设备上获得私密转写。","keywords":["如何录制 zoom 会议","录制 zoom 会议","参会者如何录制 zoom","未经许可录制 zoom","录制带声音的 zoom","zoom 录制文件保存位置","本地录制 zoom","获取 zoom 会议转写","zoom 会议转写","录制 zoom 是否合法","mac 录制 zoom","windows 录制 zoom"]},"breadcrumb":{"home":"首页","blog":"博客","article":"如何录制 Zoom 会议"},"schema":{"title":"如何录制 Zoom 会议（无论是否为主持人）","description":"了解主持人或参会者如何在本地、云端或无机器人的情况下录制 Zoom 会议，并在自己的设备上获得私密转写。"},"faqs":[{"question":"不是主持人也能录制 Zoom 会议吗？","answer":"可以。主持人可在通话中授予权限（参会者 → 悬停你的姓名 → 更多 → 允许录制），从而启用 Zoom 内置录制。否则，可用电脑上的独立工具录制屏幕和系统音频，无需 Zoom 权限即可捕捉看到和听到的全部内容。无论哪种方式，都要先告知其他参会者。"},{"question":"Zoom 录制文件保存在哪里？","answer":"本地录制默认保存在电脑“文档”下的 Zoom 文件夹中，包含 MP4 视频和独立 M4A 音频。云录制（仅付费计划）保存在 Zoom 服务器，提供分享链接，部分计划还提供自动转写。重视隐私时应选择本地录制。"},{"question":"录制 Zoom 会议合法吗？","answer":"取决于参会者所在地。有些地区只需一人同意，有些要求所有人同意；欧盟 GDPR 也将录下的声音视为个人数据。这不是法律建议。安全做法是提前说明并取得明确同意。"},{"question":"如何录制带声音的 Zoom 会议？","answer":"Zoom 内置功能会自动录制视频和会议音频。使用独立录屏工具时，请启用系统音频，而不只是麦克风，否则视频会没有声音。OpenWhispr 等本地工具可直接捕捉完整的系统音频。"},{"question":"能在主持人不知情时录制吗？","answer":"Zoom 内置录制会向所有人显示“正在录制”。外部工具不会显示，但秘密录制可能违法，也会破坏信任。请始终告知大家；录制应当形成准确、共享的记录，而不是暗中留存。"},{"question":"如何获得 Zoom 会议转写？","answer":"付费计划可自动转写云录制。本地录制可在任何计划下自行转写。OpenWhispr 在设备上生成支持搜索的准确转写，支持 100+ 种语言，并可用 AI 汇总笔记和行动项。"},{"question":"可以在手机上录制 Zoom 吗？","answer":"付费计划的主持人可在 Zoom 移动应用中录制到云端。参会者可使用 iPhone 控制中心或 Android 的屏幕录制，并开启麦克风或内部音频。之后可转写该文件。"},{"question":"怎样最私密地录制 Zoom？","answer":"选择本地录制，让文件不离开设备，并避免把会议上传给第三方的云端笔记机器人。OpenWhispr 在电脑上完成系统音频录制和转写：没有机器人入会，也不上传数据，适合人力、法律和客户数据等敏感会议。"}],"sections":[{"id":"host","title":"主持人录制"},{"id":"participant","title":"不是主持人时"},{"id":"local","title":"本地录制，无机器人"},{"id":"mobile","title":"手机录制"},{"id":"transcript","title":"获取转写"},{"id":"legal","title":"同意与合法性"},{"id":"storage","title":"保存位置"},{"id":"troubleshoot","title":"故障排除"},{"id":"faq","title":"常见问题"}],"hero":{"title":"如何录制 Zoom 会议（无论是否为主持人）","subtitle":"以主持人或参会者身份，在云端或自己的设备上录制带声音的 Zoom 通话，再将其变成可搜索的转写，并了解同意规则。","date":"2026年7月6日","tag":"指南"},"opening":{"lead":"如果你是主持人或已获授权，请点击工具栏中的“录制”来录制 Zoom 会议。","body":"没有权限的参会者可请主持人启用录制，或用电脑上的独立工具录制屏幕和系统音频；后者无需 Zoom 权限，可捕捉完整通话。无论采用何种方式，都要先告知其他人。","scope":"本文涵盖主持人的本地与云录制、参会者录制、无机器人设备端录制、手机录制、转写、同意规则，以及录制按钮变灰时的处理方法。","updated":"最后更新：2026年7月6日。"},"host":{"title":"主持人录制（Zoom 内置功能）","intro":"由你发起会议时，只需点击工具栏中的“录制”。付费计划会要求选择两个位置之一：","options":[{"title":"在此计算机上录制","body":"会议结束时在电脑上保存 MP4 视频和 M4A 音频。所有计划均可使用，包括免费计划。"},{"title":"录制到云端","body":"上传到 Zoom 服务器并提供分享链接；部分计划还有自动转写。仅限付费计划。"}],"outroPrefix":"开始后，Zoom 会向所有人显示“正在录制”。会议结束后，本地文件会转换并默认放入“文档”中的","outroSuffix":"文件夹。"},"participant":{"title":"不是主持人时","intro":"参会者必须得到主持人允许才能使用 Zoom 的录制功能。可选两种方式：","options":[{"title":"请求权限。","body":"主持人打开“参会者”，悬停你的姓名，点击“更多”并选择“允许录制”，录制按钮便会出现。"},{"title":"录制屏幕和系统音频。","body":"独立工具可在没有主持人权限时捕捉屏幕和扬声器声音，是主持人无法或不愿启用录制时的可靠方案。"}],"warningPrefix":"普通录屏常常只录到你的麦克风，漏掉其他人的声音。请启用","systemAudio":"系统音频","warningSuffix":"，或使用专门捕捉会议声音的工具。"},"local":{"title":"本地录制，不用机器人和云端","cloudPrefix":"Otter、Fireflies、Fathom 等云端笔记服务会让","bot":"机器人","cloudSuffix":"加入通话并把音频传到服务器。即使没有机器人，有些服务仍会把会议发到设备外处理，不适合敏感会议。","approachPrefix":"采用相反方式，直接在电脑上捕捉系统音频：","noBot":"没有机器人加入通话","approachSuffix":"，音频也不离开设备。它可作为参会者在任意 Zoom 计划及 macOS、Windows、Linux 上使用。你仍需告知录制，但没有上传或第三方访问。","bullets":["录制整场通话，而不只是麦克风。","参会列表无机器人，也不上传云端。","在设备上以 100+ 种语言转写和总结。"],"github":"在 GitHub 上查看"},"mobile":{"title":"在手机上录制 Zoom 会议","paragraph1":"付费计划主持人可在 Zoom 应用中点“更多”，再选“录制到云端”。参会者可用 iPhone 或 Android 自带录屏，并开启麦克风或内部音频，以免只得到无声视频。","paragraph2":"手机录制就是视频文件，会议结束后可像其他录制一样转写。"},"transcript":{"title":"把录制变成转写和笔记","paragraph1":"只有可阅读、可搜索时，录制才真正有用。付费计划可自动转写云录制；本地录制或免费计划需自行转写保存的文件。","paragraph2Prefix":"将 MP4 或 M4A 导入","paragraph2Suffix":"，即可在设备上得到带时间戳的准确转写，并让 AI 总结决定、行动项和后续事项。由于完全在本地运行，机密录制也不会上传。"},"legal":{"title":"同意与合法性","paragraph1":"是否合法取决于参会者所在地。有些地方一人同意即可，有些要求所有人同意；欧盟 GDPR 将录下的声音视为个人数据。这不是法律建议。请在开始前说明并获得明确同意。","paragraph2":"Zoom 内置录制会自动在屏幕上通知。外部工具不会，因此告知责任在你。秘密录制存在法律和信任风险；目的应是创建所有人都能依赖的共享记录。"},"storage":{"title":"录制文件最终保存在哪里","localLabel":"本地录制","localBody":"以 MP4 和 M4A 留在电脑上，默认位于“文档”的 Zoom 文件夹。除非你分享，否则不会离开设备。","cloudLabel":"云录制","cloudBody":"保存在 Zoom 服务器上，受计划容量和组织保留策略限制。","botLabel":"机器人笔记服务","botBody":"会将整场会议上传给第三方，部分服务商可能用其训练模型。人力、法律、医疗或客户数据应通过本地录制与转写保持控制。"},"troubleshoot":{"title":"故障排除：录制按钮缺失或变灰","items":[{"title":"你不是主持人且没有权限。","body":"请主持人允许，或使用本地屏幕与音频录制。"},{"title":"账户层面禁用了录制。","body":"受管账户可能需要管理员在 Zoom 网页设置中启用。"},{"title":"没有云录制。","body":"云录制需要付费计划；免费计划仅支持本地录制。"},{"title":"录制没有声音。","body":"录屏只捕捉了麦克风。请开启系统音频或使用专用工具。"}]},"faq":{"title":"常见问题"},"closing":{"title":"私密地录制并转写会议","body":"OpenWhispr 在设备上把 Zoom 音频变成笔记，无机器人、无上传。macOS、Windows 和 Linux 免费使用。","github":"在 GitHub 上查看"}},"blogStreamingStt":{"metadata":{"title":"CPU 流式语音转文字：本地实时运行 | OpenWhispr","description":"OpenWhispr 如何在 CPU 上运行 NVIDIA Nemotron 流式 ASR：通过本地 WebSocket 进行缓存感知流式处理，实时返回部分结果，无需 GPU 和云端。","keywords":["流式语音转文字","本地流式asr","nemotron asr","缓存感知流式asr","sherpa onnx","本地实时转录","实时听写预览","流式asr electron","端侧语音识别"]},"breadcrumb":{"home":"首页","blog":"博客","article":"在 CPU 上运行流式语音转文字"},"schema":{"title":"在 CPU 上运行流式语音转文字：将 NVIDIA Nemotron 集成到桌面应用","description":"OpenWhispr 如何在 CPU 上运行 NVIDIA Nemotron 流式 ASR：通过本地 WebSocket 进行缓存感知流式处理，实时返回部分结果，无需 GPU 和云端。"},"faqs":[{"question":"什么是流式语音转文字？","answer":"流式（online）语音转文字会在说话尚未结束时处理音频，并在固定延迟范围内输出部分结果，通常为80毫秒到约1秒。Whisper 等批处理（offline）模型则要等录音结束后一次性处理全部内容。实时字幕、语音智能体和边说边显示的听写预览都依赖流式处理。"},{"question":"没有 GPU，能否在本地运行流式语音转文字？","answer":"可以。NVIDIA Nemotron 流式模型有600M参数，以约650MB的INT8量化ONNX文件提供，通过无需Python或PyTorch的原生二进制运行时sherpa-onnx，可在现代CPU上实时运行。OpenWhispr为macOS、Windows和Linux提供了这套配置。"},{"question":"什么是缓存感知流式 ASR？","answer":"这是 NVIDIA cache-aware FastConformer 架构：模型会在音频分块之间保留编码器内部状态。每个新分块只处理一次，并使用此前语音的缓存上下文，而不是重新编码相互重叠的音频窗口。因此，低延迟流式识别可以轻量到在笔记本CPU上运行。"},{"question":"实时预览文字为何显示后还会变化？","answer":"流式模型会随着更多音频到达而修正假设。单词边界或同音词有时只能靠后文确定，因此部分结果会原地替换。这是有意设计。停止听写时，OpenWhispr会刷新模型尾部，并将稳定后的流式文字写入转录；若刷新中断或结果被截断，应用会自动重新解码完整录音，绝不会把不稳定的流式结果粘贴出去。"},{"question":"流式转录是否不如批处理准确？","answer":"略低，因为流式模型无法看到未来音频。Nemotron 英语流式模型在 Open ASR Leaderboard 数据集上使用1.12秒分块时，平均词错误率为6.93%。这已非常接近批处理模型，但仍落后于 NVIDIA 最佳英语批处理模型的5.91%。选择流式模型，就是以这点差距换取实时文字和停止即得的转录；若最高准确率更重要，可选择offline Parakeet模型。"},{"question":"OpenWhispr 的流式模式支持哪些语言？","answer":"目前有两个 Nemotron 模型：仅英语模型，以及支持自动语言检测的 Nemotron 3.5。后者有15种可直接转录的语言（英语、西班牙语、法语、意大利语、葡萄牙语、荷兰语、德语、土耳其语、俄语、阿拉伯语、印地语、日语、韩语、越南语、乌克兰语）。Parakeet 和 Whisper 等 offline 模型覆盖更多语言，但预览使用缓冲分块而不是实时流。"}],"sections":[{"id":"the-problem","title":"问题：分块式预览"},{"id":"cache-aware","title":"缓存感知流式处理是什么"},{"id":"architecture","title":"系统架构"},{"id":"gotchas","title":"实现过程中踩过的坑"},{"id":"single-pass","title":"直接提交流式结果"},{"id":"footprint","title":"性能与资源占用"},{"id":"limitations","title":"坦诚的局限"},{"id":"model-choice","title":"如何选择模型"},{"id":"faq","title":"常见问题"}],"hero":{"title":"在 CPU 上运行流式语音转文字：将 NVIDIA Nemotron 集成到桌面应用","subtitle":"OpenWhispr 如何通过本地 WebSocket 上的缓存感知流式模型，在不用 GPU 和云端的情况下边说边显示文字，以及让它可靠运行所需的工程工作。","date":"2026年7月18日","tag":"技术"},"opening":{"lead":"流式语音转文字不必等待录音结束，而是在你说话时同步转录。从版本1.7.6开始，OpenWhispr完全在设备端完成这项工作：NVIDIA Nemotron 流式模型（600M参数、INT8量化）通过本地sherpa-onnx WebSocket服务器在CPU上运行；部分结果到达时听写预览会实时更新，停止时则立即将流式文字提交为正式转录。","body":"无需 GPU，无需 Python 环境，音频也不会离开设备。","guidePrefix":"本文记录具体工程实现：旧版实时预览为何迟缓，缓存感知流式处理究竟是什么，整个管线如何装进Electron应用，版本固定错误为何导致全面误解码，以及我们如何从每次重新解码整段录音改为直接提交流。模型之间的比较请参阅","comparisonLink":"Parakeet vs Whisper vs Nemotron","guideSuffix":"。","updated":"最后更新于2026年7月18日。实现细节对应 OpenWhispr 1.7.6 发布的流式转录功能；文中内容均可在开源代码库中核查。"},"factCheck":{"title":"事实速查（第一手来源）","items":[{"lead":"Nemotron 的 cache-aware FastConformer 只处理新增音频，并复用缓存的编码器上下文，","body":"运行时可将分块大小配置为80ms到1.12s。","sources":[{"href":"https://huggingface.co/nvidia/nemotron-speech-streaming-en-0.6b","label":"Nemotron Streaming EN 模型卡"},{"href":"https://huggingface.co/blog/nvidia/nemotron-speech-asr-scaling-voice-agents","label":"NVIDIA 对缓存感知流式处理的说明"}]},{"lead":"英语模型在1.12s分块下流式运行，平均WER为6.93%；","body":"多语言3.5模型支持40种语言与地区组合，并能自动检测。","sources":[{"href":"https://huggingface.co/nvidia/nemotron-3.5-asr-streaming-0.6b","label":"Nemotron 3.5 ASR 模型卡"}]},{"lead":"全部推理通过 sherpa-onnx 完成；它是无需 Python 的原生 ONNX 运行时，","body":"我们使用其online（流式）WebSocket服务器，并固定和打包版本1.13.4。","sources":[{"href":"https://github.com/k2-fsa/sherpa-onnx","label":"GitHub 上的 sherpa-onnx"}]},{"lead":"INT8 ONNX模型在磁盘上分别占632MB（英语）和650MB（多语言），","body":"只需下载一次，之后保存在本地缓存中。","sources":[{"href":"https://github.com/OpenWhispr/openwhispr","label":"OpenWhispr 模型注册表（开源）"},{"href":"https://openwhispr.com/models","label":"OpenWhispr 模型页面"}]}]},"problem":{"title":"问题：在批处理模型上搭建实时预览","paragraphs":[{"body":"OpenWhispr一直有转录预览：听写时用一个小窗口显示模型听到的内容。1.7.6之前，我们采用了让批处理模型看起来实时的唯一办法：缓存约1.5秒麦克风音频，完整转录这个分块，追加文字，然后重复。"},{"body":"这种方式有三个结构性问题。第一是延迟：缓冲区填满前文字无法出现，因此预览总会落后1.5秒加推理时间。第二是边界：跨越两个分块的单词会被切断，而分块独立转录，双方都无法修正。第三是上下文：第五块不知道第四块说了什么，模型每1.5秒都要重新判断同一类歧义。"},{"body":"缩短分块会显得更快，却因上下文减少、边界增多而大幅降低准确率。重叠分块能修复边界，却要重复转录同一音频。真正需要的是为转录音频流而训练的模型，这就是Nemotron。"}]},"cacheAware":{"title":"缓存感知流式处理究竟是什么","paragraphs":[{"lead":"缓存感知流式模型会在分块之间保留内部状态。","body":"下一段音频到达时，编码器不会重读此前全部语音，只处理新帧，并查询已计算上下文的缓存。它和让聊天机器人更快的KV缓存原理相同，只是用在语音编码器上。"},{"body":"这种设计带来两个结果。计算量不再随录音长度增长：无论录到10秒还是10分钟，每块成本都一样，因此纯CPU流式处理成为可能。上下文也会延续：模型利用此前全部语音解决歧义，后续音频给出新证据时会修正较早的词。文字在你说完后稍微“稳定”一下，不是故障，而是模型用更充分的依据改正判断。"},{"body":"延迟预算可以调节。Nemotron提供80ms（最快，每步上下文最少）到1.12s（准确率最高，在leaderboard数据集上平均6.93% WER，与最佳批处理模型相差约一个百分点）的分块。训练采用NVIDIA cache-aware FastConformer和transducer解码器；其"}],"engineeringPostLink":"工程文章","engineeringPostSuffix":"深入解释了相关原理。"},"architecture":{"title":"OpenWhispr 内部架构","paragraphs":[{"body":"管线刻意保持简单。AudioWorklet以16kHz采集麦克风PCM，将帧转换为float32传输格式，再通过WebSocket推送到sherpa-onnx online服务器。OpenWhispr在本地启动这个原生二进制，并通过127.0.0.1通信，每次录音保持一条长连接。服务器以INT8权重运行Nemotron并返回JSON部分结果；预览窗口会原地替换全文，而不是不断追加。"},{"body":"为何采用独立服务器进程，而不在进程内推理？为了隔离。原生推理运行时可能崩溃；ONNX内核一次错误的内存分配不应拖垮整个桌面应用。辅助进程崩溃只需重连，不会丢失听写。它也让原生代码完全离开Electron主进程：应用负责协议，辅助进程负责计算。"},{"body":"同一模式也用于我们的"}],"diarizationLink":"本地说话人分离","diarizationSuffix":"——一套sherpa-onnx工具链，多种任务，全都在设备端完成。"},"gotchas":{"title":"实现过程中踩过的坑","items":[{"label":"运行时版本比预想的更重要。","body":"Nemotron需要sherpa-onnx 1.13.4或更新版本才能准确解码。在为Parakeet打包的旧运行时上，它不会明显报错，而是直接转录错误。修复很普通：升级并重新固定各平台二进制。普遍教训是，对模型运行时而言，“能运行”和“结果正确”是两项测试，只有后者算数。"},{"label":"流会失败，所以旧路径必须保留。","body":"端口可能被占用，模型文件可能缺失，服务器也可能在录音中退出。若长连接无法启动，预览会自动回退到缓冲分块路径——更慢，但不会空白。offline模型（Parakeet、Whisper）本来就保留分块预览；注册表中的每个模型声明运行时为online或offline，应用据此选择。"},{"label":"部分结果需要不同的渲染方式。","body":"分块路径追加文字，流式假设却会自我修正，因此UI在每个部分结果到达时替换整段预览。如果把会变的假设不断追加，就会产生断断续续的重复文字——这是我们很早修复的真实UX问题。"}]},"singlePass":{"title":"直接提交流：从两次解码到一次","paragraphs":[{"body":"第一版只把流式结果当作预览：停止听写后，应用会从头重新解码完整录音，并粘贴那份结果。这样很稳妥，因为最终处理能看到完整上下文；但每次听写都付出两次解码的成本，而且你已经看着文字稳定下来，却仍要等待几乎总是相同的第二意见。"},{"body":"于是我们在同一版本中把流设为事实来源。无论预览窗口是否打开，听写都会走长连接；停止时，应用刷新模型尾部，直接提交流式文字，不再进行第二次解码。结束延迟只剩一次尾部刷新，每次听写的CPU开销约减半。"},{"body":"安全网没有消失，而是成为回退机制。停止刷新会感知截断，只要结果仍在到达便延长期限。若连接中断或结果被截断，应用会丢弃流，改用经典的“先录音、后转录”方式处理完整录音。只有既快又正确时才走快速路径。"}]},"footprint":{"title":"性能与资源占用","items":[{"label":"模型：","body":"Nemotron Speech Streaming EN 0.6B（632MB）和Nemotron 3.5 ASR Streaming 0.6B（650MB、15种可转录语言、自动检测）——INT8 ONNX，只下载一次并在本地缓存。"},{"label":"运行时：","body":"sherpa-onnx 1.13.4，自包含的原生二进制。无需Python、PyTorch或CUDA。"},{"label":"硬件：","body":"现代笔记本CPU可实时处理；Apple Silicon和较新的x86都能轻松跟上说话速度。"},{"label":"体感：","body":"部分文字落后语音远不到1秒；停止时一次尾部刷新即可提交转录，无需再次解码。"}],"privacyPrefix":"这条路径完全不接触互联网。应用通过127.0.0.1与服务器通信（Windows上还用限定范围的防火墙规则阻止网络访问该端口），模型文件位于本地缓存，音频从不离开设备。这与","privacyLink":"我们的本地转录体系","privacySuffix":"其余部分采用相同的隐私标准。"},"limitations":{"title":"坦诚的局限","items":[{"label":"流式处理以少量准确率换取即时性。","body":"在相同基准上，流式约为~6.9%，NVIDIA最佳英语批处理模型为5.91%。现在正式转录来自流，因此你确实接受了这项取舍。若只在意最终结果，offline Parakeet仍是准确率更高的选择。"},{"label":"语言覆盖范围小于Whisper。","body":"多语言模型有15种可直接转录语言，Whisper则有99种。普通话和泰语流式识别目前尚不可用。"},{"label":"部分文字会变化。","body":"假设会随上下文修正。我们认为看着它逐渐稳定是一项优点；若觉得分心，可关闭预览，听写仍和以前完全一样。"},{"label":"磁盘会再占用650MB","body":"如果已经装有offline模型，这就是为不同任务优化第二套引擎的代价。"}]},"modelChoice":{"title":"在 OpenWhispr 中选择模型","wantHeader":"你的需求","pickHeader":"选择","rows":[{"want":"听写时实时预览英语文字","pick":"Nemotron Speech Streaming EN 0.6B"},{"want":"实时预览西班牙语、日语、阿拉伯语、印地语等","pick":"Nemotron 3.5 ASR Streaming 0.6B"},{"want":"英语准确率最高，不需要实时文字","pick":"Parakeet Unified EN 0.6B"},{"want":"使用NVIDIA范围外的语言","pick":"Whisper (turbo 或 large)"}],"catalogPrefix":"这些模型都可在“设置”的下拉菜单中选择；完整的容量和取舍信息见","catalogLink":"模型页面","catalogSuffix":"。OpenWhispr免费、开源，并支持macOS、Windows和Linux。","githubCta":"在 GitHub 上加星"},"faq":{"title":"常见问题"}},"blogParakeetNemotron":{"metadata":{"title":"Parakeet、Whisper 与 Nemotron：2026 年最佳本地 STT | OpenWhispr","description":"客观比较主流本地语音转文字模型：准确率（WER）、速度、语言覆盖和流式识别。数据均来自一手资料。","keywords":["parakeet 与 whisper 对比","nemotron asr","最佳本地语音转文字","whisper 替代方案","nvidia parakeet","本地语音转文字模型","开源语音转文字","2026 最佳语音转文字模型","nemotron 与 whisper 对比","离线语音识别"]},"breadcrumb":{"home":"首页","blog":"博客","article":"Parakeet、Whisper 与 Nemotron 对比"},"schema":{"title":"Parakeet、Whisper 与 Nemotron：2026 年最佳本地语音转文字方案","description":"客观比较主流本地语音转文字模型：准确率（WER）、速度、语言覆盖和流式识别，并引用一手资料中的数据。"},"faqs":[{"question":"2026 年最佳本地语音转文字模型是什么？","answer":"对于英语和主要欧洲语言，NVIDIA Parakeet TDT 0.6B v3 的综合表现最佳：它在 Open ASR Leaderboard 上的平均词错误率为 6.34%，优于 Whisper large-v3 约 7.4% 的成绩，体积仅为后者的四分之一，CPU 吞吐量也高得多。如果需要边说边显示文字，NVIDIA Nemotron 流式模型是最佳选择。如果所需语言不在 NVIDIA 模型的覆盖范围内，请使用 Whisper。"},{"question":"Parakeet 比 Whisper 更准确吗？","answer":"在 Open ASR Leaderboard 的英语基准中，是的。Parakeet TDT 0.6B v3 的平均 WER 为 6.34%，Whisper large-v3 约为 7.4%，仅支持英语的 Parakeet Unified 更达到 5.91%。但需要说明的是，日常听写中约 ~1 个 WER 百分点的差距很难察觉。Parakeet 更明显的优势在于速度、效率，以及不会在静音时生成幻觉文本；后者是 Whisper 的已知问题。"},{"question":"在本地运行这些模型需要 GPU 吗？","answer":"不需要。三者都能在现代 CPU 上运行。Parakeet 和 Nemotron 使用针对 CPU 推理设计的 INT8 量化 ONNX 模型（磁盘占用约 630–680MB）。Whisper 通过 whisper.cpp 在 CPU 上运行，不过其 3GB 大模型能从 GPU 加速中受益。OpenWhispr 支持 Apple Silicon 上的 Metal、NVIDIA 上的 CUDA 以及 AMD/Intel GPU 上的 Vulkan，并可自动回退至 CPU。"},{"question":"Nemotron ASR 支持哪些语言？","answer":"NVIDIA Nemotron 3.5 ASR 在单个 600M 参数模型中覆盖 40 个语言区域，并支持自动语言检测。OpenWhispr 已启用其中 15 种可直接转录的语言：英语、西班牙语、法语、意大利语、葡萄牙语、荷兰语、德语、土耳其语、俄语、阿拉伯语、印地语、日语、韩语、越南语和乌克兰语。另有仅支持英语的 Nemotron 流式模型。"},{"question":"Whisper 已经过时了吗？","answer":"没有。Whisper 仍支持 99 种语言，远多于 Parakeet 的 25 种或 Nemotron 的 40 个语言区域。对于日语、韩语和越南语之外的亚洲语言、低资源语言以及频繁切换语言的音频，它仍是最稳妥的选择。它也是经过最多实际检验、生态最成熟的模型。变化只在于 Whisper 已不再占据英语准确率和速度的首位。"},{"question":"OpenWhispr 支持其中哪些模型？","answer":"全部支持。OpenWhispr 通过 whisper.cpp 提供六种大小的 Whisper，通过 sherpa-onnx 提供两款 Parakeet 和两款 Nemotron 流式模型，并可在设置中切换。所有处理均在设备上完成，音频绝不会离开您的电脑；应用免费且开源。"}],"sections":[{"id":"contenders","title":"三个候选模型"},{"id":"accuracy","title":"准确率"},{"id":"speed","title":"速度与效率"},{"id":"languages","title":"语言覆盖"},{"id":"streaming","title":"流式识别：真正的分水岭"},{"id":"hardware","title":"硬件要求"},{"id":"which-to-choose","title":"应该选哪一个？"},{"id":"try-them","title":"三个都试一遍"},{"id":"faq","title":"常见问题"}],"hero":{"title":"Parakeet、Whisper 与 Nemotron：2026 年最佳本地语音转文字方案","subtitle":"如今，三款开放模型都能完全在您的电脑上将语音转成文字。本文依据一手资料中的可靠数据，比较它们在准确率、速度、语言覆盖和流式识别方面的真实表现。","date":"2026 年 7 月 18 日","tag":"对比"},"opening":{"lead":"对于英语和主要欧洲语言，NVIDIA Parakeet TDT 0.6B v3 是 2026 年最佳本地语音转文字模型：其实测准确率优于 Whisper large-v3，体积仅为后者四分之一，在普通 CPU 上也快得多。如果希望文字在说话时实时出现，NVIDIA 新推出的 Nemotron 流式模型是首批专门为此设计的本地模型。如果您使用的是 NVIDIA 模型尚未覆盖的众多语言之一，那么支持 99 种语言的 OpenAI Whisper 仍是正确选择。","paragraph2":"2026 年最出人意料的变化是，Whisper 失去了保持三年的开放模型准确率桂冠。不过，“哪款模型最好”已经真正变成三方竞争，因为它们各有所长：Parakeet 优化单位功耗下的准确率，Nemotron 优化延迟，Whisper 则优化语言广度。本文使用模型卡和公开 Leaderboard 的数据进行比较，不凭感觉，也不夸大差异。","disclosurePrefix":"最后更新于 2026 年 7 月 18 日。我们开发","disclosureSuffix":"，这是一款同时提供三大模型系列的开源听写应用，因此我们会在生产环境中测试并支持每一种模型。这也说明了我们的立场：我们不销售其中任何模型，下面所述的实际取舍与我们向用户解释的完全相同。"},"factCheck":{"title":"关键事实速览（一手资料）","items":[{"lead":"Parakeet TDT 0.6B v3 在 Open ASR Leaderboard 的八项基准中平均词错误率为 6.34%，吞吐速度达到实时的 3,332×。","body":"600M 参数、25 种语言、CC-BY-4.0 许可证。","sources":[{"href":"https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3","label":"Parakeet TDT v3 模型卡"},{"href":"https://huggingface.co/spaces/hf-audio/open_asr_leaderboard","label":"Open ASR Leaderboard"}]},{"lead":"Whisper large-v3 在同一榜单上约为 7.4%，并支持 99 种语言","body":"拥有 1.55B 参数，采用 MIT 许可证。","sources":[{"href":"https://github.com/openai/whisper","label":"OpenAI Whisper 代码仓库"},{"href":"https://huggingface.co/openai/whisper-large-v3","label":"Whisper large-v3 模型卡"}]},{"lead":"Nemotron Speech Streaming EN 0.6B 在流式识别时的平均 WER 为 6.93%","body":"（1.12 秒音频块），在 LibriSpeech test-clean 上达到 2.32%，训练数据包含 530,000 小时音频。","sources":[{"href":"https://huggingface.co/nvidia/nemotron-speech-streaming-en-0.6b","label":"Nemotron Streaming EN 模型卡"}]},{"lead":"Nemotron 3.5 ASR 以单个 600M 流式模型覆盖 40 个语言区域","body":"支持自动语言检测，于 2026 年 6 月 4 日以 OpenMDW 许可证发布。","sources":[{"href":"https://huggingface.co/nvidia/nemotron-3.5-asr-streaming-0.6b","label":"Nemotron 3.5 ASR 模型卡"}]},{"lead":"下文提到的所有磁盘大小和应用内行为均可核实","body":"可在 OpenWhispr 的开源模型注册表中查证。","sources":[{"href":"https://github.com/OpenWhispr/openwhispr","label":"GitHub 上的 OpenWhispr"},{"href":"https://openwhispr.com/models","label":"OpenWhispr 模型页面"}]}]},"contenders":{"title":"三个候选模型","intro":"三者都是可免费下载、离线运行的开放权重模型。架构上的不同，几乎解释了实际使用中的所有差异。","items":[{"title":"OpenAI Whisper (2022)","body":"让开放语音识别进入主流的模型。它是一个编码器—解码器 Transformer，使用 680,000 小时音频训练，提供从 75MB（tiny）到 3GB（large-v3）的六种大小，覆盖 99 种语言。它像语言模型一样逐个 token 生成文字，因此也是三者中最慢的。MIT 许可证。"},{"title":"NVIDIA Parakeet (2024–2025)","body":"为批量转录设计的 600M 参数转导器（TDT）模型。它不采用自回归方式生成 token，而是在一次读取音频的过程中输出文字，速度快得多，也不会在静音时编造文字。两款重要版本是多语言 v3（25 种语言，INT8 ONNX 为 680MB）和仅支持英语的 Unified（631MB）；后者目前拥有领先的英语准确率。CC-BY-4.0 许可证。"},{"title":"NVIDIA Nemotron ASR (2026)","body":"面向流式识别的最新系列：具有缓存感知能力的 FastConformer 转导器会随音频到达而转录，无需等待录音结束。英语模型于 2026 年 1 月推出，覆盖 40 个语言区域的多语言 3.5 模型于 6 月跟进。两者均为 600M 参数（INT8 ONNX 为 632–650MB），采用开放的 OpenMDW 许可证。"}]},"accuracy":{"title":"准确率：Parakeet 胜出，但差距没有看起来那么大","definitionPrefix":"语音模型通常使用","definitionTerm":"词错误率（WER）","definitionSuffix":"衡量，即模型识别错误的单词比例，越低越好。标准参考是 Hugging Face Open ASR Leaderboard，它对八个不同的英语数据集（会议、财报电话会、TED Talks、有声书等）的 WER 取平均值，避免单个简单数据集人为抬高模型成绩。","analysis":"应当客观看待图表：这里的每款模型都已达到生产级别，总差距仅约 1.5 个百分点。在日常听写的清晰音频上，6.3% 与 7.4% 的体验接近；差别会在口音、噪声和专业术语等困难音频中显现。NVIDIA 数据值得关注的是取得成绩所需的成本：Parakeet 仅用 600M 参数便达到该准确率，而 Whisper 需要 1.55B；Nemotron 更是在完全看不到后续音频的流式条件下取得 6.93%。","calloutTitle":"Whisper 的已知弱点：在静音中产生幻觉","calloutBody":"Whisper 的解码器像语言模型一样工作，因此在静音或噪声中可能生成流畅但完全虚构的句子；整个生态都记录过这种故障。Parakeet 和 Nemotron 等转导器在结构上更能抵抗它：没有音频依据，就没有 token。对于开头和结尾常有静音的听写来说，这比基准中的一个百分点更重要。"},"speed":{"title":"速度与效率：完全不是一个级别","paragraph1":"在 Leaderboard 使用的硬件上，Parakeet TDT v3 的转录速度为实时的 3,332×，约一秒即可处理一小时音频。Whisper large-v3 的速度远不到它的十分之一。原因在于架构：Whisper 逐个 token 写出文本，每一步都要等待上一步；转导器只读取一次音频，并随进度输出文字。没有循环，也没有逐词往返处理。","paragraph2Prefix":"在笔记本电脑上，绝对数字会下降，但比例依然成立：Whisper large 需要数秒处理的一段听写，Parakeet 用 CPU、无需 GPU 就能近乎瞬间完成。独立基准还显示，在质量相当时，NVIDIA 转导器每小时音频大约少用","energyLink":"8–10× 的能耗","paragraph2Suffix":"；如果经常用电池长时间听写，这一点很实际。","paragraph3Prefix":"Whisper 的应对方案是 turbo（1.6GB），它是 large-v3 的蒸馏版本，以少量准确率换取约 8× Whisper-large 的速度。它是适合听写的 Whisper，但只能缩小差距，无法追平。完整的型号大小对比请参阅","guideLink":"Whisper 模型大小指南","paragraph3Suffix":"。"},"languages":{"title":"语言覆盖：Whisper 的护城河","headers":{"model":"模型","languages":"语言","notes":"说明"},"rows":[{"model":"Whisper large-v3","languages":"99","notes":"开放模型中覆盖最广，包括低资源语言"},{"model":"Nemotron 3.5 ASR","languages":"40 个语言区域","notes":"自动语言检测；15 种语言可直接转录，包括日语、韩语、阿拉伯语和印地语"},{"model":"Parakeet TDT v3","languages":"25","notes":"欧洲语言以及英语和俄语；自动检测"},{"model":"Parakeet Unified / Nemotron EN","languages":"1","notes":"仅支持英语，以换取同类最佳准确率"}],"rule":"经验法则很简单。英语或主要欧洲语言：三者都可用，按速度选择。日语、韩语、越南语、阿拉伯语和印地语：Nemotron 3.5 如今可提供流式识别，这是本地模型真正的首次突破。普通话、泰语、印度尼西亚语、斯瓦希里语或大量语言切换：Whisper 仍是唯一严肃的选择，短期内不太可能改变。"},"streaming":{"title":"流式识别：真正的分水岭","lead":"在流式识别方面，Nemotron 不只是小幅领先，而是在完成另外两者受架构限制无法完成的事。","paragraph1":"Whisper 以 30 秒窗口处理音频，并重新编码看到的全部内容。因此所谓“实时”Whisper，其实是把同一个批处理模型反复运行在相互重叠的音频上，成本高、延迟大，接缝处也不稳定。Nemotron 的缓存感知架构专为流式场景训练：它只处理最新音频块，延续内部状态，并在可配置的 80 毫秒至 1.12 秒延迟范围内输出部分文字。","paragraph2Prefix":"在大规模部署中，效率差异非常显著：NVIDIA 报告同一 GPU 上可实现比上一代流式模型","sourceLink":"多 17× 的并发流","paragraph2Middle":"。而在笔记本电脑上，您感受到的优势更直观：模型在本机 CPU 上运行，文字随说话出现。我们在","deepDiveLink":"流式 ASR 技术深入解析","paragraph2Suffix":"中详细介绍了将其用于桌面应用所需的工程工作。","paragraph3":"如果不在乎实时文字，只是说完后获取最终转录，那么流式识别没有额外价值，Parakeet 批处理模型更合适。流式识别适用于实时字幕、语音智能体以及边说边显示的听写预览。"},"hardware":{"title":"硬件：三者都能在普通笔记本上运行","paragraph1":"这些模型都不需要 GPU。NVIDIA 模型以 INT8 量化 ONNX 文件（631–680MB）提供，通过无需 Python 和 PyTorch 的原生二进制运行时 sherpa-onnx，可在任何较新的 CPU（Apple Silicon 或 x86）上流畅运行。Whisper 通过 whisper.cpp 可在各平台运行；small 和 turbo 适合 CPU，large-v3（3GB、~10GB RAM）则确实能从加速中获益。","paragraph2":"如果有 GPU，建议用于 Whisper：Apple Silicon 内置 Metal，OpenWhispr 一键提供 CUDA（NVIDIA）和 Vulkan（AMD 与 Intel）运行时，并在 GPU 运行失败时自动回退到 CPU。NVIDIA 转导器在 CPU 上已经足够快，对听写而言 GPU 加速意义不大。"},"decision":{"title":"应该选哪一个？","headers":{"situation":"您的需求","use":"建议使用"},"rows":[{"situation":"英语听写，追求最高准确率","use":"Parakeet Unified EN 0.6B"},{"situation":"欧洲语言，快速批量转录","use":"Parakeet TDT 0.6B v3"},{"situation":"说英语时实时显示文字","use":"Nemotron Speech Streaming EN"},{"situation":"以西班牙语、日语、韩语、阿拉伯语、印地语等实时显示文字","use":"Nemotron 3.5 ASR Streaming"},{"situation":"NVIDIA 模型未覆盖的语言","use":"Whisper large-v3 或 turbo"},{"situation":"旧设备或内存较少，记录简短笔记","use":"Whisper tiny 或 base"}],"privacyPrefix":"如果隐私是您选择本地方案的原因，三者同样合格：音频在本机处理，绝不上传。我们在","privacyLink":"本地转录与云端转录对比","privacySuffix":"中进一步讨论了这种取舍。"},"cta":{"title":"两分钟试完三个模型","paragraphPrefix":"最客观的选择方法，是用每个模型听写同一段话。OpenWhispr 在 macOS、Windows 和 Linux 上免费开源，并通过一个设置下拉菜单提供本文全部模型：六种大小的 Whisper、两款 Parakeet 和两款 Nemotron。下载模型，按下快捷键，然后说话。我们的","modelsLink":"模型页面","paragraphSuffix":"列出了全部模型的大小和取舍。","github":"在 GitHub 上加星"},"faqSection":{"title":"常见问题"}},"blogHipaaNoteTakers":{"metadata":{"title":"符合 HIPAA 的 AI 会议记录与听写软件（2026） | OpenWhispr","description":"哪些 AI 会议记录工具会签署 BAA，哪些又把 AI 功能排除在外。2026年的清晰对比，以及我们在 SOC 2 Type II 与 ISO 27001 上的立场。","keywords":["符合hipaa的ai会议记录","符合hipaa的笔记应用","hipaa ai 医疗记录助手","符合hipaa的转录","符合hipaa的听写软件","符合hipaa的语音转文字","otter ai 是否符合 hipaa","hipaa ai 记录心理治疗师","ai 会议记录 soc 2 hipaa","符合hipaa的会议转录"]},"breadcrumb":{"home":"首页","blog":"博客","article":"符合 HIPAA 的 AI 会议记录工具"},"schema":{"title":"2026年符合 HIPAA 的 AI 会议记录工具与听写软件","description":"哪些厂商会签署业务伙伴协议，哪些把 AI 功能排除在外，签约前该问什么，以及本地转录如何改变整个判断。"},"hero":{"title":"符合 HIPAA 的 AI 会议记录工具：BAA 究竟覆盖什么","subtitle":"多数自称「符合 HIPAA」的工具确实会签 BAA，然后把 AI 功能留在外面。以下是在患者数据交给厂商之前的核查方法，以及我们在 SOC 2 Type II、ISO 27001 和 HIPAA 上的立场。","date":"2026年8月11日","tag":"合规"},"sections":[{"id":"where-we-stand","title":"我们的立场：SOC 2、ISO 27001、HIPAA"},{"id":"what-hipaa-requires","title":"HIPAA 实际要求什么"},{"id":"ai-exclusion","title":"陷阱：当 BAA 排除了 AI"},{"id":"comparison","title":"哪些工具会签署 BAA"},{"id":"checklist","title":"向厂商提出的五个问题"},{"id":"local-processing","title":"本地处理为何改变这道算式"},{"id":"frameworks","title":"SOC 2 和 ISO 27001 不等于 HIPAA"},{"id":"faq","title":"常见问题"}],"opening":{"lead":"并不存在「通过 HIPAA 认证」的 AI 会议记录工具。","body":"没有任何机构会把软件认证为符合 HIPAA（美国《健康保险流通与责任法案》）。当厂商网站写着「已通过 HIPAA 认证」时，这句话出自市场团队，而不是审计师。真正重要的问题更为具体，也容易核实得多：他们是否签署业务伙伴协议（BAA），而该协议是否覆盖你打算用在患者数据上的那项功能？","thesis":"团队栽跟头的正是后半句。几家大型平台会为核心产品签下 BAA，却把 AI 功能留在外面——而那恰恰是你为之而来的功能。","updated":"最后更新于 2026年8月。厂商条款变动很快，在向厂商发送任何内容之前请先确认当前的覆盖范围。"},"whereWeStand":{"title":"我们的立场：SOC 2、ISO 27001、HIPAA","disclosure":"OpenWhispr 是我们做的，它也在下面的表格里。这是利益冲突，所以我们先把自己的立场讲清楚：请带着这一点阅读本页其余内容。","body":"我们是一支小团队。我们比同等规模的公司更早完成了 SOC 2 Type II、ISO 27001 和 HIPAA 就绪度评估，因为当有人问患者音频去了哪里时，「相信我们」不是一个答案。具体来说，目前存在的是这些：","items":[{"lead":"SOC 2 Type II 与 ISO 27001。","body":"均由独立机构评估，66 项控制措施持续监控，13 家次级处理方公开列出。最新状态见我们的信任中心。"},{"lead":"HIPAA，附带 BAA。","body":"自 Business 套餐起可应请求提供。我们并非「通过 HIPAA 认证」，因为这种证书根本不存在，对声称如此的人应当保持怀疑。"},{"lead":"第三方渗透测试。","body":"由外部机构执行。报告可应客户请求共享。"},{"lead":"一个报告问题的渠道。","body":"security@openwhispr.com，48 小时内确认收到。公布一个地址，只有在有人回复时才有意义。"},{"lead":"本地模式，它比上述任何一项都重要。","body":"选择本地 Whisper 模型，音频就在你自己的机器上完成转录，从不外传。既然录音从未到达我们这里，也就没有什么需要 BAA 去覆盖。"}],"limitsLabel":"这些认证不能说明的事。","limitsBody":"审计是底线，不是保证。它说明一家公司有控制措施并加以执行，却没有说明某项具体功能是否契合你的工作流程。OpenWhispr 通过全局快捷键可以在任何电子病历字段中听写，因此 Epic、Cerner、athenahealth 和网页版病历都无需集成即可使用。它不做的是通过 API 回写：没有 FHIR 或 HL7 集成，不知道当前打开的是哪位患者的病历，也不产生编码化或结构化数据。它同样不是经过临床验证的环境记录助手。如果你需要把记录归档到某次具体就诊并附带编码建议，专门的记录助手产品会比我们更合适。","trustLine":"SOC 2 报告、ISO 27001 证书、次级处理方名单和受监控的控制措施都发布在我们的<trustLink>信任中心</trustLink>，我们如何处理数据则写在<securityLink>安全页面</securityLink>。"},"whatHipaaRequires":{"title":"HIPAA 对 AI 会议记录工具的实际要求","intro":"如果厂商代表你创建、接收、存储或传输受保护的健康信息（PHI），HIPAA 就把它视为业务伙伴。由此产生一份简短而具体的清单：","items":[{"lead":"一份已签署的 BAA。","body":"这份合同让厂商对你交付的 PHI 承担法律责任。没有它，向厂商发送 PHI 本身就是违规，无论加密做得多好。"},{"lead":"安全规则要求的保护措施。","body":"访问控制、审计日志、传输与静态加密，以及书面的事件响应流程。"},{"lead":"泄露通知。","body":"他们必须不得无故拖延地告知你未受保护 PHI 的泄露，好让你赶上自己的期限。"},{"lead":"最小必要原则。","body":"只能按 BAA 允许的方式使用 PHI。这排除了二次用途，其中最需要盯住的就是用你的患者数据训练 AI 模型。"},{"lead":"向分包方传递义务。","body":"如果你的音频转到第三方模型提供商，那就是分包方，需要签署自己的 BAA。问清楚链路上都有谁。"}],"mythLabel":"关于认证的迷思。","mythBody":"HIPAA 没有认证机构，所以「通过 HIPAA 认证」并不是一个可以核实的说法。诚实的表述是「符合 HIPAA」「我们支持 HIPAA 合规」或「我们签署 BAA」。宣传证书的厂商，要么是在简化「我们完成了第三方就绪度评估」，要么就是没读懂法规。无论哪种，请索取 BAA，而不是那枚徽章。"},"aiExclusion":{"title":"陷阱：一份排除了 AI 的 BAA","intro":"2026年常见的失误不是厂商拒签 BAA，而是签了一份悄悄不覆盖 AI 的 BAA。该品类中两家最大的平台正是如此：","zoomLabel":"Zoom。","zoomBody":"Zoom 为 Zoom for Healthcare 以及符合条件的 Business 和 Enterprise 套餐提供 BAA，覆盖 Meetings、Webinars、Phone、Team Chat 和 Rooms。但启用 HIPAA 模式会停用一部分 AI Companion 功能，剩下的功能中也只有智能录制、会议摘要等少数落在 BAA 范围内。合规的那套配置反而能力更弱。","zoomSource":"Zoom 的 HIPAA 文档","notionLabel":"Notion。","notionBody":"Notion 会签署 BAA，但仅限席位数超过下限的 Enterprise 客户，且 Notion AI 在所有套餐中都在该覆盖范围之外。你可以把 PHI 存进 Notion 数据库。但不能对它使用 Notion AI。","notionSource":"Notion 的 HIPAA 帮助页面","takeaway":"这个模式反复出现，与这些产品的构建方式有关。合规团队审查平台，AI 稍后以另一套法律前提上线，而这项排除从未出现在销售对话中。所以请以书面形式、逐项功能地索取 BAA 的覆盖范围，而不是就整家公司要一个「是」或「否」。"},"comparison":{"title":"哪些 AI 会议记录工具会签署 BAA","intro":"2026年8月的状况，依据各厂商公开文档。最后一栏最关键，也是定价页面最少回答的一栏。","headers":{"tool":"工具","baa":"是否签署 BAA","plan":"所需套餐","aiCovered":"是否覆盖 AI 功能"},"vendors":[{"name":"OpenWhispr","baa":"是，可应请求提供","plan":"Business 套餐及以上","aiCovered":"覆盖，或 PHI 不离开设备","note":"自 Business 套餐起即可应请求提供 BAA，比 Otter 和 Fathom 设置的「仅限 Enterprise」门槛低一档。在本地模式下这个问题根本不会出现：转录使用开源 Whisper 模型在你的机器上运行，音频从不发送给我们。开发 OpenWhispr 的 Gizmo Labs 持有 SOC 2 Type II 和 ISO 27001，并运行 HIPAA 与 GDPR 合规项目。","href":null},{"name":"Otter.ai","baa":"是","plan":"仅限 Enterprise","aiCovered":"覆盖（在 BAA 范围内）","note":"Otter 在独立评估后于 2025年7月宣布符合 HIPAA，并签署覆盖 PHI 使用与披露的 BAA。只有 Enterprise 符合条件。Basic、Pro 和 Business 客户无法获得，因此升级到合规层级意味着实打实的价格跳涨。","href":"/compare/otter"},{"name":"Fathom","baa":"是","plan":"仅限 Enterprise","aiCovered":"覆盖（在 BAA 范围内）","note":"在 Enterprise 层级提供有 BAA 支撑的 HIPAA 合规。其模型提供商 Anthropic、OpenAI 和 Google 在合同上被禁止使用用户数据训练，这正是对分包方问题的正确回答。","href":"/compare/fathom"},{"name":"Fireflies.ai","baa":"是，可应请求提供","plan":"医疗层级","aiCovered":"覆盖（在 BAA 范围内）","note":"Fireflies 于 2025年9月推出面向医疗行业的方案，包含符合 HIPAA 的层级。BAA 不会自动生效：默认套餐不在覆盖范围内，必须主动申请。Fireflies 表示已与自身供应商签署 BAA，禁止其训练或存储 ePHI。","href":"/compare/fireflies"},{"name":"Zoom AI Companion","baa":"部分覆盖","plan":"Healthcare / 符合条件的套餐","aiCovered":"仅部分功能","note":"BAA 覆盖 Zoom 的核心产品，但 HIPAA 模式会关掉 AI Companion 的一系列功能，只有一部分仍在范围内。在依赖它撰写临床文档之前，请逐项确认。","href":null},{"name":"Granola","baa":"否","plan":"—","aiCovered":"不覆盖","note":"Granola 目前既不提供 HIPAA 合规，也不提供 BAA。产品不错，但在这一点改变之前不能用于 PHI。","href":"/compare/granola"},{"name":"Notion AI","baa":"部分覆盖","plan":"Enterprise，100 个席位以上","aiCovered":"不覆盖——AI 被排除","note":"Notion Enterprise 会签署覆盖页面、数据库、Wiki 和文件上传的 BAA，但 Notion AI 在所有套餐中都在其外。存放文档尚可，作为处理 PHI 的 AI 记录助手则不行。","href":null}],"compareLink":"对比 OpenWhispr 与 {name}","disclaimer":"根据 2026年8月各厂商公开文档整理。套餐名称、层级和 BAA 覆盖范围经常变动，请把本表当作自行尽职调查的起点，而非替代。请向厂商及贵方合规顾问核实当前条款。"},"checklist":{"title":"记录 PHI 之前该问的五个问题","intro":"多数安全页面只回答容易的问题，跳过那些决定你能否真正使用该产品的问题。以下五个请以书面形式提出：","questions":[{"lead":"你们签 BAA 吗？需要哪个套餐？","body":"如果答案是「仅限 Enterprise」，在对产品动感情之前先问价。合规层级的费用常常是最初报价的数倍。"},{"lead":"BAA 有没有点名 AI 功能？","body":"不是问「产品是否被覆盖」，而是问转录、摘要以及任何 AI 聊天或智能体功能是否各自在范围内。Zoom 和 Notion 与自家宣传出现落差的地方正在于此。"},{"lead":"哪些次级处理方会拿到 PHI？","body":"如果音频流向 OpenAI、Anthropic、Google 或某个托管语音服务，它们各自都是分包方，需要各自的 BAA 和不用于训练的承诺。请索取名单，而不是一句宽慰。"},{"lead":"我们的数据会用于训练或产品改进吗？","body":"「我们不出售你的数据」是另一回事，而且是弱得多的承诺。你要的是合同层面禁止训练、禁止人工查看你的录音。"},{"lead":"音频保留多久？能否设为零？","body":"实时处理并立即删除，和保留九十天再给一个删除按钮，完全不是一回事。无法配置的保留期，就是将来要在审计中辩解的保留期。"}]},"localProcessing":{"title":"本地处理为何改变这道算式","intro":"上面所有问题之所以存在，都是因为 PHI 离开了你的机构、落在别人的服务器上。当音频从不离开设备，绝大部分分析就不再适用：","points":["不存在向业务伙伴的披露，因为厂商从未收到 PHI。","没有需要审计的分包方链条，因为路径上没有第三方模型提供商。","没有需要谈判的保留策略，因为服务器端不存在副本。","不会因厂商日后改政策而受影响，因为模型跑在你自己掌控的硬件上。","无需联网即可运行，这在临床设备被刻意保持离线的机构里很重要。"],"tradeoff":"取舍是真实存在的，值得说明。本地模型需要一台还算新的机器，最大最准确的那些在旧硬件上会更慢，而且你放弃了让机器人代你参会的便利。对许多临床听写来说这笔账划算。对大型诊所的大批量环境记录来说，未必如此。","closing":"OpenWhispr 正是围绕这一点设计的，也正因如此，我们认为本地这个选项比我们持有的任何证书都更重要。"},"frameworks":{"title":"SOC 2 和 ISO 27001 不等于 HIPAA","intro":"这三者在采购沟通中常被混为一谈，但它们回答的确实是不同的问题。分清楚能为安全审查的双方省下时间：","items":[{"lead":"SOC 2 是一份报告，不是证书。","body":"由独立审计师依据 Trust Services Criteria 出具。Type I 看的是某个时点上控制措施如何设计。Type II 检验它们在一段时间内是否真的有效，通常为三到十二个月，值得索取的是后者。自称「通过 SOC 2 认证」的人，描述的其实是一份根本不以这种方式运作的文件。"},{"lead":"ISO 27001 确实是一项认证。","body":"由认可机构依据标准对组织的信息安全管理体系进行认证，三年为一周期并配合监督审核。请索取证书编号、颁发机构，以及最关键的适用范围声明，它说明实际覆盖了什么。"},{"lead":"HIPAA 是一部法律，根本没有证书。","body":"合规是一项持续义务，通过保护措施、政策、风险分析和已签署的 BAA 来证明。SOC 2 和 ISO 27001 能有力说明厂商认真对待安全。但两者都不能替代 BAA。"}]},"faq":{"title":"常见问题","legalNote":"这只是关于各厂商合规安排的一般性信息，不构成法律意见。你在 HIPAA 下的义务取决于贵机构的情况、风险分析以及法律顾问的判断。"},"faqs":[{"question":"Otter.ai 符合 HIPAA 吗？","answer":"符合，但仅限 Enterprise 套餐。Otter 在独立评估后于 2025年7月宣布符合 HIPAA，并签署覆盖 PHI 使用与披露的 BAA。Basic、Pro 和 Business 客户无法获得 BAA，因此在这些层级上并不符合。如需启动流程，请联系 Otter 的销售团队。"},{"question":"Fathom 符合 HIPAA 吗？","answer":"符合，在 Enterprise 层级并有 BAA 支撑。Fathom 的 AI 提供商 Anthropic、OpenAI 和 Google 在合同上被禁止使用用户数据训练。较低层级不在覆盖范围内，因此在录制任何临床内容之前，请确认你的套餐符合条件。"},{"question":"Notion 符合 HIPAA 吗？","answer":"部分符合。Notion 为席位数超过下限的 Enterprise 客户签署 BAA，覆盖页面、数据库、Wiki 和文件上传。但 Notion AI 在所有套餐中都被排除在该覆盖之外，因此即便你是已签 BAA 的 Enterprise 客户，也不能把 Notion 的 AI 功能用在受保护的健康信息上。"},{"question":"Zoom AI Companion 符合 HIPAA 吗？","answer":"只是部分符合。Zoom 会为 Zoom for Healthcare 以及符合条件的 Business 和 Enterprise 套餐签署 BAA，但启用 HIPAA 模式会停用多项 AI Companion 功能。智能录制、会议摘要等在范围内，其余则不在。在用于临床文档之前，请确认你需要的具体功能确实被覆盖。"},{"question":"Fireflies 符合 HIPAA 吗？","answer":"在其医疗层级上符合。Fireflies 于 2025年9月推出符合 HIPAA 的方案，但默认套餐不在覆盖范围内，必须明确申请 BAA。Fireflies 还表示已与自身次级处理方签署 BAA，禁止其训练或存储 ePHI。"},{"question":"Granola 符合 HIPAA 吗？","answer":"不符合。Granola 目前既不提供 HIPAA 合规，也不提供业务伙伴协议，因此不应用于记录或处理受保护的健康信息。"},{"question":"心理治疗师最适合用哪款符合 HIPAA 的 AI 记录工具？","answer":"这取决于你是否需要会议机器人。个人执业和小型诊所往往更适合设备端听写：会谈音频不离开笔记本电脑，这一步也不需要 BAA。如果需要能加入远程问诊通话的机器人，Otter Enterprise、Fathom Enterprise 和 Fireflies 的医疗层级都会签署 BAA。无论选哪个，都请查看保留设置和次级处理方名单，因为心理治疗记录属于 HIPAA 涵盖范围内最敏感的资料。"},{"question":"如果转录在我自己的电脑上本地运行，还需要 BAA 吗？","answer":"就该步骤而言通常不需要。如果音频完全在设备上转录、从未发送给厂商，那么厂商并没有代表你创建、接收、存储或传输 PHI，因此在这条链路中不充当业务伙伴。但你自己的保护措施依然必要：设备加密、访问控制，以及关于生成记录存放位置的政策。请向贵方合规顾问确认这一判断。"},{"question":"SOC 2 或 ISO 27001 是否意味着某款工具符合 HIPAA？","answer":"不意味着。SOC 2 是审计师关于安全控制的报告，ISO 27001 认证的是信息安全管理体系。两者都能有力说明安全工作是认真的，但都无法建立 HIPAA 所要求的合同关系。只有已签署的 BAA 才能做到。"},{"question":"「符合 HIPAA」和「通过 HIPAA 认证」有什么区别？","answer":"「通过 HIPAA 认证」不是一种真实存在的状态，因为没有任何政府机构或认可机构会依据 HIPAA 认证软件。使用这一说法的厂商通常指的是完成了第三方就绪度评估。你真正能核实的是：他们是否签署 BAA、该 BAA 覆盖什么，以及记录了哪些保护措施。"},{"question":"有没有免费且符合 HIPAA 的 AI 会议记录工具？","answer":"OpenWhispr 的本地模式免费且开源，由于音频从不离开你的设备，也就没有需要覆盖的披露行为。如果你想改用我们的云服务，自 Business 套餐起可签署 BAA。多数云端记录工具都把 BAA 绑定在 Enterprise 定价上，因此免费的云端层级几乎无法用于 PHI。在认定免费选项可行之前，请先确认所需套餐。"},{"question":"AI 会议记录工具会用患者数据训练吗？","answer":"在拟定得当的 BAA 之下不会。协议应当明确禁止厂商及其每一个次级处理方将 PHI 用于模型训练或产品改进。请要求把该条款写进书面文件。隐私政策里关于不出售数据的一句话，远比不上合同层面的训练禁令。"}],"cta":{"title":"留在你机器上的听写与会议记录","body":"OpenWhispr 在本地运行开源 Whisper 模型，你的音频无需离开设备。若确实需要我们的云服务，自 Business 套餐起可签署 BAA。免费、开源，支持 macOS、Windows 和 Linux。","github":"在 GitHub 上点亮星标"}},"useCaseLawyers":{"breadcrumb":{"home":"Home","useCases":"Use Cases","lawyers":"For Lawyers"},"metadata":{"title":"Free Dictation Software For Lawyers & Attorneys - OpenWhispr","description":"Dictation software for lawyers. OpenWhispr offers Cloud transcription or local processing for sensitive and offline work.","keywords":["legal dictation software","lawyer dictation","attorney dictation software","legal transcription","dictation for law firms","offline legal dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Legal dictation software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source dictation software for lawyers. Choose Cloud transcription without a local model download, or local processing for sensitive and offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation software for lawyers & attorneys","title":"Speak the letter. It lands in Word. No typist ever hears the file.","body":"Dictate a letter, a file note or a memo and the clean text drops in wherever your cursor already is: Word, Clio, an email. Choose Cloud for fast transcription or local for highly sensitive and offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Man speaking beside an open laptop on a sunny terrace"}},"trustBar":{"label":"Dictate into any app, including:","names":["Clio","iManage","NetDocuments","MyCase","Smokeball","Microsoft Word"]},"pain":{"label":"What you're choosing between","title":"Saying the letter takes two minutes. Getting it typed is the rest of the afternoon.","subtitle":"Every route from your voice to the file should suit the matter, not keep you typing.","cards":[{"title":"A stranger typing up privileged notes.","body":"With a local model and no Cloud fallback, the dictation is transcribed on the machine you spoke into. No outsourced transcription service handles it."},{"title":"Sensitive matters need a local option.","body":"Choose local models when the matter is highly sensitive or you will work without signal. Your device processes the recording."},{"title":"An hour of typing is an hour you can't bill.","body":"Typing up your own dictation is the slowest way to get a letter out, and you do it after the thinking is already done. Dictate it while the matter is fresh and you review and send instead of type."},{"title":"A tool that only fits one desk.","body":"OpenWhispr runs on macOS, Windows, and Linux. Start with the Free plan, then upgrade when the workflow calls for it."}]},"features":{"label":"From your voice to the file","title":"From your voice to a first draft.","subtitle":"Get the letter out while the matter is fresh. Choose the transcription path that fits the work.","items":[{"image":{"label":"Woman speaking while using a laptop at a desk"},"title":"Cloud transcription without a local model download","body":"Speak a letter, a memo or a file note into your mic. Cloud gives you fast transcription without a local model download, then immediately discards the transcription audio."},{"image":{"label":"Man reviewing a printed document beside his laptop"},"title":"Letters and memos, drafted for you","body":"Tell it what you want and use local AI models for device-only drafting. You read every word before it leaves your desk."},{"image":{"label":"Man speaking and pointing toward a laptop at a table"},"title":"Dictate into Word without an add-in","body":"One hotkey, and the clean punctuated text pastes itself where you're already typing: Word, an email, Clio, your document system. There's nothing for IT to install into any of them."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Privileged audio sent to us","value":"None","footnote":"with a local transcription model the recording never leaves your computer"},{"label":"Cost per lawyer","value":"$$0","footnote":"unlimited local dictation in the Free plan"},{"label":"Processing options","value":"2","footnote":"Cloud transcription or local models"}]},"howItWorks":{"label":"Download to first letter","title":"Three steps from download to draft.","subtitle":"Choose Cloud transcription or download a local model.","image":{"label":"Man standing beside a laptop in a glass-walled office"},"steps":[{"title":"Download it free.","body":"Mac, Windows or Linux. Setup offers OpenWhispr Cloud or local models. Pick local for privileged work and it downloads them once."},{"title":"Dictate the letter.","body":"Speak it the way you'd say it to your assistant, and it comes back as text."},{"title":"Read it, paste it, delete the audio.","body":"Edit the draft, drop it into Word or your document system, and delete the recording. Every one of those is yours to control."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Why firms keep it","title":"Ten reasons your IT or security team can check for themselves.","subtitle":"Choosing dictation software for lawyers is a confidentiality decision before it's anything else. Most of these are facts about the code, so IT or security can check them instead of taking them from us.","items":[{"title":"The recording stays on your machine","body":"Choose a local transcription model for device-only processing. An optional Cloud fallback can send audio if you turn it on."},{"title":"Free for the whole firm","body":"Unlimited local dictation at $0 for every desk, including your assistant and paralegals."},{"title":"Nothing goes to a typing pool","body":"With local transcription and no Cloud fallback, no outsourced transcription service handles the recording."},{"title":"Works with the wifi off","body":"A courthouse hallway, a deposition, a client's office, a flight. Local models need no signal, so nothing stalls mid-sentence."},{"title":"Dictate into Word or any case system","body":"One system-wide hotkey drops clean text where your cursor already is: Word, Clio, iManage, an email. On a Mac it uses the accessibility permission you grant at setup."},{"title":"Letters and memos, not just transcripts","body":"Ask for a case memo and a local AI model can draft it on your machine. Local AI models are included on every plan."},{"title":"You control storage and deletion","body":"Recordings are files on your machine. Audio is deleted after 30 days out of the box, and Settings moves that or switches it off."},{"title":"Nothing to take on faith","body":"Your IT or security team can read every line of the app on GitHub before a partner installs it."},{"title":"Party names and citations, spelled your way","body":"Put party names, case names, statutes, and citations in the custom dictionary to guide transcription and cleanup."},{"title":"100+ languages","body":"Dictate in 100+ languages."}]},"faq":{"label":"Before you install it","title":"Eight questions law firms ask before they install anything.","subtitle":"Straight answers on confidentiality, cost, accuracy, and what IT will say. More questions? Ask in Discord.","items":[{"question":"Does the audio ever leave my device?","answer":"With local transcription, audio is transcribed on your device. Turn off the optional Cloud fallback when every attempt must stay local. Cloud sends audio only to transcribe it, then immediately discards it."},{"question":"Is this consistent with my duty of confidentiality?","answer":"Confidentiality is your professional duty and no tool discharges it for you. Local processing is the deliberate choice when a highly sensitive matter calls for device-only work. Cloud transcription discards audio immediately after transcription. This is not legal advice, and the rules of your bar or law society still apply."},{"question":"Does dictation software for lawyers get case names and citations right?","answer":"Add party names, case names, Latin, statutes, and citations to the custom dictionary. OpenWhispr passes that list to transcription and cleanup to guide the text."},{"question":"Will IT let me install it on the firm laptop?","answer":"It is a desktop app, not an add-in for Word or your document system. On a Mac, grant accessibility permission for auto-paste. Without it, copy the text from your clipboard instead."},{"question":"Will it work in a courthouse or on a flight with no signal?","answer":"Yes, once the models are on your machine. Local models run entirely offline after the first download. Dictate in the courthouse hallway, after a deposition, or on the flight home, then file the note when you are back in range."},{"question":"Does it draft the letter, or just transcribe?","answer":"Both. Dictation turns speech into text, then an AI model can shape a client letter or case memo. Transcription and AI cleanup are separate settings, so choose a local AI model when the draft must stay on-device."},{"question":"Do I need a separate legal dictation tool?","answer":"OpenWhispr runs on macOS, Windows, and Linux. Try the Free plan with a local model, then move to Cloud transcription when it fits your workflow."},{"question":"Is it actually free?","answer":"Free includes unlimited local dictation and local AI models, plus 2,000 Cloud words a week. Pro adds unlimited Cloud transcription, 20 hours of meeting recordings, sync, mobile, and API/MCP access. Business adds unlimited meeting recordings, Agent mode, and chat over data."}]},"reviews":{"label":"What people tell us","title":"Built by people who wouldn't hand their audio to a stranger either.","subtitle":"From our Discord, GitHub and inboxes. None of them is a lawyer, so read them for the tool, not the practice.","score":{"value":"0","label":"Audio stored on our servers with local transcription","body":"With a local transcription model the audio never leaves your device, so there's nothing on our side to lose, breach or hand over."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free to start.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for lawyers","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Lawyers who want local dictation and 2,000 Cloud words a week"}]},{"parts":[{"text":"Lawyers who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Firms documenting at volume with Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local dictation plus 2,000 Cloud words a week"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Enterprise adds SSO, audit logs, retention controls, and dedicated support."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"A client tells you things they'd tell nobody else.","paragraphs":["OpenWhispr started with a simple objection: confidential dictation should have a local option.","We build in public. The app is open source, so your firm can inspect how it works before anyone installs it.","Community support is available in Discord. Paid plans add email or priority support."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"$$0 a seat · open source · no add-in","title":"Speak the next letter while the matter is fresh.","body":"Download OpenWhispr, choose Cloud transcription or local processing, and dictate the next letter.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free on macOS, Windows, and Linux. Paid plans add Cloud transcription, mobile, and workflow tools."}}},"useCaseLegalDocuments":{"breadcrumb":{"home":"Home","useCases":"Use Cases","legalDocuments":"Legal Documents"},"metadata":{"title":"Free AI Dictation For Legal Documents - OpenWhispr","description":"AI dictation for legal documents: talk a clause, a brief section or a motion through and OpenWhispr's agent drafts it. Free forever, and it runs locally.","keywords":["ai dictation for legal documents","dictate a contract","brief and motion dictation software","discovery response drafting tool","deposition summary dictation","paralegal and legal ops dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Legal document drafting software","operatingSystem":"macOS, Windows, Linux","description":"OpenWhispr is an open-source dictation app for the people who produce legal documents: paralegals, legal assistants and lawyers. They dictate contracts, briefs, motions, discovery responses and deposition summaries, and the AI agent redrafts them into structured text, either in place in the document or in the app's assistant panel.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"AI dictation software for legal documents","title":"Stop typing first drafts. Talk the section through.","body":"Talk a contract clause, a motion section or a discovery response through, then tell the AI agent what to make of it: headings, numbered points, a summary. Pick a local model and it's drafted on your own machine, free and unlimited on every plan.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman speaking beside a laptop at an office desk"}},"trustBar":{"label":"Dictate:","names":["Contract clauses","Motion sections","Discovery responses","Deposition summaries","Case memos","Engagement letters"]},"pain":{"label":"Why dictating a contract stalls","title":"Talking it through was never the slow part. Shaping it is.","subtitle":"Speak an engagement letter or a case chronology and you'll get every word back in one unbroken block. Cutting that block into clauses, headings and numbered points is what eats the afternoon, and plain dictation hands that part straight back to you.","cards":[{"title":"One block of speech, not a clause in it.","body":"Dictation cleanup hands back what you said, tidied. It will break the speech into paragraphs and number a list you clearly dictated. What it won't do is impose a heading or restructure a deposition summary into sections. That job belongs to the AI agent, and you have to ask it for it."},{"title":"The motion is the sensitive part, not the audio.","body":"A transcript is words. A motion is the argument you're about to run and the position your client will take. Choose local processing when you want the composing to run on your machine, or cloud tools when they fit your workflow."},{"title":"Paralegals and legal ops draft it too. Per-seat pricing adds up.","body":"An associate dictates the contract, a paralegal checks the cites, an assistant cleans it up, legal ops files it. Charge per seat and the cost follows the draft across four desks. OpenWhispr charges nothing for a seat on Free, so each of them can use it."},{"title":"Party names and statutes come back mangled.","body":"A general-purpose model has never seen your matter. Put the case names, the parties, the Latin and the citations your practice uses in the custom dictionary. A dictation vocabulary only fixes what was heard; this one also reaches the model that drafts, so it sees those names before it writes a word. It moves the odds rather than guaranteeing them, so read what carries legal weight before the document goes out."}]},"features":{"label":"From microphone to document","title":"You dictate a section. You get back something worth editing.","subtitle":"A transcript isn't a document. Dictation, cleanup and the AI agent each hand you something different.","items":[{"image":{"label":"Woman speaking while working on a laptop"},"title":"Say it once, read it back punctuated","body":"Talk an interrogatory response through the way you'd say it out loud, and it comes back transcribed and punctuated with the filler and the false starts taken out. Address the agent instead and it works from the same dictation, handling the disfluencies itself."},{"image":{"label":"Woman reading a printed document beside her laptop"},"title":"Headings and numbered points, not a wall of speech","body":"Cleanup breaks your speech into paragraphs and numbers a list you clearly dictated, and stops there. Ask the AI agent instead, on its own hotkey or by name mid-dictation, and it composes what you asked for: the headings, the numbered points and the lists you talked through."},{"image":{"label":"Man speaking while reviewing a printed document beside a laptop"},"title":"Nothing to install into Word or your DMS","body":"One global hotkey copies the text and pastes it at your cursor, so the clause lands in Word, a matter record, an email or a web form the same way. Auto-paste is on by default, so you edit it where it lands."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Audio kept on your own machine by default","value":"30 days","footnote":"Audio Retention ships on at 30 days. Settings takes it from 1 day to 90, or switches it off"},{"label":"Languages you can dictate a matter in","value":"100+","footnote":"a German framework agreement comes back as German text"},{"label":"Cost to add the paralegal and legal ops","value":"$$0/month","footnote":"no card details, and no per-seat licence to buy"}]},"howItWorks":{"label":"How a contract gets drafted","title":"Talk through the motion, then edit what comes back.","subtitle":"Three steps from talking the argument through to editing it in the document.","image":{"label":"Man speaking beside a laptop in a book-lined office"},"steps":[{"title":"Install it and choose where it runs.","body":"Setup puts a local model and OpenWhispr Cloud in front of you. Going local takes extra model downloads, but keeps the drafting path on your machine. Once that choice is made, place the cursor where the legal text belongs or highlight the passage you want the AI agent to reshape."},{"title":"Talk the clause through, ums and all.","body":"Say the deposition summary the way you'd explain it to a colleague, ums and all. Cleanup hands the same words back with the filler and the false starts stripped out."},{"title":"Ask the agent for headings and numbered points.","body":"Highlight the rough passage, press the agent hotkey and tell it what you want, and it rewrites that passage in place in the document. With nothing highlighted the draft arrives in the app's assistant panel instead, ready to copy across. Either way it's plain markdown text rather than Word styles, so you shape it in the document itself."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Ten claims you can check","title":"What OpenWhispr does to a legal document, and where it stops.","subtitle":"What it does, what it won't do, and where the document ends up. No asterisks.","items":[{"title":"The setting that would upload a contract arrives switched off","body":"Pick a local model at setup and the audio, the transcript and the composed section all stay on the machine you spoke into. Two deliberate acts send content to us: switching on paid cloud sync, which arrives off and stores your notes, transcripts, agent conversations and your custom dictionary, party names included; and sharing a note or moving it into a team space."},{"title":"Cleanup keeps your wording and won't add a sentence you didn't say","body":"Cleanup is told to keep your voice, your wording and your intent, and never to add a sentence you didn't say. It still fixes grammar and punctuation, splits run-on sentences, drops false starts, and writes spoken numbers and dates out in standard form. That's an instruction to a model rather than a guarantee, so read an indemnity back before it goes out."},{"title":"The agent turns talk into a numbered section","body":"Say its name mid-dictation or press its hotkey, tell it what you want, and it composes that from what you talked through: headings, numbered points, a summary of the argument. Highlight a passage first and it rewrites that passage in place; with nothing highlighted the draft lands in the assistant panel. Pick local at setup and it composes on the LLM you downloaded, on your own machine. Styling stays a Word job."},{"title":"One hotkey, and no add-in for IT to approve","body":"The text goes to the clipboard and pastes itself where your cursor is. On macOS, without Accessibility permission it can copy for you but not paste."},{"title":"Know where the ceiling is before you hit it","body":"For selection edits, select 6,000 characters or fewer."},{"title":"Dictate it into a note now, shape it later","body":"Save the case chronology as a note while it's fresh, and a note action turns it into clean markdown when you come back to it."},{"title":"Recordings are files, so deleting one really deletes it","body":"Recordings are audio files on your machine you can open or delete yourself; notes and transcripts sit in a local database the app deletes rows from. Audio Retention ships on at 30 days, and Settings takes it from 1 day to 90 or switches it off. Transcripts are the other way round, kept until you set a window of your own, so set one alongside the audio. OpenWhispr doesn't encrypt that local data itself, so the folder wants the care you'd give any other matter folder."},{"title":"Send procurement the repo instead of a policy PDF","body":"OpenWhispr is open source. When IT asks what happens to a contract you dictated, they can inspect the code rather than take our word for it."},{"title":"The paralegal and the legal ops lead get the same hotkey","body":"Legal assistants and legal ops draft these documents alongside lawyers. Free is a download rather than a licence, so the person who assembles the interrogatory response has the same hotkey as the person who signs it."},{"title":"Dictate the deal in the language it was done in","body":"Whisper-class models cover 100+ languages, so the deal you did in Spanish comes back transcribed in Spanish."}]},"faq":{"label":"Before the first document","title":"What actually comes back when you dictate a contract.","subtitle":"Anything we've missed, bring it to our Discord, where the people who build OpenWhispr answer.","items":[{"question":"Which legal documents can AI dictation draft: a contract clause, a motion section, a discovery response?","answer":"All three, plus case memos, deposition summaries and file notes. Worth knowing which part does what: plain dictation hands back your own words, tidied up. The AI agent, on its own hotkey or called by name mid-sentence, drafts them into a structured section when you ask it for one. Highlight a passage first and it rewrites that passage in place; with nothing highlighted the draft arrives in the assistant panel to copy across. What you get is markdown text, so ## Heading arrives in Word as those literal characters and not a Word style. Pleading captions and court-ready tables aren't its job either, so the formatting your jurisdiction wants still happens in Word."},{"question":"Who else ends up with a copy of an unsigned contract?","answer":"Nobody, as long as you picked a local model at setup and left the cloud alone. Whisper-class transcription runs on your machine, and the composing runs on the LLM you downloaded, served by llama.cpp at 127.0.0.1, so the clause never travels. Two things send content to us, and both are deliberate acts: cloud sync, a paid setting that stays off until you switch it on, which stores your notes, your dictation transcripts, your agent conversations and your custom dictionary, party names included, in our cloud database; and sharing a note or putting it in a team space, which uploads that note on any signed-in account. One more thing if you sign in: the assistant panel can run a web search, which sends the query the model composed, not your document, to us. Stay signed out and it is never offered."},{"question":"Can it write a whole brief in one go?","answer":"For a brief, work section by section: facts, then argument, then relief."},{"question":"Will party names and citations survive a deposition summary?","answer":"Load the custom dictionary first. It holds the words your matters actually turn on: party names, case names, statutes, the Latin, the citations your practice uses. It's a hint the model sees rather than a find-and-replace, so it moves the odds a long way in your favour without promising a clean sweep. It reaches local Whisper transcription, the cleanup pass, the AI agent and translation, across 100+ languages."},{"question":"Can a paralegal take over the contract after I've dictated it?","answer":"Yes. The hotkey copies the text and pastes it wherever the cursor is, so the words land in the file the two of you already share: Word, the matter record, an email. Auto-paste is on by default. On macOS it needs Accessibility permission, or it falls back to leaving the text on your clipboard."},{"question":"What does it cost to run a matter's documents through it?","answer":"Nothing on the local path, where dictation and drafting are unlimited on every plan, Free included, and neither one is metered. The cloud side has two separate meters. On Free you get 2,000 words a week of cloud speech-to-text, and 5 hours a month of cloud meeting recording. Pro at $8 a month makes cloud speech-to-text unlimited and takes meeting recordings to 20 hours a month. Business at $16 per user makes meeting recordings unlimited and adds team features."}]},"reviews":{"label":"Straight from our users","title":"Rambling in, editable text out.","subtitle":"Nobody was asked for these. One of them stumbles mid-sentence and still gets a clean paragraph back, which is what dictating a clause actually sounds like.","score":{"value":"0","label":"Client documents sitting on our servers","body":"In local mode, signed out, with cloud sync off and nothing shared, nothing reaches us. Turn cloud sync on, a paid setting that arrives switched off, and your notes, dictation transcripts, agent conversations and custom dictionary are stored in our cloud database."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free to get started.","Upgrade when you want more."],"caption":"What each plan covers for legal document work","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Paralegals and legal teams drafting contracts, briefs and motions, who don't mind downloading a speech model and an LLM first"}]},{"parts":[{"text":"Legal assistants who'd rather dictate through OpenWhispr Cloud than manage model downloads, at the desk and on an iPhone"}]},{"parts":[{"text":"Legal ops teams handling document work across a lot of matters, and recording calls all month"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Free plan: $0."}]},{"parts":[{"text":"The cloud speech-to-text meter comes off, and device sync and the iPhone app come with it"}]},{"parts":[{"text":"Cloud meeting recordings go unlimited and team features unlock, at $16 per user per month"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud and does not require local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Read the code yourself","title":"A contract shouldn't need a round trip to get written.","paragraphs":["Turning a transcript into clauses and numbered points takes an AI model. OpenWhispr's AI agent can do that work with a model you downloaded yourself, served by the llama.cpp binary that ships inside the app, on 127.0.0.1.","You don't have to take our word for that. OpenWhispr is open source, so your risk team can inspect the code rather than read a policy about it.","Talk the section through while the argument is fresh, have the AI agent shape it into headings or numbered points, then read what carries legal weight before the document goes out."],"image":{"label":"The people who build it"}},"finalCta":{"tagline":"$$0 a seat · No card details","title":"Say the clause out loud. Get a numbered section back.","body":"Download it free, pick a local model at setup, then hand the AI agent your rough dictation. Highlight the passage and it rewrites the clause or the case memo in place, as plain markdown text. No card details, no seat to buy.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Available on macOS, Windows, and Linux. Get community support in Discord."}}},"useCaseVeterinary":{"breadcrumb":{"home":"Home","useCases":"Use Cases","veterinary":"For Vets"},"metadata":{"title":"Free Offline Veterinary Dictation Software - OpenWhispr","description":"OpenWhispr is free, open-source veterinary dictation software that turns spoken findings into SOAP notes on your device, even on a farm call with no WiFi.","keywords":["veterinary dictation software","vet dictation app","veterinary transcription","SOAP note dictation","veterinary practice software","offline vet dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Veterinary dictation software","operatingSystem":"macOS, Windows, Linux","description":"Local-first, open-source veterinary dictation software. Dictated findings are transcribed on the vet's own device and never leave it. Free for unlimited local dictation, and works fully offline.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Veterinary dictation software for vets & vet techs","title":"Say it once. Get the SOAP note.","body":"OpenWhispr turns that dictation into a SOAP note, a discharge or a client email on your own computer. No per-DVM subscription, and it still works in a barn with no signal.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Veterinarian speaking beside a dog and laptop in a clinic"}},"trustBar":{"label":"Dictate into:","names":["ezyVet","Cornerstone","Avimark","Covetrus Pulse","Shepherd","Microsoft Word"]},"pain":{"label":"The documentation squeeze","title":"Documentation shouldn't cost your evenings and $99 a vet a month.","subtitle":"You want the note written for you. You don't want a subscription per DVM, a recorder in the exam room, or a tool that dies the moment you leave WiFi.","cards":[{"title":"Charting follows you home.","body":"Turn dictated text into a structured note with a custom action."},{"title":"A per-DVM subscription is another monthly line item.","body":"The Free plan includes unlimited local AI models and 100+ language support at $0."},{"title":"No signal, no scribe.","body":"Barn calls, farm visits, a back room with one bar. A cloud scribe can stall exactly where you are, leaving you to rebuild the note from memory that night. OpenWhispr runs on your device, fully offline."},{"title":"A recorder in the euthanasia room.","body":"Some conversations shouldn't be uploaded to a stranger's server to become a note. You dictate the findings afterwards, and with a local model everything is processed on your machine. Nothing sits on the table while you talk a client through it."}]},"features":{"label":"What you get","title":"Records closed before you lock up. No per-DVM bill.","subtitle":"An AI scribe listens to the room so you never have to say it. OpenWhispr waits for you to dictate, then a local model drafts the note from a prompt you saved yourself, on your own laptop.","items":[{"image":{"label":"Veterinarian speaking beside a dog on an exam table and a laptop"},"title":"Dictated at the treatment table, transcribed on your laptop","body":"Signalment, history, exam findings, assessment and plan, straight into the mic. Pick a local model and it's transcribed on your laptop, never uploaded, and you decide what's kept and when it goes."},{"image":{"label":"Veterinarian handing paperwork to a dog owner at a clinic doorway"},"title":"One dictation, two documents","body":"Save your own SOAP, discharge and client-email prompts as one-click actions, then run whichever one the case needs. The same dictation becomes the record for the file and the instructions the client reads in the car park. You check both before either is filed."},{"image":{"label":"Veterinarian using a laptop beside a dog in a rural clinic"},"title":"Types into the PIMS field you're already in","body":"Press a hotkey and dictate into the Cornerstone note field, an ezyVet consult tab, a discharge template or a referral email to the specialist. Clean, punctuated text, processed locally."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Client conversations uploaded","value":"None","footnote":"in local mode, not the estimate conversation, not the goodbye"},{"label":"Cost per DVM, per tech","value":"$$0","footnote":"unlimited local dictation, including the relief vet who works Saturdays"},{"label":"Setup takes","value":"No fixed time","footnote":"Install the app, then choose a local model or OpenWhispr Cloud."}]},"howItWorks":{"label":"How it works","title":"Dictate on the walk back to treatment.","subtitle":"This is one app on the practice laptop, with no PIMS migration or IT project.","image":{"label":"Veterinarian carrying a tablet past a dog in a treatment room"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Pick a local model during setup."},{"title":"Dictate the case.","body":"Signalment, presenting complaint, exam findings, assessment and plan. In the room, in the treatment area, or in the truck between farms. It's transcribed on your device, and one click turns it into the draft."},{"title":"Review, paste, close the record.","body":"Fix what it got wrong, drop it into the patient's record in ezyVet or Cornerstone, and delete the audio. All of it happens on your laptop."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons to choose OpenWhispr","subtitle":"In a three-vet practice this is a budget decision the owner signs off on, not an IT project. Here are the facts, and the practice manager can hand the repo link to whoever does your IT.","items":[{"title":"Choose local when dictation must stay on your device","body":"Dictation, and note drafts with a local model, are processed on your machine. The open-source code shows there's no upload step."},{"title":"Every DVM and every tech, $0","body":"Unlimited local dictation, forever. Scribes charge per DVM and leave the techs writing records for free anyway."},{"title":"Works in the barn and in the truck","body":"The local model runs with the laptop in airplane mode. Dictate in the truck, file the note when you're back in range."},{"title":"Dictate into the PIMS you already run","body":"A system-wide hotkey drops clean text wherever your cursor is: ezyVet, Cornerstone, Avimark, a discharge template, a referral email. No integration to build, and no reason to change PIMS."},{"title":"One dictation, the SOAP note and the discharge","body":"Save the prompt once. A local model, or your own AI key, turns the same dictation into the record for the file and the instructions the client takes home."},{"title":"No recorder in the room","body":"You dictate afterwards. Nothing sits on the table during the estimate, and nothing is running when you put a hand on a client's shoulder."},{"title":"Choose how long local records stay on the practice laptop","body":"When you enable local audio or transcript retention, files stay on this device and are deleted after the retention period."},{"title":"Read the code before the owner signs anything","body":"Every line is on GitHub. The practice manager can send the link to whoever does your IT and have an answer the same week."},{"title":"Knows cephalexin from Cerenia","body":"The custom dictionary learns the drug names, breeds and procedures you correct, and you add the abbreviations your clinic already writes: OHE, TPR, BAR, QAR."},{"title":"100+ languages","body":"Take the history from a client who would rather explain it in Spanish, transcribed on your laptop."}]},"faq":{"label":"Before you download","title":"The questions vets ask first.","subtitle":"Straight answers on cost, terminology and what happens when you leave WiFi. If yours isn't here, ask us in Discord.","items":[{"question":"Do you record the appointment, like a scribe does?","answer":"No. Nothing joins the consult and nothing sits on the table. You dictate afterwards, at the treatment table or in the truck, and in local mode that dictation is transcribed on your own laptop with no upload step. The code is public, so you can check that instead of taking our word for it."},{"question":"Do vets even need this? HIPAA doesn't cover animals.","answer":"Right. HIPAA covers human patients, not pets. But the record is still confidential, and the audio carries the client's name, their address and the hardest conversation of your day. Local-first keeps all of it off a vendor's server."},{"question":"Does veterinary dictation software get drug names and breeds right?","answer":"Yes, once you teach it yours. It runs Whisper- and Parakeet-class models trained on a wide range of accents and languages, and the custom dictionary learns the terms you correct: Apoquel, Cerenia, brachycephalic, Bichon Frise, OHE, plus whatever abbreviations your clinic already writes."},{"question":"Will it work on a farm call with no signal?","answer":"Yes. Local models run entirely offline. Dictate in the barn or the truck, and file the note once you're back in range. Cloud scribes can't do this, because they stall the moment you leave WiFi."},{"question":"Can it write the discharge as well as the SOAP note?","answer":"Yes, from the same dictation. Run a local model (Llama, Qwen, Gemma) or connect your own AI key, then save your own SOAP, discharge and client-letter prompts as one-click actions and run whichever one the case needs. A local model keeps it all on your machine; your own key means the model and data path are ones you already trust."},{"question":"Is it actually free?","answer":"Yes. The Free plan includes unlimited local AI models, 100+ languages, and desktop availability on macOS, Windows and Linux at $0. Pro is $8/month and adds unlimited OpenWhispr Cloud transcription, sync across devices, and the iPhone and iPad app."}]},"reviews":{"label":"What users say","title":"Built by people who wouldn't put a recorder in the room either.","subtitle":"Three OpenWhispr users describe how they use it.","score":{"value":"0","label":"Consults on a server","body":"You dictate afterwards, so there is no appointment recording to begin with. In local mode the dictation is never uploaded either."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for veterinarians","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Vets and techs who want free, private, offline dictation and are fine with a quick local setup"}]},{"parts":[{"text":"Vets who want easy, reliable cloud transcription"}]},{"parts":[{"text":"Practices and teams documenting at volume"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Free offline dictation, with consult audio staying on your device"}]},{"parts":[{"text":"Unlimited cloud transcription without managing local models or keys"}]},{"parts":[{"text":"Unlimited recordings plus advanced workflow features for a team"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Enterprise includes team features and admin controls, plus dedicated support."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"The goodbye appointment doesn't belong on a vendor's server.","paragraphs":["OpenWhispr started with a simple objection: the three conversations that take the most out of you, the diagnosis, the estimate and the goodbye, shouldn't have to pass through a stranger's server to become a note.","We build in public. The code is on GitHub. When you choose local processing, dictation stays on your device. A practice can verify that in the code.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Last appointment at six. Records closed at six.","body":"Download OpenWhispr and dictate the next case on the walk back to treatment. Choose local models when dictation needs to stay on your laptop and work offline.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Unlimited local AI models are free on macOS, Windows and Linux."}}},"useCaseJournalists":{"breadcrumb":{"home":"Home","useCases":"Use Cases","journalists":"For Journalists"},"metadata":{"title":"Free AI Speech To Text Tool For Journalists - OpenWhispr","description":"Free, open-source AI speech to text for journalists. In local mode the interview is transcribed on your laptop, never uploaded. No per-minute bill.","keywords":["ai speech to text tool for journalists","transcription software for journalists","interview transcription software","offline interview transcription","private transcription for reporters","local speech to text","free interview transcription"]},"schema":{"name":"OpenWhispr","applicationCategory":"Journalism transcription software","operatingSystem":"macOS, Windows, Linux","description":"Local-first, open-source speech to text software for journalists. In local mode interview audio is transcribed on the reporter's own device and never leaves it. Free for unlimited local transcription.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"AI speech to text for journalists & reporters","title":"Drop in the tape. Get the whole transcript.","body":"Get the whole transcript without uploading the tape. In local mode, OpenWhispr transcribes on your machine, free and open source, with no per-minute bill or signal once the model is downloaded.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Person typing on a laptop beside a field recorder in an open vehicle"}},"trustBar":{"label":"Paste your quotes into:","names":["Google Docs","Microsoft Word","WordPress","Arc XP","Slack","Substack"]},"pain":{"label":"What an upload actually costs you","title":"Source protection shouldn't end at the upload button.","subtitle":"You want the tape transcribed on deadline. You don't want it living in someone else's account to get there.","cards":[{"title":"Keep source audio on your device in local mode.","body":"Upload an interview and a vendor holds the recording. Some services route the audio to human transcriptionists. In local mode OpenWhispr transcribes on your device, so there is no server for the tape to sit on."},{"title":"Your tape becomes someone else's record.","body":"Every upload leaves unpublished material in an account you don't administer, on a retention schedule you didn't set. In local mode there is no upload and no account, so there is nothing on our side for anyone to subpoena."},{"title":"A quote you can't check.","body":"Speech to text mishears names and crosstalk, and every model will occasionally write a sentence nobody said. A summary will not show you where that happened. Record the interview in the app and the transcript is stamped line by line. To check a quote before you file, compare it with the original recording you keep."},{"title":"\"Trust us\" is not source protection.","body":"Closed tools ask you to take a privacy policy on faith. OpenWhispr's code is public on GitHub under the MIT license. Your security desk, or any developer you trust, can read exactly what it does and what it doesn't."}]},"features":{"label":"What you get","title":"Filed faster. The tape stays yours.","subtitle":"Transcription tools usually trade source protection for speed, or bill you by the minute. This does neither. In local mode the interview is transcribed on your machine, at no cost per minute.","items":[{"image":{"label":"Two people talking across a table with an open laptop"},"title":"Interview files transcribed on-device","body":"Drop in an MP3, WAV, M4A or half a dozen other formats, one file or a whole batch. In local mode it's transcribed on your machine, with no length limit and no per-minute bill."},{"image":{"label":"Person reviewing printed pages beside a laptop and phone"},"title":"The whole transcript, never a summary","body":"You get the full transcript, not a digest. Record the interview in the app and every line is stamped on your device, ready to export as .txt, .md, .json or .srt. On an upload, speaker detection stamps each turn locally; per-line stamps need your own Groq or Mistral key, or OpenAI set to whisper-1."},{"image":{"label":"Two people talking at a desk with a field recorder and notebook"},"title":"Live transcript from the room","body":"Put your laptop on the table and it transcribes an in-person interview as you talk, on any platform. On a remote interview it captures your computer's own audio, so no bot joins and nothing shows in the participant list. That capture needs Windows, macOS 14.2 or later, or a Linux desktop whose audio portal allows it. Recording is its own setting, so point that one at a local model too."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Source audio sent to us","value":"None","footnote":"in local mode, your audio is transcribed on your own device"},{"label":"Cost per hour of tape","value":"$$0","footnote":"unlimited local transcription, every reporter, forever"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first transcript, plus the one-time model download"}]},"howItWorks":{"label":"How it works","title":"Transcribed before you write the lede.","subtitle":"No account or credit card required. Choose a local model in Settings, then drop in the tape or record in the app.","image":{"label":"Person at a desk with a laptop, notebook, and audio recorder"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Skip the account and setup hands you a local model. Either way, point uploads and recording at one in Settings."},{"title":"Drop in the tape.","body":"MP3, WAV, M4A, FLAC and more, one file or a batch. In local mode it's transcribed on your device, however long the interview ran."},{"title":"Check it, then file.","body":"Read the transcript in the app and check any line against the audio. A recorded interview exports to .txt, .md, .json or .srt with every line stamped."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"On the record","title":"10 reasons to choose OpenWhispr","subtitle":"Picking an AI speech to text tool is a source-protection decision before it's anything else. Here are the facts, and the ones about your audio you can check in the code yourself.","items":[{"title":"Local mode keeps source audio on your device","body":"Transcription runs on your machine, with no upload step."},{"title":"No bot joins a remote interview","body":"It records your computer's own audio, on Windows, macOS 14.2 or later, or a Linux desktop that allows it. Nothing appears in the participant list, and on a local model no vendor gets a copy of the call."},{"title":"Offline, once the model is downloaded","body":"Courthouse steps, a jail visit, a rural county meeting. Once the model is downloaded, no signal required."},{"title":"No import limits, no per-minute bill","body":"Local transcription is unlimited at $0, on every plan."},{"title":"Your security desk can read the code","body":"MIT licensed, every line on GitHub. A newsroom security review can read exactly what happens to the audio, before you put a source's tape through it."},{"title":"Recorded interviews, stamped line by line","body":"Export to .txt, .md, .json or .srt with every line stamped, so you can point standards or the lawyers at the exact moment."},{"title":"No length limit on your own machine","body":"A ninety-minute interview costs the same as a nine-minute one, and takes as long as your laptop needs."},{"title":"Who said what, on every plan","body":"Two voices in one interview, labelled. Speaker detection runs on your device, so it works on a local model with nothing uploaded."},{"title":"Nothing rewrites your quotes","body":"An uploaded interview is never rewritten. The note is the transcript as the model heard it. In dictation the untouched text sits beside the cleaned-up version, and you can switch cleanup off entirely."},{"title":"100+ languages","body":"Interview a source in their own language, not their second one. In local mode, transcription stays on your machine."}]},"faq":{"label":"Before you download","title":"The questions reporters ask first.","subtitle":"Straight answers, including where it falls short. Anything missing, ask in Discord.","items":[{"question":"Does my source's audio ever leave my device?","answer":"Not in local mode. Transcription runs on your machine and the open-source code proves there's no upload step. Cloud transcription is the default, so switch to a local model in Settings before your first interview. Dictation, recordings and uploaded files each carry their own transcription setting, so change the one you use for tape. If you sign in, point the AI cleanup model at a local one too, or switch cleanup off: that setting writes the note title and it defaults to the cloud."},{"question":"Can it transcribe a file I already recorded?","answer":"Yes. Drag in an MP3, WAV, M4A, FLAC, AAC, OGG, Opus or WebM file, pick it from a file dialog, or queue a batch. On a local model there's no length limit and no per-minute charge. Local video files aren't supported, but you can paste a YouTube or video link and it pulls the audio, up to six hours or 500 MB."},{"question":"Is it accurate enough to quote from?","answer":"It runs Whisper-class and NVIDIA Parakeet models, and no speech to text is perfect on accents, names or crosstalk. That's why you get the whole transcript, never a summary: record the interview in the app and it's stamped line by line, so you can check the quote against the tape before you file. Treat it as a first pass you verify, not a transcript you publish unread."},{"question":"Can it tell my source's voice from mine?","answer":"Yes. On a local model the speaker detection runs on your device too, with nothing uploaded. It separates the voices in an in-person interview and labels who said what, on every plan; the transcript is live as you talk and the labels land when you stop. A recording expects two speakers and goes up to eight, enough for a panel. On an uploaded file, switch labels on before you transcribe and it works out the count for you."},{"question":"If we get a subpoena, what do you have?","answer":"In local mode, nothing. There is no upload and no account, so there is no tape and no transcript on our side to hand anyone. What exists is on your laptop, under your own retention and source-protection rules. That is also why offline works: once the model is downloaded, a courthouse or a jail visit needs no connection."},{"question":"Is it actually free?","answer":"Yes. Unlimited local transcription, speaker labels, 100+ languages, all three desktop platforms, $0 forever. Pro ($8/month) adds unlimited OpenWhispr Cloud speech-to-text and mobile dictation. Business ($16/user/month) adds unlimited cloud meeting recordings and priority support."}]},"reviews":{"label":"What users say","title":"Different jobs. The same instinct about their own audio.","subtitle":"From our Discord, GitHub and inboxes. None of them is a reporter, so read them for the tool, not the beat.","score":{"value":"0","label":"Interviews on a vendor's server","body":"In local mode the interview never leaves your device. There is no copy on our side, because nothing is sent to us."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for journalists","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Reporters who want free, offline transcription and don't mind setting up a local model"}]},{"parts":[{"text":"Reporters who'd rather not manage a local model and are fine with cloud transcription"}]},{"parts":[{"text":"Desks that transcribe at volume, or need unlimited cloud meeting recordings"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"The tape never leaves your laptop, at $0"}]},{"parts":[{"text":"Unlimited cloud transcription without managing local models or keys"}]},{"parts":[{"text":"Unlimited cloud meeting recordings and priority support for a desk"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some conversations only happen off the record. We built for those.","paragraphs":["OpenWhispr started with a simple objection: a tape of a source who asked for confidentiality shouldn't end up on a stranger's server just so you can read back what they said.","We build in public. The code is on GitHub. When we say the audio stays local, that isn't just a promise in a privacy policy. Anyone on your desk can inspect the code and verify what happens to the audio.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Your tape stays between you, your source and your laptop.","body":"Download OpenWhispr, point it at a local model, and your next interview is transcribed on your machine and nowhere else.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseContentCreators":{"breadcrumb":{"home":"Home","useCases":"Use Cases","contentCreators":"For Content Creators"},"metadata":{"title":"Free Dictation Software For Content Creators - OpenWhispr","description":"OpenWhispr is free, open-source dictation software for content creators. Paste your own video link or an audio file, get the transcript, talk out the rest.","keywords":["dictation software for content creators","dictation for youtubers","youtube video transcript software","transcribe your own youtube video","voice to text for creators","speech to text for video scripts"]},"schema":{"name":"OpenWhispr","applicationCategory":"Dictation software for content creators","operatingSystem":"macOS, Windows, Linux, iOS","description":"Local-first, open-source transcription and dictation software for content creators. Paste the link to a video you posted yourself, or drop in an audio file, and get the transcript in a note you can search and quote. Then talk out the title, the description and the captions and the clean text lands wherever the cursor already is. With a local model chosen the audio is transcribed on the creator's own device. Free for unlimited local transcription and dictation on macOS, Windows and Linux, and it works fully offline.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation software for content creators","title":"Paste your own video link. Get the transcript.","body":"Get a transcript of your own YouTube video or audio file on your laptop for $0. Then use the same hotkey to dictate the title and description.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"A person speaks to a camera in a home studio with a ring light and laptop nearby"}},"trustBar":{"label":"Dictate into any app, including:","names":["YouTube Studio","CapCut","Notion","Patreon","Buffer","Discord"]},"pain":{"label":"What you're choosing between","title":"Everything you said is in the video. None of it is in text.","subtitle":"A transcript is what turns one upload into chapters, a newsletter and a thread. Getting one means paying per minute, or sitting through your own video with a keyboard.","cards":[{"title":"Your back catalogue is locked inside the videos.","body":"Turning last year's uploads into a newsletter means sitting through them again. Paste the link to one of your own videos and OpenWhispr pulls the audio down and transcribes it. Paste a list of your links, up to 50 at a time, and it works through the queue."},{"title":"The interview is on your disk and nowhere else.","body":"The raw recording, the voice memo, the podcast cut that never went up. Drop the audio file in and it comes back as a transcript in a note you can search. mp3, wav, m4a, flac, aac, opus, ogg and webm, straight off the disk."},{"title":"The description box, still blank at midnight.","body":"The export is done and the title, the description and the pinned comment are not. Press a hotkey in the tab you already have open and say them. Clean, punctuated text lands where the cursor is."},{"title":"Another subscription before the channel pays for itself.","body":"Editor, thumbnails, scheduler, transcripts, newsletter. Five tools billing one channel that does not cover them yet. Unlimited local transcription and dictation on OpenWhispr is $0, on every machine you own, and the free plan is not a trial."}]},"features":{"label":"From the upload to the words around it","title":"The transcript is there before you sit down to write.","subtitle":"Pick local models and both the transcription and the clean-up run on your own computer, unlimited, at no cost, in 100+ languages.","items":[{"image":{"label":"A person works at a laptop in a home editing setup with audio equipment nearby"},"title":"A transcript out of a video you already posted","body":"Paste the link to a video you posted yourself, or a list of them one per line, and OpenWhispr downloads the audio and transcribes it into a note you can search and quote. Up to 50 links at a time, 500 MB and six hours a video. Playlists and live streams are out."},{"image":{"label":"A person speaks beside a camera and laptop at a kitchen table"},"title":"The audio file that never got posted","body":"Interviews, voice memos, the podcast cut, the raw take. Drop in mp3, wav, m4a, flac, aac, opus, ogg or webm and it comes back as a transcript. On a local model there is no per-minute bill, and a file already on your disk has no size limit at all."},{"image":{"label":"A person speaks while using a laptop beside handwritten notes"},"title":"One hotkey, every tab you publish from","body":"Talk into YouTube Studio, your script doc, Notion, an email to a sponsor. The clean text pastes itself where you were already typing, with no plugin in any of them. On a Mac you grant accessibility once, and that is the part that lets it paste at your cursor."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Sent to our servers","value":"None","footnote":"with a local transcription model the audio is transcribed on your own computer"},{"label":"Cost per creator","value":"$$0","footnote":"unlimited local transcription and unlimited local dictation, on every machine you own"},{"label":"Setup takes","value":"A few extra steps","footnote":"download, pick a model, paste a link or set a hotkey"}]},"howItWorks":{"label":"Download to first transcript","title":"Three steps from your posted video to searchable text.","subtitle":"Nothing to connect to the platforms you publish on.","image":{"label":"A person speaks into a microphone at a desk with a laptop and camera gear nearby"},"steps":[{"title":"Download it free.","body":"Mac, Windows or Linux, whatever the edit rig runs. Setup offers OpenWhispr Cloud or local models, and picking local downloads them once."},{"title":"Paste the link, or drop the file.","body":"Your own video's URL, a list of them one per line, or an audio file off your disk. OpenWhispr pulls the audio down and transcribes it into a note."},{"title":"Read it, use it.","body":"Search it, quote it, cut it into chapters and a newsletter. Then hit the hotkey in YouTube Studio and talk the title and the description out."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Why creators keep it","title":"10 reasons content creators choose OpenWhispr","subtitle":"Every one is in the app or in the repo. Check before you download anything.","items":[{"title":"Paste your own video link, get the transcript.","body":"Give OpenWhispr the YouTube URL of a video you posted, a direct audio URL, or an audio file from your disk. It transcribes the audio and can queue up to 50 links; playlists and live streams are not supported."},{"title":"It types into everything you publish from.","body":"Use the system-wide hotkey in YouTube Studio, your script doc, newsletter or sponsor email. On a Mac, one accessibility permission lets OpenWhispr paste at the cursor."},{"title":"Unlimited local transcription costs nothing.","body":"The free plan is not a trial. With a local model there is no per-minute meter on a file, no word cap on dictation, and no card to enter."},{"title":"Your niche words come out spelled right.","body":"The custom dictionary takes your channel name, your recurring guests and the vocabulary only your audience uses. Dictation hands that list to the transcription model as a hint, and to the clean-up model after it."},{"title":"The ums and the false starts come out.","body":"The clean-up model strips filler, stutters and restarts, and is told to keep your voice, your wording, your formality and your intent, with your jargon and names exactly as spoken."},{"title":"100+ languages.","body":"Dictate the description in the language your audience actually comments in. Same hotkey, same model, no separate tool."},{"title":"Location shoots with no signal.","body":"A rented studio, a hotel, a plane, a shoot in the middle of nowhere. Local models need no network at all."},{"title":"Nothing is uploaded when you keep it local.","body":"Pick a local transcription model and the audio is transcribed on your own machine, so none of it reaches us. Point the clean-up at a local model too and the text stays put as well."},{"title":"The code is public.","body":"OpenWhispr is open source on GitHub, where 5,000+ engineers have starred it. Every privacy claim on this page is a file you can open and read."},{"title":"Mac, Windows and Linux, one free download.","body":"Whatever you edit on. The same app, the same free price, and a hotkey you choose yourself."}]},"faq":{"label":"Before you install it","title":"The questions creators ask before they download it","subtitle":"Short answers, and where in the app or the repo you can check them.","items":[{"question":"Can I get a transcript of a video I already published?","answer":"Paste the link to your own video. OpenWhispr accepts a YouTube URL or a direct audio file URL, downloads the audio and transcribes it into a note. Paste a list of your links, one per line and up to 50 at a time, and it works through them as a queue. Playlists and live streams are not supported, and the download caps at 500 MB and six hours a video. On a local model that cap is the only one. On OpenWhispr Cloud the free plan transcribes files up to 25 MB, and past that it is a paid plan."},{"question":"Can I upload the video file itself?","answer":"Not the video file. The upload panel takes audio: mp3, wav, m4a, webm, ogg, oga, flac, aac and opus. For something you already posted, paste its link instead and OpenWhispr pulls the audio down for you. For a cut that is not posted anywhere yet, export the audio out of your editor and drop that in."},{"question":"Is my audio uploaded anywhere?","answer":"Not if you pick a local transcription model: the audio is transcribed on your own computer and nothing is sent. One thing worth knowing, because it is two settings and not one. Transcription and the AI clean-up are chosen separately, and if you are signed in the clean-up still defaults to OpenWhispr Cloud. On that default the transcript text, never the audio, is sent to us to be tidied. Set the clean-up model to a local one too and neither leaves the machine."},{"question":"Is it really free, or is the free plan a trial?","answer":"Free is free, with no card. Unlimited local dictation and unlimited local transcription are $0 on every plan, forever. Pro ($8/month) makes OpenWhispr Cloud transcription unlimited, and Business ($16/month) adds unlimited cloud meeting recordings and agent mode. Nothing on the free plan expires."},{"question":"Will it get my channel name and my niche jargon right?","answer":"Add them to the custom dictionary once. Your channel name, your recurring guests, the in-jokes and the product names your audience uses. Dictation passes that list to the transcription model as a hint, and the clean-up model gets the same list. One exception worth knowing: the local Parakeet model does not take the hint, so use a local Whisper model if the dictionary matters."},{"question":"Does it work on a shoot with no WiFi?","answer":"Yes, once the local models are downloaded. Local transcription and a local AI model both run on your own machine with no network, so a hotel, a plane or a location with no signal is the same as your desk. Cloud transcription is the option that needs a connection."}]},"reviews":{"label":"What people tell us","title":"They tried it on their own machine and stopped typing.","subtitle":"From our Discord, GitHub and inbox, word for word. None of them runs a channel, so read them for the tool.","score":{"value":"0","label":"Plugins to install in YouTube Studio","body":"The hotkey works the same in a browser tab, your edit suite and your newsletter draft. Nothing to connect, nothing to log into, no integration to keep alive."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for content creators","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"One-person channels that want unlimited local transcription and dictation at no cost and will set up a local model"}]},{"parts":[{"text":"Creators who would rather pay $8 than look after a local model"}]},{"parts":[{"text":"Channels where an editor, a writer and a producer all touch the same content"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Transcripts of your own videos, and everything you write around them, for nothing, with the audio staying on your machine"}]},{"parts":[{"text":"Cloud transcription that never runs out, with no models or API keys to look after"}]},{"parts":[{"text":"Unlimited cloud meeting recordings and agent mode for everyone on the channel"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"The fastest way to sound like yourself is to say it out loud.","paragraphs":["OpenWhispr started with an objection: talking is how people are fluent and typing is how they get stiff, and the software that turns one into the other kept routing everything through somebody else's server.","We build in public. Every release and every line of the app is on GitHub. So when we say the dictation can stay on your machine, you do not have to take our word for it. You can open the file and read it.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"$$0 · open source · no plugin to install","title":"Get the transcript, then write around it.","body":"Download OpenWhispr, pick a local model at setup, and paste the link to the video you posted this morning. No card, no per-minute bill, no second subscription.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free forever on Mac, Windows & Linux."}}},"useCasePodcasters":{"breadcrumb":{"home":"Home","useCases":"Use Cases","podcasters":"For Podcasters"},"metadata":{"title":"Best Speech-to-Text Software For Podcasting - OpenWhispr","description":"OpenWhispr is free, open-source speech-to-text for podcasters. Transcribe unlimited episode audio on your own machine and never pay per minute again.","keywords":["speech to text for podcasting","podcast transcription software","free podcast transcription","dictation software for podcasters","podcast show notes software","offline podcast transcription","open source transcription software"]},"schema":{"name":"OpenWhispr","applicationCategory":"Podcast transcription software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Local-first, open-source speech-to-text for podcasters. With a local model chosen, episodes and show notes are transcribed on the creator's own device. Free for unlimited local transcription and dictation, and works fully offline.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text software for podcasters","title":"Record the episode. Get the transcript. Pay nothing per minute.","body":"Turn each recorded episode into a transcript, then dictate show notes, descriptions and posts into any app. Choose local transcription for offline work, or Cloud for high accuracy and speed without downloading models.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Person speaking into a studio microphone beside a laptop and handwritten notes"}},"trustBar":{"label":"Write show notes in:","names":["Notion","Buzzsprout","Transistor","Spotify for Creators","Substack","YouTube Studio"]},"pain":{"label":"The problem with per-minute transcription","title":"Transcription shouldn't cost more than your hosting.","subtitle":"A transcript and show notes are part of shipping many episodes. Many tools bill by the minute or cap usage by the month.","cards":[{"title":"Paying by the minute to transcribe your own voice.","body":"A weekly one-hour show, metered per minute, turns into a real monthly bill that never ends. OpenWhispr transcribes on your machine: unlimited hours, $0, and the Free plan is not a trial."},{"title":"Show notes take longer than the episode.","body":"You talked for an hour, then typed for two. Press a hotkey and speak the summary, the guest bio and the links straight into your host or your docs. Clean, punctuated text, at speaking speed."},{"title":"Guest names turn show notes into a correction job.","body":"When your guest's name or show title comes out wrong in dictated notes or AI-drafted summaries, you have another correction job. Add those terms to OpenWhispr's dictionary to improve how they appear."},{"title":"Pulling one quote means scrubbing the whole episode.","body":"Clips, quote graphics and newsletters all start with finding the line. With a transcript on your machine you search text, copy the quote and move on. No scrubbing."}]},"features":{"label":"What you get","title":"Transcripts, show notes, summaries. No meter on any of them.","subtitle":"Make transcripts and show notes on your own machine in 100+ languages, without adding another monthly transcription bill.","items":[{"image":{"label":"Person monitoring audio at a microphone with a laptop and recording interface"},"title":"Unlimited local transcription","body":"Import the episode you already cut, and it is transcribed on your device with local Whisper-class models. No per-minute fees or monthly hour caps for local transcription, with nothing uploaded in local mode."},{"image":{"label":"Person speaking into a desktop microphone while working on a laptop"},"title":"Show notes at speaking speed","body":"Press a hotkey and speak your summary, description or newsletter straight into Notion, Buzzsprout, Substack, wherever your cursor is."},{"image":{"label":"Two people recording with microphones while one reviews notes beside a laptop"},"title":"A dictionary that knows your show","body":"Add your show's title, recurring guests and niche vocabulary to the custom dictionary. It improves how those names and terms are handled in dictation and AI-drafted notes."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Unreleased audio uploaded","value":"None","footnote":"in local mode the episode never leaves your machine"},{"label":"Cost per episode","value":"$$0","footnote":"unlimited local transcription and dictation, every episode, forever"},{"label":"Transcription mode","value":"Local or Cloud","footnote":"Local runs on your device. Cloud prioritizes accuracy and speed with no model download."}]},"howItWorks":{"label":"How it works","title":"A transcript and show notes before you close the editor.","subtitle":"Choose local transcription when you need offline work or want audio to stay on your machine. Choose Cloud for high accuracy and speed without downloading local models.","image":{"label":"Person holding headphones beside a closed laptop and handwritten notes"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Pick a local model during setup."},{"title":"Record or dictate.","body":"Record a session for a local transcript, or press the hotkey and speak your show notes into any app."},{"title":"Review, paste, publish.","body":"Edit the text, paste it into your host or your docs, and ship the episode. You control all of it."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons podcasters choose OpenWhispr","subtitle":"Transcription cost recurs with every episode. These are the practical details to compare.","items":[{"title":"Unlimited local transcription","body":"No per-minute fees or monthly hour caps for local transcription on the Free plan."},{"title":"Free for your whole show","body":"Unlimited local dictation and transcription at $0: you, your co-host, your editor. No per-seat licence."},{"title":"Dictate straight into your host","body":"A system-wide hotkey drops clean text wherever your cursor is: Buzzsprout, Notion, Substack, YouTube Studio."},{"title":"Summaries drafted, not just transcribed","body":"Connect your own AI key, or run a local model, to turn a transcript into an episode summary or description."},{"title":"A dictionary for your guests' names","body":"Teach it the show title, recurring guests and niche terms once. The custom dictionary improves how it handles them in dictated notes and drafted summaries."},{"title":"Works fully offline","body":"Transcribe on the flight home from the conference. Local models never need a signal."},{"title":"Pull quotes without scrubbing","body":"The transcript is text on your machine. Search it, copy the line, make the clip."},{"title":"Unreleased audio stays on your machine","body":"In local mode nothing is uploaded. Embargoed interviews and unshipped episodes never touch a server."},{"title":"Open source, auditable by anyone","body":"OpenWhispr is open source under the MIT License."},{"title":"100+ languages","body":"Record in the language your audience actually speaks, transcribed locally."}]},"faq":{"label":"Before you download","title":"The questions podcasters ask first.","subtitle":"Straight answers on cost, accuracy and privacy. If yours isn't here, ask us in Discord.","items":[{"question":"What's the best speech-to-text software for podcasting?","answer":"OpenWhispr is a strong option if you want to transcribe recorded episodes locally without per-minute fees. It can turn imported audio into a transcript and lets you dictate show notes wherever your cursor is. Its open-source code makes the workflow inspectable. Trying it costs a download."},{"question":"Is it really free?","answer":"Yes. Local dictation and transcription are included at $0 on macOS, Windows and Linux. The Free plan also includes 2,000 words a week of OpenWhispr Cloud transcription and 5 hours a month of meeting recordings. Pro is $8/month with unlimited Cloud transcription and 20 meeting-recording hours. Business includes unlimited meeting recordings."},{"question":"Will it get my guests' names and niche terms right?","answer":"OpenWhispr supports 100+ languages. Its custom dictionary lets you add your show's title, guest names and niche vocabulary to improve how they appear in dictation and AI-drafted show notes."},{"question":"Does it transcribe my recorded episodes, or only what I say live?","answer":"Both. Import an episode you've already recorded, or record a session in the app, and the transcript is made on your device. Press the hotkey to dictate show notes into any app, and connect your own AI key, or a local model, to draft summaries from the transcript."},{"question":"Does it work offline?","answer":"Yes, once the models are downloaded at setup. After that they run entirely offline. Transcribe an interview on the plane and publish when you land."},{"question":"Do unreleased episodes stay private?","answer":"In local mode, the audio stays on your machine. OpenWhispr is open source under the MIT License. Cloud transcription is optional: choose local for offline work or Cloud for high accuracy and speed without downloading models."}]},"reviews":{"label":"What users say","title":"Built for people who'd rather talk than type.","subtitle":"What people say about OpenWhispr.","score":{"value":"0","label":"Invoices for your own words","body":"Local transcription runs on your machine. There is no meter, no hour cap, and no bill at the end of the month."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for podcasters","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Get unlimited Cloud transcription"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Podcasters who want free, unlimited local transcription and are fine with a quick local setup"}]},{"parts":[{"text":"Podcasters who want easy cloud transcription and mobile dictation"}]},{"parts":[{"text":"Networks and teams producing shows at volume"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local transcription, with your audio staying on your machine"}]},{"parts":[{"text":"Unlimited cloud speech-to-text without managing local models or keys"}]},{"parts":[{"text":"Unlimited cloud recordings plus advanced workflow features for a team"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Your words are the show. You shouldn't rent them back.","paragraphs":["OpenWhispr started with a simple objection: turning your own voice into text shouldn't cost by the minute or pass through someone else's server.","We build in public on GitHub. Local transcription is not just a privacy-policy promise. It is a property of the code you can inspect.","And if anything breaks the night before an episode drops, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Stop paying by the minute for your own voice.","body":"Download OpenWhispr, pick a local model at setup, and your next episode gets transcribed, summarized and shipped on your own computer.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free desktop app for macOS, Windows and Linux."}}},"useCaseProductTeams":{"breadcrumb":{"home":"Home","useCases":"Use Cases","productTeams":"For Product Teams"},"metadata":{"title":"Free AI Meeting Notes For Product Teams, No Bot - OpenWhispr","description":"OpenWhispr is free, open-source AI meeting notes for product teams. No bot joins the call, and with a local model the roadmap never leaves your machine.","keywords":["AI meeting notes","meeting notes for product teams","local meeting transcription","private meeting notes","bot-free notetaker","offline meeting notes"]},"schema":{"name":"OpenWhispr","applicationCategory":"Meeting notes software","operatingSystem":"macOS, Windows, Linux","description":"Local-first, open-source AI meeting notes for product teams. With a local model the call is transcribed on the user's own machine with no bot joining, and written up as decisions, action items and owners. Local meeting notes are unlimited and free, and work fully offline.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"AI meeting notes for product teams","title":"Run the meeting. Get the decisions written up.","body":"Get a clean meeting write-up with decisions and next steps. OpenWhispr captures the call on your own computer; pick a local model and the recording never reaches a server, ours or anyone's.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Two people listen to a colleague during a meeting around a table"}},"trustBar":{"label":"Paste the notes into:","names":["Linear","Jira","Notion","Confluence","Slack","Productboard"]},"pain":{"label":"The trade-offs product teams notice","title":"Product decisions shouldn't disappear when the call ends.","subtitle":"The PM should be facilitating, not typing. OpenWhispr turns the discussion into decisions, action items and owners, without adding a bot. A local model keeps transcription on your computer.","cards":[{"title":"A bot shows up to every meeting.","body":"Meeting bots can make a sensitive discovery call feel different. OpenWhispr captures your computer's audio, so it does not join the call."},{"title":"Your strategy on a vendor's server.","body":"Roadmap, pricing, competitive analysis, reorg talk: product meetings can cover sensitive work. Choose a local model when you want transcription to run on your device."},{"title":"Someone stops contributing to take notes.","body":"The PM types instead of facilitating, and the decision still gets lost. OpenWhispr writes up the decisions, action items and owners, so everyone stays in the discussion."},{"title":"Per-seat pricing for the whole org.","body":"Choose the mode and plan that fit the team. Local transcription is unlimited on every plan; Free includes 5 hours of meeting recordings a month, Pro adds unlimited Cloud transcription and 20 recording hours, and Business includes unlimited meeting recordings."}]},"features":{"label":"What you get","title":"Turn the product conversation into decisions and next steps.","subtitle":"Capture the call without a bot, review a structured write-up, then paste it into the tools where the team works. Choose a local model when the discussion needs to stay on your computer.","items":[{"image":{"label":"A person gestures while speaking to a colleague across a table"},"title":"Captures the call from your computer, no bot","body":"Listens to your computer's audio on Zoom, Meet, Teams or a Slack huddle, or to the mic for a standup in the room. Transcribed on your device with a local model, and nothing joins the call."},{"image":{"label":"Two colleagues review a laptop together during a group discussion"},"title":"Decisions, action items and owners","body":"Turns your rough notes and the transcript into a structured write-up: what was decided, what happens next, who owns it. You review before it's shared."},{"image":{"label":"A person gestures while working on a laptop at a table"},"title":"Paste it where the work lives","body":"Copy the clean write-up into Linear, Jira, Notion or Slack. It's text, so there's nothing to connect to your tracker."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Roadmap calls sent to us","value":"None","footnote":"pick a local model at setup and there is no upload step in the code"},{"label":"Cost per person on the call","value":"$$0","footnote":"unlimited local meeting notes, the whole org, forever"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first meeting written up"}]},"howItWorks":{"label":"How it works","title":"Leave the call with the decisions already written down.","subtitle":"Download the app, choose a model, then start the call.","image":{"label":"A person holding a tablet speaks in an open office"},"steps":[{"title":"Download for free.","body":"macOS 14.2+, Windows or Linux. Pick a local model during setup."},{"title":"Start the meeting.","body":"Open Zoom, Meet or Teams as usual. OpenWhispr captures your computer's audio and transcribes it on your device."},{"title":"Review, paste, move on.","body":"Hit Generate Notes, edit the write-up and paste it where the work lives. With a local model, transcription runs on your device."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons product teams choose OpenWhispr","subtitle":"Choosing a notetaker is a security decision before it's a note-taking one. Here are the facts, and your security team can verify the privacy ones line by line in the repo.","items":[{"title":"The roadmap call never reaches a server","body":"Pick a local model and transcription runs on your machine. The open-source code shows no upload step on that path."},{"title":"No bot in the call","body":"It captures your computer's audio, so nothing appears in the participant list. Your customer never sees a notetaker."},{"title":"Every seat is free, including the quiet ones","body":"Unlimited local meeting notes at $0, for everyone who ever sits in a call. Only the cloud side is metered."},{"title":"Works with any meeting app","body":"Zoom, Meet, Teams, Slack huddles, Webex, a standup in the room. It captures audio, not a specific app."},{"title":"Decisions and owners, not just transcripts","body":"The write-up is structured with decisions, action items and follow-ups. Where an owner is clear in the transcript, it uses that speaker label. You don't write the prompt."},{"title":"No meeting-app integration to approve","body":"No meeting-app plugin or bot account to approve. It's a desktop app that listens to your computer's audio."},{"title":"The recording is deleted after the meeting","body":"Temporary diarization audio is removed after processing. You control the recordings and transcripts saved in the app."},{"title":"Read the code before you pick the tool","body":"OpenWhispr is open source. Your security team can read the code on GitHub."},{"title":"No signal, same write-up","body":"A standup in a room with bad WiFi, a customer site, a plane. Local models need no signal."},{"title":"Who said what, worked out on your machine","body":"Speaker labels are computed on your own device, free on every plan. The write-up says who owns cutting scope, rather than merely that someone said it."}]},"faq":{"label":"Before you download","title":"The questions product teams ask first.","subtitle":"Straight answers on bots, your data and cost. If yours isn't here, ask us in Discord.","items":[{"question":"Does the meeting audio ever leave my device?","answer":"Not in local mode. Pick a local model during setup and transcription runs on your machine with no upload step. Cloud transcription is optional, and we never train on your data."},{"question":"Does a bot join the call?","answer":"No. OpenWhispr captures your computer's audio, so nothing appears in the participant list. Zoom, Meet, Teams, a Slack huddle or a standup in the room all work the same way because it listens instead of joining. With a local model, transcription runs on your device. Tell people you're recording, same as you would with a bot."},{"question":"Will it get our feature names and acronyms right?","answer":"Add them to the custom dictionary, free on every plan. The write-up is told to use those exact spellings, so codenames and feature names come out right in the notes you paste."},{"question":"Do the AI meeting notes capture decisions and owners, not just a transcript?","answer":"Yes. Jot rough notes during the call, then run the write-up when it ends. AI combines them with the transcript into decisions, next steps and action items. Where an owner is clear, it uses that speaker label. You review before anything is shared."},{"question":"Will it work for an in-person standup with no call?","answer":"Yes. It records from your mic, and local models run entirely offline. Capture the standup, then paste the write-up into Linear or Notion when you're done."},{"question":"Is it actually free?","answer":"Yes. Unlimited local meeting notes, 100+ languages, all three desktop platforms: $0 forever, for every seat. Only the cloud side is metered: cloud meeting recordings run 5 hours a month free, 20 on Pro ($8/month), unlimited on Business ($16/user/month)."}]},"reviews":{"label":"What users say","title":"Built by people who don't want a bot in their meetings either.","subtitle":"From our Discord, GitHub and inboxes, the same instinct that matters when the roadmap's on screen.","score":{"value":"0","label":"Roadmap calls on a vendor's server","body":"In local mode, transcription runs on your device. Cloud transcription is optional, and its audio is discarded after transcription."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for product teams","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Product teams who want private meeting notes and are fine with a quick local setup"}]},{"parts":[{"text":"Product teams who want easy, reliable cloud transcription"}]},{"parts":[{"text":"Product orgs in calls all day"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Free local meeting notes, with the audio staying on your device"}]},{"parts":[{"text":"Unlimited cloud transcription without managing local models or keys"}]},{"parts":[{"text":"Unlimited recordings plus advanced workflow features for a team"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for top accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","tooltip":"Cloud meeting recordings. Recording and transcribing meetings with a local model is unlimited on every plan.","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some meetings decide the roadmap. We built for those.","paragraphs":["OpenWhispr started with a simple objection: the meeting where you decide the roadmap shouldn't have to pass through a stranger's server to become a write-up.","OpenWhispr is open source. Your security team can inspect how local transcription works before putting it in roadmap meetings.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Take the decisions into the work.","body":"Download OpenWhispr, pick a local model, and your next meeting gets transcribed, written up and ready to paste on your computer.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseStudents":{"breadcrumb":{"home":"Home","useCases":"Use Cases","students":"For Students"},"metadata":{"title":"Free Speech-to-Text Software For Students - OpenWhispr","description":"OpenWhispr is free, open-source speech-to-text software for students. Record the lecture, get the notes written up on your own laptop, at $0 with no card.","keywords":["speech to text software for students","free dictation software for students","lecture transcription app","record lectures and get notes","speech to text for university students","free speech to text for students","student lecture notes app"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text software for students","operatingSystem":"macOS, Windows, Linux","description":"Local-first, open-source speech-to-text software for students. Lectures are recorded and transcribed on the student's own laptop, written up as structured notes with a local AI model, and essays are dictated the same way. Unlimited local transcription is free, and it works fully offline.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text software for students","title":"Hit record in the lecture. Get the notes written up.","body":"Turn a lecture into readable notes on your laptop. OpenWhispr records the room or the stream, and one tap on Generate Notes turns the transcript into headings and key points for $0 with a local model.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"A person smiling with a laptop while relaxing on a sofa."}},"trustBar":{"label":"Record lectures from:","names":["Zoom","Microsoft Teams","Panopto","Echo360","Google Meet","YouTube"]},"pain":{"label":"The problem with lecture notes","title":"You cannot listen and write at the same time.","subtitle":"A lecture runs at the speed of speech. Writing notes pulls your attention away from following the argument.","cards":[{"title":"The notes you take are the lecture you missed.","body":"Every sentence you write down is a sentence you stop listening to. Let the recording take the words and keep your attention on the whiteboard instead."},{"title":"A two-hour recording takes two hours to replay.","body":"Audio is hard to revisit when you cannot skim it. OpenWhispr hands back text and structured notes you can scan before replaying anything."},{"title":"Pick the right mode for exam month.","body":"Use local transcription for unlimited on-device transcription at $0, or Cloud for top accuracy and speed without downloading a model. Free includes 2,000 words of Cloud transcription a week and 5 hours of meeting recordings a month; Pro adds unlimited Cloud transcription and 20 recording hours."},{"title":"Your lectures, sitting on a company's server.","body":"Most transcription tools upload the audio to process it. In local mode OpenWhispr keeps the recording on your laptop, and the code is public on GitHub so you can check that yourself."}]},"features":{"label":"What you get","title":"The lecture, written up before you get home.","subtitle":"Recording is the easy half. What makes a lecture useful in week nine is text you can search and notes you can read.","items":[{"image":{"label":"A person listening beside a laptop in a tiered lecture hall."},"title":"Records the room or the stream","body":"In the lecture hall it records through the mic. For a Zoom, Teams or Panopto lecture it captures the audio playing on your laptop. Mac, Windows and Linux."},{"image":{"label":"A person sitting beside a laptop in a campus lounge."},"title":"Notes, not a wall of transcript","body":"The transcript goes through a write-up step that returns headings, key points and action items in markdown. With a local AI model that step runs on your laptop too."},{"image":{"label":"A person speaking while working on a laptop at a library desk."},"title":"Dictation for everything after","body":"Press a hotkey anywhere and talk. Clean text lands in Google Docs, Word or Overleaf, so the essay that follows the lecture gets said out loud instead of typed."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Lecture audio sent to us","value":"None","footnote":"in local mode, recording and transcription run on your laptop"},{"label":"Cost per student","value":"$$0","footnote":"unlimited local transcription and write-ups, no card, no expiry"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first recorded lecture"}]},"howItWorks":{"label":"How it works","title":"Three steps, and you leave with the write-up.","subtitle":"The whole setup is one download and one model choice. There is nothing to sign up for.","image":{"label":"A person wearing an earbud beside a laptop and handwritten notes."},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Pick a local model during setup and the whole thing runs offline."},{"title":"Hit record when the lecture starts.","body":"It takes the room through your mic, or the stream through your laptop's audio. Then you just listen."},{"title":"Tap Generate Notes.","body":"The transcript comes back as headings, key points and action items. Search those in revision week instead of scrubbing audio."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons students choose OpenWhispr","subtitle":"On a student budget, cost comes first.","items":[{"title":"Free forever, no card details","body":"Unlimited local transcription and note write-ups at $0, with no trial or semester limit."},{"title":"No bot joins the call","body":"OpenWhispr captures audio directly from your system; it does not join meetings as a participant."},{"title":"Your own notes go in the same file","body":"Type a line during the lecture and the write-up folds it in with what was said."},{"title":"Works with no WiFi","body":"Local models run offline. A basement lecture theatre, the train home, a flight: it still records and still transcribes."},{"title":"Recordings stay on your laptop","body":"In local mode the audio is transcribed on your device, and the open-source code shows the local path never uploads it."},{"title":"Searchable in revision week","body":"Search a term in your write-ups instead of scrubbing audio."},{"title":"Use a custom dictionary in the write-up","body":"Add lecturer names, module codes and technical terms to the custom dictionary to give the write-up the spellings you use."},{"title":"100+ languages","body":"Sit a lecture in your second language and read the notes in it, or dictate your essay in your first."},{"title":"Upgrade for unlimited cloud transcription","body":"The free plan includes 2,000 words a week of OpenWhispr Cloud transcription. Pro is $8/month for unlimited Cloud transcription. Students get 40% off for 12 months; email support@openwhispr.com with proof of enrolment for a code."},{"title":"Open source, under the MIT license","body":"Read the code before you install it. No privacy policy to take on faith."}]},"faq":{"label":"Before you download","title":"The questions students ask first.","subtitle":"Straight answers on cost, permission and what your university will think. If yours is not here, ask us in Discord.","items":[{"question":"Is it actually free, or free for a month?","answer":"Actually free. Local transcription, local note write-ups, 100+ languages and all three desktop platforms cost $0 forever, with no card details. The free plan also includes 2,000 words a week of OpenWhispr Cloud transcription and 5 hours a month of cloud meeting recordings. Pro is $8/month if you want unlimited Cloud transcription. Students get 40% off for 12 months; email support@openwhispr.com with proof of enrolment for a code."},{"question":"Am I allowed to record a lecture?","answer":"That is your university's call, not ours, and it varies by institution and by country. Most have a policy on recording teaching, often tied to accessibility. Check yours, and ask the lecturer. OpenWhispr does not record anything until you press the button."},{"question":"Does it work for online lectures as well as in-person ones?","answer":"Yes. In a lecture hall it records through your mic. For a Zoom, Teams, Google Meet or Panopto lecture it captures the audio playing on your laptop, on Windows, Linux and macOS 14.2 or newer. Already have the file, from your phone or a lecture-capture download? Drop it in and it transcribes that too, with no size limit on a local model."},{"question":"Will it get my subject's terminology right?","answer":"It runs Whisper-class models trained on a wide range of accents and 100+ languages. For the words a general model has never seen, add them to the custom dictionary in Settings: lecturer names, module codes and technical terms. Those spellings carry into your dictation and into the write-up of your notes."},{"question":"Is using the AI write-up cheating?","answer":"Transcription is not: the words are the ones your lecturer said, and reading them back is no different from reading your own notes. The write-up step is different, because a model summarises the transcript for you, so check your university's rules on AI before any of it goes into work you hand in."},{"question":"Where does my recording actually go?","answer":"It depends on the mode you pick during setup. OpenWhispr Cloud is the default: audio is sent only to transcribe it, and is not stored afterwards. Choose a local model instead and the answer is nowhere. It is transcribed on your laptop, and the open-source code shows the local path has no upload step. You can switch in Settings any time."}]},"reviews":{"label":"What users say","title":"Built by people who take notes in bad WiFi too.","subtitle":"From our Discord, GitHub and inboxes.","score":{"value":"0","label":"Lecture recordings kept on our servers","body":"In local mode the recording never leaves your laptop. On cloud, audio is sent only to transcribe it and is not stored after that."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for students","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Students recording lectures on their own laptop, with unlimited local transcription and notes"}]},{"parts":[{"text":"Students who want cloud transcription that just works, with no local models to download"}]},{"parts":[{"text":"Research groups and societies recording at volume"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"No meter and no card on the local side, with the recording staying on your laptop"}]},{"parts":[{"text":"Unlimited cloud speech-to-text without managing local models or keys"}]},{"parts":[{"text":"Unlimited cloud meeting recordings plus speaker labels for group work"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"A lecture you can read shouldn't need a subscription.","paragraphs":["OpenWhispr is for students who want to turn lectures into readable notes without a subscription.","We build in public. The code and releases are on GitHub, where 5,000+ engineers have starred the project. When we say the recording stays on your laptop, that's not a promise in a privacy policy. It's a property of the code, and you do not have to take our word for it.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Turn your 9am into notes with one tap.","body":"Download OpenWhispr, hit record when the lecture starts, and read the notes on the way home. On your laptop, at $0, with no card details.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free forever on Mac, Windows & Linux."}}},"useCaseSpanishVeterinary":{"breadcrumb":{"home":"Home","useCases":"Use Cases","spanishVeterinary":"For Spanish-speaking vets"},"metadata":{"title":"Free Spanish Veterinary Dictation Software - OpenWhispr","description":"OpenWhispr turns Spanish veterinary dictation into text for QVET, Provet Cloud, or any clinical record, with free local models available.","keywords":["spanish veterinary dictation software","veterinary dictation software spanish","spanish vet dictation","veterinary transcription spanish","dictado veterinario en español","software de dictado veterinario"]},"schema":{"name":"OpenWhispr","applicationCategory":"Spanish veterinary dictation software","operatingSystem":"macOS, Windows, Linux","description":"Free, open-source dictation software for veterinary teams who document in Spanish. Select Spanish, dictate, review the transcript, and paste it into the clinical record you already use.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Spanish veterinary dictation software","title":"Dictate in Spanish. Get text ready for your PIMS.","body":"Put the cursor in the historia clínica, press the hotkey, and dictate in Spanish. Review the text before you file it.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ happy users","image":{"label":"Veterinarian working at a clinic computer beside a dog"}},"trustBar":{"label":"Use alongside:","names":["QVET","Provet Cloud","GVET","WinVet","Wirevet","Microsoft Word"]},"pain":{"label":"The problem with Spanish clinical documentation","title":"The consult ends. The historia clínica is still blank.","subtitle":"You still owe the anamnesis, exploración, diagnóstico, tratamiento, informe de alta, and sometimes a prescription. Every line still needs your review.","cards":[{"title":"The Spanish note starts after the patient leaves.","body":"The findings are fresh, but the historia clínica is empty. Dictate the note while you still remember the sequence, then review it before filing."},{"title":"Drug names, doses, and accents need a final check.","body":"Select Spanish before you dictate. Store recurring patient names, drug names, and clinic abbreviations in the custom dictionary, then check every dose and clinical term before filing."},{"title":"Your PIMS does not need a new integration.","body":"OpenWhispr does not connect to QVET, Provet Cloud, GVET, WinVet, Wirevet, or Word. It produces text you can paste into a standard editable field, so you keep the workflow you already use."},{"title":"A prescription still needs a veterinarian.","body":"Dictation can type the details. It cannot examine the animal, choose treatment, validate required fields, or sign the prescription. You stay responsible for reviewing the record in the approved system."}]},"features":{"label":"What you get","title":"Pick Spanish. Keep your PIMS.","subtitle":"Select Spanish, press the hotkey, then review what lands in the field. No migration and no new record system.","items":[{"image":{"label":"Veterinarian examining a cat beside an open laptop"},"title":"Spanish is an actual speech setting","body":"Select Spanish during setup or in settings. The shipped app lists Spanish for local Whisper and Parakeet models as well as supported cloud transcription."},{"image":{"label":"Veterinarian using a laptop beside a cat in a carrier"},"title":"From the microphone to a PIMS-ready transcript","body":"Use the system-wide hotkey in a clinical history, discharge, referral, email, or Word document. Automatic paste is a setting and needs the required OS permission. The clipboard option keeps text available for manual paste."},{"image":{"label":"Man using a laptop in a stable tack room"},"title":"A local option when the connection drops","body":"Download a local model once, then Spanish dictation can run on the computer without internet. Local models are unlimited and included on every plan."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"New PIMS required","value":"None","footnote":"paste into the veterinary record system you already use"},{"label":"Cost to start","value":"$$0","footnote":"unlimited local models are included on the Free plan"},{"label":"Setup takes","value":"5 min","footnote":"from download to the first Spanish dictation"}]},"howItWorks":{"label":"How it works","title":"From spoken Spanish to the right clinical field.","subtitle":"OpenWhispr handles the dictation step. You keep the PIMS, the clinical judgment, and the final review.","image":{"label":"Veterinarian using a laptop while comforting a dog on an exam table"},"steps":[{"title":"Download for free.","body":"Install on Mac, Windows, or Linux. Select Spanish as a spoken language."},{"title":"Put the cursor in the record.","body":"Open the right QVET, Provet Cloud, GVET, WinVet, Wirevet, or Word field, then press the hotkey and speak."},{"title":"Review every clinical term.","body":"Check patient names, drug names, doses, accents, and required fields before you save or sign anything."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons Spanish-speaking vets choose OpenWhispr","subtitle":"Spanish is a speech setting in the app. The hotkey, local models, dictionary, and automatic-paste or clipboard workflow stay the same across your desktop.","items":[{"title":"Spanish appears in the language picker","body":"Spanish is a selectable spoken language in the shipped onboarding and settings flow."},{"title":"Two local model families list Spanish","body":"The app's model registry includes Spanish for local Whisper and multilingual Parakeet transcription."},{"title":"No dedicated PIMS integration required","body":"If a PIMS field accepts pasted text, use automatic paste or the clipboard. No dedicated QVET, Provet Cloud, GVET, WinVet, or Wirevet integration is claimed."},{"title":"Use the sections your team already knows","body":"Dictate anamnesis, exploración, diagnóstico, tratamiento, evolución, or informe de alta without changing the record's structure."},{"title":"Keep recurring terms in one list","body":"The custom dictionary stores the patient names, drug names, procedures, and abbreviations you add. You still verify each result."},{"title":"Local Spanish dictation can work offline","body":"After the local model is downloaded, transcription runs on the computer without an internet connection."},{"title":"Unlimited local models cost $0","body":"The Free plan includes unlimited local models, Spanish support, and the custom dictionary."},{"title":"Mac, Windows, and Linux","body":"Use the same Spanish dictation workflow across all three supported desktop platforms."},{"title":"The cleanup instruction is set for Spanish","body":"The instruction asks for Spanish grammar, spelling, and punctuation while keeping naturally spoken English technical terms. Review the result before filing."},{"title":"It does not pretend to prescribe","body":"OpenWhispr types what you dictate. The veterinarian still examines the animal, chooses treatment, checks the required fields, and signs in the approved system."}]},"faq":{"label":"Before you download","title":"The questions Spanish-speaking vets ask first.","subtitle":"Straight answers on Spanish support, terminology, PIMS workflow, offline use, prescriptions, and cost.","items":[{"question":"Does OpenWhispr actually support Spanish dictation?","answer":"Yes. Spanish is a selectable speech language in the shipped app. The code lists it for local Whisper, local multilingual Parakeet, and supported cloud transcription. This page does not claim equal accuracy for every accent or vocabulary."},{"question":"Will it recognize every veterinary term and drug name?","answer":"No blanket accuracy claim. The closed product sources do not document veterinary-specific or Spanish-veterinary tuning for the included local models. Add recurring terms to the custom dictionary, then review patient names, drugs, doses, abbreviations, and diagnoses before filing."},{"question":"Does it work with QVET, Provet Cloud, GVET, WinVet, and Wirevet?","answer":"No dedicated integration is claimed. If the field accepts pasted text, press the system-wide hotkey and dictate. Automatic paste needs the relevant setting and OS permission; the clipboard option supports manual paste."},{"question":"Can Spanish veterinary dictation work without internet?","answer":"Yes, in local mode after you download a model. Local models run on the computer with no internet required. Cloud transcription still needs a connection."},{"question":"Can OpenWhispr create or submit a veterinary prescription?","answer":"No. It can type the details you dictate into the active field, but it does not prescribe, validate a jurisdiction's required fields, submit to a prescription platform, or sign on your behalf. The veterinarian reviews and completes that workflow."},{"question":"Is it actually free?","answer":"Yes. The Free plan is $0 and includes unlimited local models, Spanish support, the custom dictionary, and Mac, Windows, and Linux. OpenWhispr Cloud on Free is limited to 2,000 words a week; Pro is $8 a month and makes cloud transcription unlimited."}]},"reviews":{"label":"What users say","title":"Real users. Their words, unchanged.","subtitle":"These reviews are not presented as veterinary endorsements. They describe the dictation experience the page is built on.","score":{"value":"ES","label":"Selectable speech language","body":"Spanish appears in the shipped language registry for local and cloud transcription."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for Spanish-speaking veterinary teams","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Vets and techs who want unlimited local Spanish dictation and are fine with a quick model download"}]},{"parts":[{"text":"Vets who want unlimited OpenWhispr Cloud transcription in Spanish"}]},{"parts":[{"text":"Veterinary teams that want unlimited recordings and advanced workflows"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Spanish dictation with local models at $0"}]},{"parts":[{"text":"Unlimited cloud transcription without downloading a local model"}]},{"parts":[{"text":"Unlimited recordings plus Agent mode and chat over your data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" stores terms you add"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" stores terms you add"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" stores terms you add"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"Spanish should be a setting, not a workaround.","paragraphs":["OpenWhispr gives veterinary teams a direct path from spoken Spanish to the clinical field already open on screen. The app handles dictation. The veterinarian keeps the terminology review, clinical judgment, and signature.","We build in public. Every release and every line of the app is on GitHub, where 5,000+ engineers have starred and forked the project. So when we say the dictation can stay on your machine, nobody in your practice has to take our word for it. They can read the code.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Put the cursor in the historia clínica. Dictate in Spanish.","body":"Review the text while the case is still fresh, then file it in the system you already use.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCasePhysicians":{"breadcrumb":{"home":"Home","useCases":"Use Cases","physicians":"For Physicians"},"metadata":{"title":"Free Physician Dictation Software for EHR Notes - OpenWhispr","description":"OpenWhispr is free physician dictation software for HPI, exam and assessment text at your EHR cursor, with on-device transcription.","keywords":["physician dictation software","medical dictation for physicians","EHR dictation software","clinical note dictation","voice recognition for physicians","offline physician dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Physician dictation software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source physician dictation software with optional on-device transcription. Put HPI, exam and assessment text at the active cursor with unlimited local dictation at no cost.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Physician dictation software for clinical notes","title":"Press the hotkey. Fill the chart field you're in.","body":"Dictate the HPI, exam, assessment and plan into the EHR field you already opened. Choose a local model so transcription runs on your computer. OpenWhispr is free and open source.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Physician dictating at a computer in an exam room"}},"trustBar":{"label":"Physicians chart in:","names":["Epic","Oracle Health","athenaOne","eClinicalWorks","MEDITECH","Microsoft Word"]},"pain":{"label":"The problem with charting by keyboard","title":"The visit ended. The chart did not.","subtitle":"HPI, exam, assessment, plan, inbox. Every empty field is another piece of the visit waiting to be typed.","cards":[{"title":"Open charts stack up between patients.","body":"The encounter is over, but the HPI and assessment are still blank. Dictate while the details are fresh."},{"title":"Every field breaks your train of thought.","body":"Click HPI, type. Click exam, type. Click assessment, type again. OpenWhispr leaves the EHR open and puts the transcript at the active cursor."},{"title":"One drug name becomes five corrections.","body":"Generic dictation trips over medication names, eponyms and local abbreviations. Add the terms you use to OpenWhispr's custom dictionary."},{"title":"A five-minute trial needs an IT ticket.","body":"You want to test dictation on one chart, not provision a practice. Local mode starts with a desktop download and no account."}]},"features":{"label":"What you get","title":"Less keyboard. Fewer open charts.","subtitle":"Physician dictation software should fit the note template already on screen. OpenWhispr turns speech into text at your cursor without making you rebuild the workflow around it.","items":[{"image":{"label":"Physician dictating into a desktop microphone at a workstation"},"title":"HPI and assessment at the active cursor","body":"Open the note, place the cursor and press the hotkey. Your dictation is transcribed and pasted into that field, ready for review."},{"image":{"label":"Physician holding a medication bottle beside a computer"},"title":"Your medication names in the dictionary","body":"Add the drugs, procedures, clinician names and abbreviations that keep coming back wrong. OpenWhispr passes those terms to transcription."},{"image":{"label":"Clinician dictating beside a laptop inside a van"},"title":"Keep dictating when ward Wi-Fi drops","body":"Download a local model once. Then transcribe on the workstation in clinic, at home or on rounds without an internet connection."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Audio uploaded in local mode","value":"None","footnote":"in local mode, transcription runs on the physician's device"},{"label":"Cost per physician","value":"$$0","footnote":"unlimited local dictation on every desktop plan"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first dictated chart field"}]},"howItWorks":{"label":"How it works","title":"Finish the chart you already have open.","subtitle":"Keep the EHR workflow you already know. OpenWhispr adds a hotkey and local transcription, with no account or card required to start.","image":{"label":"Physician working on a laptop at night"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Choose a local model during setup."},{"title":"Put the cursor in the chart.","body":"Open the HPI, exam or assessment field, press the hotkey and dictate."},{"title":"Review and sign.","body":"Read the inserted text, correct what needs correcting, then finish the note in the EHR."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons physicians choose OpenWhispr","subtitle":"Try one workstation, one chart field and one local model. These are the ten things to check.","items":[{"title":"No copy-and-paste detour","body":"Dictate inside the open chart field. Review the inserted text there, then continue the note."},{"title":"One hotkey across desktop apps","body":"Use the same dictation control in the EHR, a portal reply, an email or a document."},{"title":"A dictionary for your clinical vocabulary","body":"Add drug names, procedures, eponyms and abbreviations as recognition hints for supported transcription and cleanup paths."},{"title":"Local mode stays on the workstation","body":"Choose a local model and dictation is transcribed on the device. Cloud fallback is a separate opt-in setting."},{"title":"Works fully offline","body":"Once the local model is downloaded, dictation keeps working without internet in clinic, at home or on a ward."},{"title":"Unlimited local dictation at $0","body":"Every desktop plan includes unlimited use of local speech-to-text models."},{"title":"Mac, Windows and Linux","body":"Use it on the desktop operating system your workstation already runs."},{"title":"Open source, inspectable by IT","body":"Your organization can review the code on GitHub instead of taking the local-mode claim on faith."},{"title":"100+ languages","body":"OpenWhispr supports 100+ languages. Availability varies by local model and cloud provider."},{"title":"Local or cloud dictation in Settings","body":"Choose Local or Cloud for dictation during setup, and change that dictation setting later when your workflow changes."}]},"faq":{"label":"Before you download","title":"What physicians ask before changing dictation software.","subtitle":"Straight answers on EHR fields, terminology, patient data, offline use and cost.","items":[{"question":"What is the best physician dictation software for EHR notes?","answer":"If you want a hotkey that puts the transcript at your active cursor, unlimited local dictation at $0, and code your IT team can inspect, try OpenWhispr on one workstation. It does not replace the physician's review or the EHR's signing workflow."},{"question":"Does OpenWhispr integrate with my EHR?","answer":"OpenWhispr is not an EHR integration. It transcribes your speech and pastes the text at the active cursor, so you can use it in editable chart fields and other desktop apps. Test it with your organization's EHR and virtual-desktop setup before a wider rollout."},{"question":"How does it handle medication names and clinical abbreviations?","answer":"OpenWhispr includes a custom dictionary. Add the medication names, procedures, eponyms and abbreviations you use, and those terms are supplied to transcription. You still review the note before signing."},{"question":"Can I dictate patient information?","answer":"Your organization decides which tools are approved for patient information. In local mode, transcription runs on the device without uploading the audio. Confirm your organization's policy before use."},{"question":"Will it work when the hospital network or Wi-Fi does not?","answer":"Yes, after you download a local model. Local transcription runs on Mac, Windows and Linux without internet. Cloud transcription remains a separate mode you can choose in Settings."},{"question":"Is it actually free?","answer":"Yes. Every desktop plan includes unlimited local dictation, a custom dictionary and 100+ languages at $0. Pro is $8 per month and adds unlimited cloud transcription and the iPhone and iPad app."}]},"reviews":{"label":"What users say","title":"What users notice when voice replaces the keyboard.","subtitle":"Verbatim notes from our Discord, GitHub and inboxes.","score":{"value":"0","label":"Per-physician licences for local dictation","body":"Unlimited local dictation is included at $0 on every desktop plan."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Start free with local dictation.","Upgrade for cloud or team features."],"caption":"Compare OpenWhispr plans for physicians","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Physicians who want free offline dictation at the active cursor and can set up a local model"}]},{"parts":[{"text":"Physicians who want unlimited cloud transcription and mobile dictation"}]},{"parts":[{"text":"Practices and teams documenting at volume"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited on-device dictation with no per-physician fee"}]},{"parts":[{"text":"Unlimited cloud dictation without downloading a local model"}]},{"parts":[{"text":"Unlimited recordings and priority support for a practice"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"The path from microphone to chart should be inspectable.","paragraphs":["Physicians should not have to guess what a dictation tool does between the microphone and the chart. OpenWhispr makes that path visible in public code.","Your organization can inspect the local-mode implementation on GitHub instead of relying only on a privacy page.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"OpenWhispr founder and community"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Finish the chart before you open the next one.","body":"Review the inserted text, sign the note and move on.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, ask us for help in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseDoctors":{"breadcrumb":{"home":"Home","useCases":"Use Cases","doctors":"For Doctors"},"metadata":{"title":"Free Doctor Dictation Software, No Middleman - OpenWhispr","description":"Doctor dictation software with no middleman: pick a local model and OpenWhispr types your letters on your own machine in seconds. Free and open source.","keywords":["doctor dictation software","physician dictation software","dictation app for doctors","voice to text for doctors","referral letter dictation","offline dictation for physicians"]},"schema":{"name":"OpenWhispr","applicationCategory":"Physician dictation software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Local-first, open-source dictation software for doctors. With a local model chosen, referral letters, reports and notes are transcribed on the doctor's own device and never leave it. Free for unlimited local dictation, and works fully offline.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation software for doctors & physicians","title":"Speak the referral. Send it before the patient leaves.","body":"Choose a local model at setup and the referral, report or patient letter is typed and drafted on your own computer. No transcription service, no per-seat licence, no one else hearing the audio.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ happy users","image":{"label":"Doctor speaking into a desk microphone at a computer"}},"trustBar":{"label":"Dictate into:","names":["DrChrono","Practice Fusion","Tebra","AdvancedMD","Outlook","Microsoft Word"]},"pain":{"label":"The problem with dictation services","title":"You already dictate. The middleman is the problem.","subtitle":"Doctors had dictation figured out decades ago. What's broken is the channel: a service that bills per line and returns the letter tomorrow, or a licence that bills per doctor forever.","cards":[{"title":"A stranger hears the patient's history.","body":"Send a dictation to a transcription service and someone you have never met hears the patient's name, history and diagnosis. In local mode, OpenWhispr transcribes on your own machine."},{"title":"The letter comes back tomorrow.","body":"Outsourced transcription returns your referral in a day or two, and then it still needs checking. OpenWhispr types it the moment you stop talking, so the referral goes out in the same sitting."},{"title":"A licence fee for every doctor, every year.","body":"Medical dictation suites are sold per seat, per year, and the quote climbs at renewal. OpenWhispr's local dictation is $0 for every doctor in the practice, forever."},{"title":"The code is the confidentiality agreement.","body":"Closed tools hand you a privacy policy and ask for faith. OpenWhispr's code is public on GitHub, so anyone in the practice can verify that in local mode, patient audio never leaves the machine."}]},"features":{"label":"What you get","title":"Instant turnaround. Nobody else in the loop.","subtitle":"Doctor dictation software used to mean a slow service or a per-seat licence. Keep the typed letter and drafted report without sending dictation through a service or buying another seat.","items":[{"image":{"label":"Doctor using a desktop computer in an office"},"title":"Typed on your machine, in seconds","body":"Dictate the referral, report or reply into your mic. The words land on your screen seconds after you stop talking, transcribed locally and never uploaded."},{"image":{"label":"Doctor writing in a notebook beside a computer"},"title":"The referral, already in letter form","body":"Speak the gist and get back a structured referral, report or patient instruction sheet. You read it, fix a word, and send it. The typist never existed."},{"image":{"label":"Doctor typing at a clinical workstation"},"title":"One hotkey. Any window, any EHR.","body":"Press the hotkey and speak into whatever is open: the chart, a reply to a colleague, a report in Word. The text lands at your cursor, punctuated, processed locally."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Audio sent to a transcription service","value":"None","footnote":"in local mode, dictation is transcribed on your device and goes nowhere else"},{"label":"Cost per doctor","value":"$$0","footnote":"unlimited local dictation with no per-doctor licence"},{"label":"Setup takes","value":"5 min","footnote":"install it yourself, no IT ticket, no contract"}]},"howItWorks":{"label":"How it works","title":"Turnaround time: the length of the dictation.","subtitle":"Old dictation goes out to a service and comes back tomorrow. Local mode runs on your machine, with no account or card required to start.","image":{"label":"Doctor opening a laptop at a desk"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Choose a local model during the five-minute setup."},{"title":"Dictate as you always have.","body":"Speak the referral, report or note. It's typed on your device seconds after you stop."},{"title":"Review, send, delete.","body":"Check the draft, paste it into the EHR or an email, and delete the audio. You keep control of all of it."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons doctors choose OpenWhispr","subtitle":"Your dictation carries a patient's history, so check the code and the settings, not just the brochure.","items":[{"title":"No transcription service in the loop","body":"The recording stays on your device instead of going out for overnight transcription."},{"title":"The audio has nowhere to go","body":"Local mode has no upload step. The recording is transcribed on your machine and stays there."},{"title":"Free for every doctor in the practice","body":"Unlimited local dictation at $0, every seat, forever. No per-doctor licence to renew each year."},{"title":"Works fully offline","body":"Home visits, a locum shift on a ward with no signal, a flight to a conference. Once the model is on your machine, dictation never stalls."},{"title":"Types into the chart, the inbox and Word","body":"One hotkey puts the text at your cursor: a referral template, a colleague's inbox, the record itself."},{"title":"Letters and reports, not just transcripts","body":"Run a local AI model, or connect your own key, to turn a dictated summary into a formatted referral, report or patient letter."},{"title":"Your letters are your files","body":"Recordings and drafts sit on your disk. Keep, encrypt or shred them on your schedule, not a vendor's retention policy."},{"title":"Auditable by whoever runs your IT","body":"Your IT team can inspect the public code instead of relying only on a vendor claim."},{"title":"Learns your drug names and eponyms","body":"A custom dictionary picks up the drugs, procedures and abbreviations your specialty uses, so they come out spelled right."},{"title":"100+ languages","body":"Dictate in the language you think in, transcribed locally. Useful when the note is in one language and the referral goes out in another."}]},"faq":{"label":"Before you download","title":"What doctors ask before switching dictation software.","subtitle":"Confidentiality, cost, accuracy, answered straight. Anything else, ask in Discord; a human replies within a day.","items":[{"question":"What's the best dictation software for doctors?","answer":"For doctors who want the letter typed the moment they stop talking, at $0 per doctor, with patient audio staying on the machine in local mode: OpenWhispr. It's open source, works offline, and drafting is built in, with a local AI model like Qwen or Llama or your own AI key. If you currently pay a transcription service or a per-seat licence, this is the thing that replaces it."},{"question":"Who hears my dictation?","answer":"In local mode, no one. It's transcribed on your machine and the open-source code shows there is no upload step. Cloud transcription is the other mode, one global setting you pick at setup or in Settings: audio is sent only to be transcribed and never stored."},{"question":"Can I use it with patient information under HIPAA?","answer":"In local mode, patient audio is transcribed on your device and never reaches a third party, so there is no vendor to cover with a BAA. For cloud transcription of PHI, OpenWhispr signs a BAA on the Business plan and up, and holds a SOC 2 Type 2 report and ISO 27001 certification. Whether your workflow is compliant still depends on how you use it; no tool can grant that on its own."},{"question":"Will it get the drug names right?","answer":"It runs Whisper-class models that handle a wide range of accents, and the custom dictionary learns the drug names, procedures and abbreviations your specialty uses. Add a term once and it comes out spelled right from then on."},{"question":"Does it work on home visits, or on a ward with no signal?","answer":"Yes. Once the local model is on your machine it runs entirely offline. Dictate at the bedside or in the car between visits, and send the letter when you're back in range. Cloud dictation tools stall the moment you lose signal."},{"question":"Is it really free for the whole practice?","answer":"Yes. Unlimited local dictation on Mac, Windows and Linux, in 100+ languages, at $0, for every doctor in the practice. Pro at $8/month adds unlimited cloud transcription and the iPhone app when you want them."}]},"reviews":{"label":"What users say","title":"Nobody here mails their voice to a stranger either.","subtitle":"From our Discord, GitHub and inboxes.","score":{"value":"0","label":"Consults heard by a transcriber","body":"In local mode the audio never leaves your device. No server stores it, and no transcriber ever hears it."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for doctors","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Doctors who want free, private, offline dictation and can spare five minutes for a local setup"}]},{"parts":[{"text":"Doctors who want easy, reliable cloud transcription and mobile dictation"}]},{"parts":[{"text":"Practices documenting at volume, or that need a signed BAA"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Free offline dictation, with patient audio staying on your device"}]},{"parts":[{"text":"Unlimited cloud transcription with no local models to manage"}]},{"parts":[{"text":"Unlimited recordings plus advanced workflow features for a practice"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"A patient's history is not training data.","paragraphs":["OpenWhispr started with a simple objection: turning speech into text shouldn't mean the audio lands on someone else's server, or in someone's training set. A patient's history least of all.","We build in public on GitHub. When we say local-mode dictation stays on your device, your practice can inspect the code instead of relying only on a privacy policy.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Keep dictating. Just stop paying someone to type it.","body":"Your next referral goes from your voice to the recipient without passing through anyone else's hands. No service, no seat licence, no waiting.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseProductivity":{"breadcrumb":{"home":"Home","useCases":"Use Cases","productivity":"For Productivity"},"metadata":{"title":"Free Dictation App For Productivity - OpenWhispr","description":"OpenWhispr is a free, open-source dictation app for productivity. Press one hotkey to put tasks, notes and replies in the app already open.","keywords":["dictation app for productivity","productivity dictation app","voice typing for productivity","speech to text productivity app","dictate tasks and notes","offline dictation app"]},"schema":{"name":"OpenWhispr","applicationCategory":"Productivity dictation app","operatingSystem":"macOS, Windows, Linux, iOS","description":"Local-first, open-source dictation app for productivity. Press a system-wide hotkey and speak tasks, notes and replies into the app already open. With local models selected, transcription and clean-up run on your own computer and work offline. Unlimited local dictation is free on macOS, Windows and Linux.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation app for productivity","title":"Say the task. Get it into the list.","body":"Capture the thought before it disappears, right in the app you already have open; no separate inbox to check later. Use OpenWhispr Cloud for fast, accurate transcription, or local models to keep the audio on your machine.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman speaking while working at a laptop beside a notebook and material samples"}},"trustBar":{"label":"Dictate into the app already open, including:","names":["Todoist","Notion","Obsidian","Gmail","Google Docs","Slack"]},"pain":{"label":"The capture problem","title":"The thought is quick. Typing it down is the slow part.","subtitle":"A task arrives while you are already doing something else. The longer capture takes, the more likely the thought stays in your head.","cards":[{"title":"The task arrives in the middle of another task.","body":"You are in Slack when you remember the invoice, or in a doc when you spot tomorrow's follow-up. Put the cursor in your task list, press the hotkey and say it before the thought disappears."},{"title":"The note is longer than the time you have.","body":"A useful idea becomes three thin bullet points because typing the whole thought feels like work. Talk it into Notion or Obsidian instead. OpenWhispr removes filler and false starts, then pastes the cleaned text at the cursor."},{"title":"Five two-minute replies become an hour of avoidance.","body":"The answer is already in your head. Press the same hotkey in Gmail or Slack, say the reply once and review the text before you send it. No separate recorder to empty later."},{"title":"Your system has names only your system understands.","body":"Project codenames, client names, acronyms and recurring places get mangled by generic dictation. Add them to the custom dictionary once so the transcription and clean-up get the spellings you use."}]},"features":{"label":"One hotkey wherever you type","title":"Keep the system you built. Replace the typing.","subtitle":"One system-wide hotkey works wherever the cursor already is. Choose local models and it keeps working without a connection.","items":[{"image":{"label":"Man speaking into a desk microphone beside a monitor and open notebook"},"title":"Your apps stay exactly where they are","body":"Todoist for tasks, Notion for projects, Obsidian for notes, Gmail for replies. OpenWhispr types into all four. It pastes the finished text into the active field, so there is no new inbox to process and no integration to maintain."},{"image":{"label":"Woman speaking beside a laptop and architectural model"},"title":"A false start does not become a messy task","body":"Say 'Thursday, no, Friday' and the clean-up keeps the correction. It removes filler, fixes punctuation and turns spoken dates into written ones while keeping your wording and intent."},{"image":{"label":"Man working on a laptop and notebook aboard a train"},"title":"The capture habit still works offline","body":"Download local speech-to-text and a local clean-up model once. Then tasks, notes and replies can be transcribed and tidied on your computer with no connection and no per-word bill."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"AI processing location","value":"On your device","footnote":"when local transcription and local clean-up models are selected"},{"label":"Cost to capture locally","value":"$$0","footnote":"unlimited local dictation on every desktop you use"},{"label":"Local setup","value":"A few steps","footnote":"download, choose models, set the hotkey and start speaking"}]},"howItWorks":{"label":"Download to first captured task","title":"Three steps from a thought to the right inbox.","subtitle":"Nothing to migrate and no productivity system to rebuild.","image":{"label":"Woman typing on a laptop backstage beside camera equipment"},"steps":[{"title":"Download it free.","body":"Install OpenWhispr on Mac, Windows or Linux. Choose OpenWhispr Cloud for speed, or download local models for offline use."},{"title":"Put the cursor where the thought belongs.","body":"Open the task field, note, email or message you already use. Press your hotkey and say the whole thought, including the context you would normally skip."},{"title":"Review it and move on.","body":"OpenWhispr cleans the wording and pastes it at the cursor. Check the text, save or send it, and get back to the work you were doing."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Why it earns a place in the stack","title":"10 reasons people use OpenWhispr in the apps they already use","subtitle":"Each one removes friction from capture without asking you to replace the apps that already work.","items":[{"title":"One hotkey across your apps.","body":"Tasks in Todoist, project notes in Notion, linked thoughts in Obsidian, replies in Gmail and Slack. If automatic paste is unavailable, the text stays on the clipboard."},{"title":"No new inbox to empty.","body":"The words land in the tool where they belong. You do not record a voice memo now and remember to move it later."},{"title":"You approve the text before it becomes a task.","body":"The cleaned text lands in the field first. Fix a name, date or label before you save or send it."},{"title":"Dates look like dates.","body":"Spoken dates, times, numbers and currency are converted to standard written forms before the text reaches your task, note or reply."},{"title":"Your project vocabulary follows you.","body":"Add client names, codenames, acronyms and product terms to the custom dictionary. The same list helps across the apps you dictate into."},{"title":"Unlimited local dictation is free.","body":"No trial clock, no word cap and no card on every desktop plan when you run the speech-to-text model on your own machine."},{"title":"Offline means the capture habit does not break.","body":"Local transcription and a local clean-up model keep working on a plane, during an outage or anywhere the connection drops."},{"title":"100+ languages.","body":"Capture the thought in the language you actually think in. The same app and the same hotkey support more than 100 languages."},{"title":"The privacy claim is code you can inspect.","body":"OpenWhispr is open source. You can inspect the local transcription and paste paths yourself."},{"title":"Mac, Windows and Linux.","body":"Use the same free app on the desktop at work and the laptop at home, with a hotkey you choose for each machine."}]},"faq":{"label":"Before you add voice input to your productivity stack","title":"What people ask before they put OpenWhispr in their stack","subtitle":"What it does, what it does not do and what stays on your machine.","items":[{"question":"Does OpenWhispr create and schedule tasks for me?","answer":"OpenWhispr types finished text where your cursor is. It does not organise or schedule tasks. Todoist, Notion and other apps still apply their own date and label rules, and you review the text before saving."},{"question":"Do I need an integration for every productivity app?","answer":"No. The system-wide hotkey pastes text into the active app, so there is no Todoist plugin, Notion connection or Gmail extension to set up. On Mac, OpenWhispr needs accessibility permission to paste automatically. If automatic paste fails, the transcription remains available to paste from the clipboard."},{"question":"Will it clean up a rambling thought?","answer":"Yes, when clean-up is enabled. The clean-up prompt removes filler words, false starts, stutters and accidental repetition. It fixes grammar and punctuation, converts dates and times to written forms and is explicitly told to preserve your wording, formality and intent."},{"question":"Can OpenWhispr process my dictation locally?","answer":"Yes. Choose local transcription and local clean-up so audio and transcript processing stay on your computer. Once text is pasted, the destination app's own storage and privacy rules apply. OpenWhispr Cloud audio is transcribed and discarded, not stored."},{"question":"Is unlimited dictation really free?","answer":"Unlimited local dictation is $0 on every desktop plan and does not expire. The Free plan also includes 2,000 words a week through OpenWhispr Cloud. Pro is $8 per user each month for unlimited cloud transcription, and Business is $16 per user each month for unlimited cloud meeting recordings and agent mode."},{"question":"Will it spell project names and acronyms correctly?","answer":"Add the exact spellings to the custom dictionary. OpenWhispr passes that vocabulary into supported transcription paths and the clean-up step so client names, project codenames and internal acronyms have more context. You can edit the dictionary any time."}]},"reviews":{"label":"What people tell us","title":"Better than typing. Clean after a stumble. Local without WiFi.","subtitle":"Read them for the tool, not for a productivity method.","score":{"value":"0","label":"Productivity apps to replace","body":"OpenWhispr is the voice input across the stack you already use. Your tasks, notes, docs, email and chat stay where they are."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for productivity","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"People who want unlimited local dictation across their personal productivity stack for $0"}]},{"parts":[{"text":"People who want unlimited cloud dictation without managing local models"}]},{"parts":[{"text":"Teams that want agent mode and unlimited cloud meeting recordings alongside dictation"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Capture tasks, notes and replies locally with no word cap, trial clock or card"}]},{"parts":[{"text":"Unlimited OpenWhispr Cloud transcription with no local model to download"}]},{"parts":[{"text":"Unlimited cloud meeting recordings and agent mode for shared workflows"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"A productivity system is only useful if the thought gets into it.","paragraphs":["OpenWhispr turns speech into clean text where the thought belongs: your task list, notes or reply field.","OpenWhispr is open source, so you can inspect the local transcription and paste paths yourself. With local models selected, dictation can stay on your machine.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"$$0 · open source · your apps stay put","title":"Press the hotkey. Get the thought into the app already open.","body":"Download OpenWhispr, choose a model and dictate into your task list, notes or reply field. No card and no migration.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Unlimited local dictation is free on Mac, Windows and Linux."}}},"useCasePathology":{"breadcrumb":{"home":"Home","useCases":"Use Cases","pathology":"For Pathologists"},"metadata":{"title":"Pathology Dictation Software for Reports - OpenWhispr","description":"OpenWhispr is free pathology dictation software for macroscopic descriptions and microscopic findings, with auto-paste in supported report fields.","keywords":["pathology dictation software","pathologist dictation software","pathology report dictation","anatomic pathology dictation","histopathology dictation software","pathology speech recognition","offline pathology dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Pathology dictation software","operatingSystem":"macOS, Windows, Linux","description":"Local-first, open-source pathology dictation software that transcribes spoken report text for automatic paste into supported fields. Free for unlimited local dictation.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Pathology dictation software","title":"Dictate the findings. Get report text at your cursor.","body":"Keep the specimen or slide in view. Dictate the macroscopic description, microscopic findings or final diagnosis. OpenWhispr types into supported fields. You review and sign out.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Laboratory professional reviewing a computer beside a microscope and slide tray"}},"trustBar":{"label":"Reporting environments to test with OpenWhispr:","names":["Epic Beaker","Oracle Health PathNet","Clinisys PowerPath","Clinisys WinPath","Sectra Digital Pathology","Microsoft Word"]},"pain":{"label":"The problem with typing pathology reports","title":"The slide is in focus. The keyboard pulls you away.","subtitle":"The finding is clear. The field is empty. Every trip to the keyboard interrupts the read.","cards":[{"title":"Your eyes leave the slide for the keyboard.","body":"You see the finding, then look away to type it. OpenWhispr lets you speak the report text while the morphology is still in view."},{"title":"The gross description waits on your hands.","body":"Measurements, margins, sections and the block key arrive in sequence at the cut-up bench. Dictate each observation while it is in front of you, then review the text before it enters the final report."},{"title":"Structured fields still need precise prose.","body":"Synoptic fields capture element-response pairs. Free-text microscopic descriptions, comments and addenda can still need your words. OpenWhispr transcribes those words without building or validating the protocol."},{"title":"One unfamiliar term becomes a correction loop.","body":"Specimen labels, eponyms and local abbreviations recur case after case. Add them to the custom dictionary as context hints, then check every occurrence before sign-out."}]},"features":{"label":"What you get","title":"Keep the case in view. Put the words in the report.","subtitle":"Pathology dictation software should fit the reporting screen already open. OpenWhispr turns speech into text at the active cursor. It does not replace your LIS, protocols or judgment.","items":[{"image":{"label":"Person typing beside magnified tissue images on two monitors"},"title":"Report text at the active cursor","body":"Open an editable report field, place the cursor and press the hotkey. OpenWhispr pastes the transcript when the field and session support automatic paste. Otherwise it copies the text for manual paste."},{"image":{"label":"Laboratory professional using a desktop microphone at a workstation"},"title":"Local transcription when you choose it","body":"Download a local speech model and select Local in Settings. With cloud fallback disabled, dictation stays on the workstation and continues without internet after the model is installed."},{"image":{"label":"Laboratory professional holding a specimen tube and smartphone beside a microscope"},"title":"Your recurring terms as context hints","body":"Add specimen names, eponyms, stains and local abbreviations to the custom dictionary. OpenWhispr feeds them to supported speech models as hints. It does not promise specialist terminology accuracy."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Direct LIS integrations included","value":"None","footnote":"OpenWhispr uses active-cursor paste in compatible fields instead"},{"label":"Cost per pathologist","value":"$$0","footnote":"unlimited local dictation on every desktop plan"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first dictated report text"}]},"howItWorks":{"label":"How it works","title":"Findings in the report before you leave the case.","subtitle":"Speech becomes editable text in the report field. You check it against the case and complete sign-out in the reporting system.","image":{"label":"Healthcare worker in blue scrubs opening a laptop in a hospital corridor"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Pick a local model during setup."},{"title":"Dictate the findings.","body":"Put the cursor in an editable field, press the hotkey and speak the macroscopic description, microscopic findings, ancillary-study results, addendum or diagnosis text."},{"title":"Review, then sign out.","body":"Check the text against the specimen, slides, ancillary studies and required report elements. The final report remains your call."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons OpenWhispr fits pathology reporting","subtitle":"A pathology report carries exact observations and a final professional judgment. OpenWhispr stays in the narrower lane of turning your speech into editable text.","items":[{"title":"Automatic paste in supported fields","body":"Use the same hotkey in a compatible report field, an email or a document. If automatic paste is unavailable, OpenWhispr leaves the text on your clipboard."},{"title":"No direct LIS plug-in required","body":"Compatible editable fields use desktop dictation and automatic paste. OpenWhispr does not claim a direct integration with Epic Beaker, PathNet, PowerPath, WinPath or another LIS. Test your reporting system and virtual-desktop setup."},{"title":"A local transcription path","body":"Choose Local in Settings and keep cloud fallback disabled. The speech model runs on the workstation and the dictation audio is not uploaded for transcription on that path."},{"title":"Works offline after model setup","body":"Once a local model is downloaded, dictation continues without internet at the microscope, gross bench or home workstation."},{"title":"A dictionary for recurring terms","body":"Add specimen names, stains, eponyms and abbreviations as context hints. The pathologist still checks the transcript before sign-out."},{"title":"Unlimited local dictation at $0","body":"Every desktop plan includes unlimited use of local speech-to-text models."},{"title":"Mac, Windows and Linux","body":"Use OpenWhispr on the desktop operating system your reporting workstation already runs."},{"title":"Open source, inspectable by IT","body":"The desktop app code is public on GitHub, so your technical team can inspect the local and cloud transcription paths."},{"title":"100+ languages","body":"OpenWhispr supports 100+ languages. Availability varies by local model and cloud provider."},{"title":"Local or cloud in Settings","body":"Choose the transcription path during setup and change it later in Settings. Cloud audio is sent only to transcribe it and is never stored by OpenWhispr."}]},"faq":{"label":"Before you download","title":"What pathologists ask before changing dictation software.","subtitle":"Straight answers on reporting fields, terminology, diagnostic boundaries, local processing and cost.","items":[{"question":"Does OpenWhispr integrate with my LIS or digital pathology viewer?","answer":"OpenWhispr is not an LIS or viewer integration. It transcribes speech and pastes text at the active cursor, so you can use it in compatible editable fields and other desktop apps. Test it with your reporting system and virtual-desktop setup before wider use."},{"question":"Does it generate a diagnosis or complete a synoptic report?","answer":"No. OpenWhispr turns what you dictate into editable text. It does not inspect slides, interpret findings, choose a diagnosis, populate required protocol elements or sign out a case. The pathologist reviews every word in the reporting system."},{"question":"Will it recognize pathology terminology?","answer":"OpenWhispr includes a custom dictionary. Add recurring specimen names, stains, eponyms and abbreviations, and supported speech models receive them as context hints. This is not a guarantee of specialist terminology accuracy, so review every report before sign-out."},{"question":"Where is the dictation processed?","answer":"Choose Local and transcription runs on the workstation. Choose OpenWhispr Cloud and the audio is sent only for transcription, then never stored by OpenWhispr. Your laboratory decides which mode and workflow are approved for its data."},{"question":"Will it work without internet?","answer":"Yes, after you download a local speech model. Local transcription runs on Mac, Windows and Linux without internet. Cloud transcription remains a separate mode you can choose in Settings."},{"question":"Is it actually free?","answer":"Yes. Every desktop plan includes unlimited local dictation and a custom dictionary at $0. OpenWhispr supports 100+ languages across available models and providers; local-model availability varies. Pro is $8 per month and adds unlimited cloud transcription, more meeting recording hours, mobile and device sync."}]},"reviews":{"label":"What users say","title":"The dictation is fast. The judgment stays yours.","subtitle":"Existing OpenWhispr user feedback. These testimonials are about the tool, not pathology reporting.","score":{"value":"0","label":"Diagnostic decisions made by OpenWhispr","body":"OpenWhispr transcribes what you say. It does not interpret a specimen or slide, choose a diagnosis or sign out a report."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for pathologists","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Pathologists who want free offline dictation at the active report cursor"}]},{"parts":[{"text":"Pathologists who want unlimited OpenWhispr Cloud transcription"}]},{"parts":[{"text":"Laboratories and teams documenting at volume"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local report dictation at $0"}]},{"parts":[{"text":"Unlimited cloud transcription without managing local models"}]},{"parts":[{"text":"Unlimited recordings and priority support for a team"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users and OpenWhispr features"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" supplies names and terms as context hints"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" supplies names and terms as context hints"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" supplies names and terms as context hints"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" runs user-configured actions on transcripts"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic OpenWhispr team use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, SAML, SCIM, audit logs, retention controls, admin controls, or dedicated support for a larger team? These are enterprise security and data-governance options, not clinical or regulatory certification.","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"The report remains yours. OpenWhispr handles the typing.","paragraphs":["OpenWhispr handles the mechanical part: speech becomes editable text at your cursor. The interpretation, structured elements, final diagnosis and sign-out remain with the pathologist and the laboratory's reporting system.","OpenWhispr is open source under the MIT License, and its source is available on GitHub. When we say dictation can run locally, your laboratory can inspect the code.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"OpenWhispr founder portrait"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Finish the report while the case is still on screen.","body":"Put the cursor in the editable field, dictate the findings and review every word before sign-out.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work with your workstation, ask us for help in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseRadiology":{"breadcrumb":{"home":"Home","useCases":"Use Cases","radiology":"For Radiology"},"metadata":{"title":"Best Radiology Dictation Software, Free & Local - OpenWhispr","description":"OpenWhispr is free, open-source radiology dictation software that transcribes locally and pastes text at the active cursor for review.","keywords":["best radiology dictation software","radiology dictation software","radiologist speech recognition software","offline radiology dictation","local radiology transcription","radiology voice to text"]},"schema":{"name":"OpenWhispr","applicationCategory":"Radiology dictation software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Local-first, open-source dictation software for radiologists. Choose local speech-to-text and the audio is transcribed on the workstation, then text is pasted at the active cursor. Cleanup is configured separately; there is no native radiology-system integration.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Best radiology dictation software for imaging specialists","title":"Keep the images in view. Get dictated text at the cursor.","body":"Choose a local speech-to-text model, place the cursor in an editable report field, and speak. Audio is transcribed on the workstation; use local cleanup or turn cleanup off to keep the text there too.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Radiologist dictating beside chest scans at a reading station"}},"trustBar":{"label":"Common reporting tools, not integrations:","names":["PowerScribe One","Epic Radiant","Sectra IDS7","Visage 7","InteleViewer","Microsoft Word"]},"pain":{"label":"Where radiology dictation breaks","title":"The images need your eyes. The report still needs your words.","subtitle":"A radiology dictation tool has to respect the active field, the vocabulary and the final read. Miss one, and the keyboard comes back.","cards":[{"title":"Your eyes leave the study for the keyboard.","body":"Place the cursor once, then hold the hotkey and dictate the finding. OpenWhispr pastes the transcript into the active field so you can stay with the images."},{"title":"The same term comes back wrong every case.","body":"Add names, modalities, anatomy and local abbreviations to the custom dictionary. They are supplied as hints to supported transcription paths."},{"title":"One missing 'no' changes the report.","body":"Speech recognition can change clinical meaning. OpenWhispr supplies text, not clinical quality control, so the radiologist still proofreads before signing."},{"title":"A simple dictation test turns into an integration project.","body":"OpenWhispr is a desktop speech-to-text layer, not a PACS or RIS replacement. Try it in an editable field without changing the reporting stack."}]},"features":{"label":"What you get","title":"Cursor dictation, not another reporting platform.","subtitle":"OpenWhispr handles speech to text. Your worklist, images, templates, measurements and signing workflow stay in the systems that already own them.","items":[{"image":{"label":"Person working at two monitors beside a desktop computer"},"title":"Findings transcribed on the workstation","body":"Choose a local speech-to-text model and audio is processed on the device. Use local cleanup or turn cleanup off if report text must stay there too; after model download, speech-to-text works offline."},{"image":{"label":"Radiologist checking a reference card beside a medical scan"},"title":"A dictionary for the terms you actually dictate","body":"Add recurring anatomy, eponyms, protocol names and abbreviations. OpenWhispr supplies the list as recognition hints, and you still check the result."},{"image":{"label":"Radiologist typing beside monitors displaying medical scans"},"title":"One hotkey at the active report cursor","body":"Open the report field, place the cursor, hold the hotkey and speak. The transcript is pasted there for review, with no claim of a native PACS, RIS or EHR integration."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Dictated audio uploaded in local mode","value":"None","footnote":"with local-to-cloud fallback off, audio stays on the radiologist's workstation"},{"label":"Cost per radiologist","value":"$$0","footnote":"unlimited local dictation for every radiologist"},{"label":"Setup takes","value":"5 min","footnote":"from download to a test dictation at the cursor"}]},"howItWorks":{"label":"How it works","title":"From spoken finding to text in the open report.","subtitle":"Keep the reporting workflow you already know. OpenWhispr adds a system-wide hotkey and a local transcription option.","image":{"label":"Radiologist working at a laptop beside brain scans"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Choose a local model during setup."},{"title":"Put the cursor in the report.","body":"Open an editable findings or impression field, then press the hotkey and dictate."},{"title":"Proofread and sign.","body":"Check terminology, numbers, negation and laterality, then finish the report in your reporting system."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons radiologists choose OpenWhispr","subtitle":"Use it for the narrow job it ships to do: turn speech into text at the active cursor. Keep the clinical and reporting workflow under your control.","items":[{"title":"Local transcription at the reading station","body":"Choose a local model and, with local-to-cloud fallback off, audio is transcribed on the device without an upload step."},{"title":"Text lands at the active cursor","body":"The desktop app pastes the transcript into the focused editable field. If automatic paste is unavailable, it can leave the text on the clipboard."},{"title":"No reporting-stack replacement","body":"Keep the worklist, viewer, report template and signing system your practice already uses."},{"title":"A dictionary for radiology vocabulary","body":"Add anatomy, eponyms, protocol names and abbreviations as hints for supported transcription and cleanup paths."},{"title":"Keeps working off the hospital network","body":"Once a local model is downloaded, dictation does not need the reading-room network or internet."},{"title":"$$0 local dictation per radiologist","body":"Every desktop plan includes unlimited use of local speech-to-text models."},{"title":"Three desktop operating systems","body":"Run the same desktop dictation app on the operating system your workstation uses."},{"title":"Your IT team can inspect the paste path","body":"Your organization can inspect the transcription and paste paths on GitHub."},{"title":"You read the images. OpenWhispr handles speech to text.","body":"It is not a diagnostic image reader and has no native patient, study or radiology-system context. Agent Mode can use optional screen context, but it does not diagnose or replace review."},{"title":"100+ spoken languages","body":"OpenWhispr supports 100+ languages. Availability and accuracy vary by local model and cloud provider."}]},"faq":{"label":"Before you download","title":"What radiologists ask before testing dictation software.","subtitle":"Straight answers on report fields, terminology, integrations, patient data, offline use and cost.","items":[{"question":"What is the best radiology dictation software?","answer":"It depends on the job. If you need a full reporting platform with patient and study context, templates, structured data, quality checks and enterprise interfaces, use a radiology reporting system built for that. If you want free, local-first speech-to-text that pastes at an active cursor, test OpenWhispr in your actual report field."},{"question":"Does OpenWhispr integrate with PACS, RIS or EHR software?","answer":"No native integration. OpenWhispr does not receive patient or study context from PACS, RIS, EHR, DICOM, HL7 or FHIR. It transcribes speech and pastes text at the active cursor. Optional Agent Mode can use screen context, which is not diagnostic integration. Test your reporting software and virtual desktop."},{"question":"Can it navigate templates or build a structured radiology report?","answer":"OpenWhispr ships no built-in radiology structured-reporting integration, report macros, voice-command field navigation or template logic. Generic Agent Mode can reformat text from your prompt, but it does not know the report fields. Use your existing system template and cursor."},{"question":"How does it handle terminology, numbers and laterality?","answer":"The custom dictionary lets you add recurring anatomy, eponyms, protocol names and abbreviations as recognition hints. It is not a radiology lexicon or clinical error checker. Proofread every report for terminology, measurements, negation and laterality before signing."},{"question":"Can I use it offline with patient information?","answer":"After a local model is downloaded, local speech-to-text runs offline and does not upload audio when the separate local-to-cloud fallback is off. Cleanup is configured separately. Your organization still decides which devices and applications are approved for patient information; cloud transcription is a separate global mode."},{"question":"Is local radiology dictation really free?","answer":"Yes. Every desktop plan includes unlimited local dictation, a custom dictionary and 100+ languages at $0. Pro is $8 per month and adds unlimited OpenWhispr Cloud transcription, more meeting hours, mobile and device sync."}]},"reviews":{"label":"What users say","title":"What users say about local dictation.","subtitle":"These quotes are verbatim. None is rewritten as a radiologist endorsement.","score":{"value":"1","label":"Job in the reporting stack","body":"OpenWhispr supplies speech-to-text at the active cursor. Your reporting system keeps the patient, study, template and signature workflow."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Start free with local dictation.","Upgrade for cloud or team features."],"caption":"Compare OpenWhispr plans for radiologists","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Radiologists who want unlimited local cursor dictation and can run a model on the workstation"}]},{"parts":[{"text":"Radiologists who want unlimited OpenWhispr Cloud transcription"}]},{"parts":[{"text":"Teams that want OpenWhispr Cloud, recordings and priority support"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local speech-to-text at $0, with audio staying on the workstation"}]},{"parts":[{"text":"Unlimited cloud transcription without managing local models or keys"}]},{"parts":[{"text":"Unlimited recordings and priority support for a team"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"We built the speech-to-text layer, not a radiology stack.","paragraphs":["Radiologists already work across a worklist, image viewer and reporting system. OpenWhispr does one narrower job: turn speech into text and paste it at the active cursor.","We build in public. Your reading-room or IT team can inspect the local transcription and paste paths on GitHub instead of relying only on a product page.","And if a report field will not accept the paste, you talk to the people who built it, in Discord."],"image":{"label":"OpenWhispr founder and community"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Test one dictated finding at the cursor.","body":"Start in a blank text field, then confirm cursor paste in an approved report field before clinical use.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, ask us for help in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseHipaaCompliant":{"breadcrumb":{"home":"Home","useCases":"Use Cases","hipaaCompliant":"For HIPAA-Regulated Teams"},"metadata":{"title":"HIPAA Compliant Dictation Software, Local + BAA - OpenWhispr","description":"OpenWhispr is HIPAA compliant dictation software with on-device transcription and a Business-plan BAA for cloud use.","keywords":["hipaa compliant dictation software","hipaa dictation software","hipaa compliant voice to text","secure medical dictation software","private healthcare dictation","baa dictation software","on-device medical transcription"]},"schema":{"name":"OpenWhispr","applicationCategory":"Healthcare dictation software","operatingSystem":"macOS, Windows, Linux, iOS","description":"HIPAA compliant, local-first, open-source dictation software for healthcare teams. Local transcription runs on the user's device. A separate BAA is available on request from the Business plan for cloud Services.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"HIPAA compliant dictation software for healthcare teams","title":"Pick a local model. Get the transcript. Send us no audio.","body":"Patient audio is transcribed on the computer in front of you. The DPA requires a signed BAA before cloud PHI; confirm it covers the exact feature path.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Professional holding a notebook and pointing toward a desktop computer"}},"trustBar":{"label":"Dictate into any text field, including:","names":["Epic","Oracle Health","athenahealth","eClinicalWorks","NextGen","Microsoft Word"]},"pain":{"label":"Choosing HIPAA compliant dictation software","title":"The compliance question starts before anyone presses record.","subtitle":"Ask where the audio goes, what text is saved, who processes it and whether a BAA covers that exact path.","cards":[{"title":"A cloud recording starts before the BAA is signed.","body":"OpenWhispr's DPA says PHI must not go through the cloud Services until a separate BAA is in place. It is available on request from Business, but confirm its scope first."},{"title":"One product, three different data paths.","body":"Local models process audio on your device. OpenWhispr Cloud sends audio to our cloud. Bring your own key sends it to the provider you chose. The setting changes the answer."},{"title":"Audio deleted. Transcript still synced.","body":"Cloud audio is processed in real time and not persisted by OpenWhispr. Transcribed text and notes can still enter cloud sync when you enable it, so map both flows."},{"title":"A HIPAA badge with no configuration.","body":"HIPAA has no software certificate. The useful questions are concrete: which path handles PHI, which agreement covers it and which safeguards your team controls."}]},"features":{"label":"Three choices to make","title":"Three paths. Three different approval questions.","subtitle":"The product gives you local, OpenWhispr Cloud and bring-your-own-key paths. Your workflow decides which one handles PHI.","items":[{"image":{"label":"Clinician dictating into a phone beside a clinical workstation"},"title":"Keep transcription on-device","body":"Pick a local model and patient audio is transcribed on your computer. If you use AI cleanup or drafting, select a local model for that step too."},{"image":{"label":"Professional reviewing a document beside a laptop"},"title":"Put the BAA before the cloud","body":"For PHI in OpenWhispr Cloud, execute the separate BAA and confirm it covers the feature and subprocessors in your path. It is available on request from Business."},{"image":{"label":"Two professionals reviewing a laptop in an office"},"title":"Map the text after transcription","body":"Cloud audio is not persisted after transcription. Notes and transcribed text may still be stored or synced, depending on the settings and tools you choose."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Patient audio sent to us","value":"None","footnote":"when a local transcription model is selected"},{"label":"Cost for local dictation","value":"$$0","footnote":"unlimited on every plan, including Free"},{"label":"Setup takes","value":"5 min","footnote":"then review the data path before using PHI"}]},"howItWorks":{"label":"How it works","title":"Approve the path before the first dictated note.","subtitle":"A HIPAA workflow is a configuration, a contract and your own safeguards. It is not a badge the download can grant.","image":{"label":"Professional reviewing paperwork at a standing desk"},"steps":[{"title":"Download and choose the path.","body":"Pick local models, OpenWhispr Cloud or your own provider. Write down which one your team approved for PHI."},{"title":"Dictate only after the contract check.","body":"Local transcription sends no audio to OpenWhispr. Keep cloud PHI off until the separate BAA is signed and its feature scope is confirmed."},{"title":"Review, file and retain deliberately.","body":"Check the note, put it in the approved record system and set local retention and optional sync to match your policy."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"What your review can verify","title":"10 reasons HIPAA-regulated teams evaluate OpenWhispr","subtitle":"Each line names a specific product or contract fact. None replaces your risk analysis, policies or workforce safeguards.","items":[{"title":"Local transcription sends us no audio","body":"Pick a local speech-to-text model and the recording is processed on your own computer."},{"title":"OpenWhispr does not persist cloud audio","body":"The DPA says cloud voice audio is processed in real time and not persisted by OpenWhispr after transcription."},{"title":"A BAA is available on request","body":"Business-plan and Enterprise customers can request the separate BAA required before PHI enters the cloud Services."},{"title":"Unlimited local dictation is free","body":"Local speech-to-text is unlimited on every plan, including Free, on macOS, Windows and Linux."},{"title":"Local mode works offline","body":"Once the local model is downloaded, transcription needs no internet connection."},{"title":"The app is open source","body":"Security teams can inspect the routing code instead of relying on a marketing description."},{"title":"Customer data is not used for AI training","body":"The DPA prohibits OpenWhispr and its subprocessors from using customer personal data to train or improve models."},{"title":"Transcription and cleanup are separate routes","body":"Review the speech-to-text model and the AI cleanup model independently before PHI use."},{"title":"Bring your own key creates a third path","body":"In BYOK mode, content goes directly from the device to the model provider you chose."},{"title":"Sync is a separate retention decision","body":"Cloud backup defaults off. Turn it on only after deciding where transcribed text and notes may be stored."}]},"faq":{"label":"Before PHI enters the workflow","title":"Six questions for a HIPAA dictation review.","subtitle":"These answers describe the product and its shipped agreements. Your organization still owns the final risk analysis.","items":[{"question":"Is OpenWhispr HIPAA compliant?","answer":"Yes. OpenWhispr is HIPAA compliant. Local transcription sends us no patient audio. For PHI in the cloud Services, its DPA requires a separate signed BAA, available on request from the Business plan. Your organization still needs to configure its chosen path and safeguards correctly."},{"question":"Do I need a BAA for local transcription?","answer":"Generally not for that transcription step. With a local model, audio is processed on the device and OpenWhispr does not receive it. You still need safeguards for the computer, the resulting note, any sync you enable and the system where you file it. Confirm the full analysis with your compliance counsel."},{"question":"Can I use OpenWhispr Cloud with PHI?","answer":"Only after a separate BAA is executed and you confirm it covers the feature path you plan to use. The shipped DPA says end users must not process PHI through the cloud Services without one. Business-plan and Enterprise customers can request it at support@openwhispr.com."},{"question":"What does OpenWhispr store?","answer":"OpenWhispr does not persist cloud voice audio after transcription. Local retention settings can keep dictation audio and transcripts on the device, and optional cloud sync can store transcribed text and notes. Review each setting against your retention policy."},{"question":"Does local transcription keep AI cleanup local too?","answer":"Not by itself. Transcription and AI cleanup use separate models. The local setup can configure both, but verify both routes in Settings before PHI use. A bring-your-own-key cleanup model sends the text to that provider."},{"question":"Is the Free plan enough for a HIPAA workflow?","answer":"Free includes unlimited local dictation on macOS, Windows and Linux. It does not include an OpenWhispr BAA for cloud PHI. If your approved path needs OpenWhispr Cloud, request the BAA on Business or Enterprise before using PHI."}]},"reviews":{"label":"What users say about the tool","title":"Product reviews, quoted word for word.","subtitle":"These are verbatim product reviews. None is presented as evidence for your HIPAA analysis.","score":{"value":"3","label":"Verbatim reviews","body":"The wording and attribution are unchanged from the real source reviews."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for HIPAA-regulated healthcare teams","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Healthcare teams using a local-only transcription path for PHI"}]},{"parts":[{"text":"Individuals using cloud transcription for work that does not contain PHI"}]},{"parts":[{"text":"Teams evaluating OpenWhispr Cloud for PHI under a separate signed BAA"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local dictation, with patient audio processed on your device"}]},{"parts":[{"text":"Unlimited cloud transcription for approved non-PHI workflows"}]},{"parts":[{"text":"A BAA on request, unlimited recordings and advanced team workflows"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"A HIPAA checkbox cannot tell you where the audio went.","paragraphs":["OpenWhispr started with a concrete choice: let people transcribe on their own machine when a cloud data path is the wrong fit.","We build in public. Your security team can inspect the local and cloud routing in the app, then compare it with the DPA and the settings your workforce will actually use.","And if your review finds a question the documents do not answer, you can ask the people who built it in Discord or request the BAA through support."],"image":{"label":"OpenWhispr founder discussing the product's data paths"}},"finalCta":{"tagline":"Local path · cloud path · BAA on Business","title":"Choose the data path before you dictate the patient name.","body":"Start with free local transcription, or request the Business-plan BAA and confirm its scope before PHI enters OpenWhispr Cloud. Then document the settings your team approved.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes about 5 minutes to set up. Free local dictation on Mac, Windows & Linux. Your organization remains responsible for its HIPAA workflow."}}},"useCaseAbaSessionNotes":{"breadcrumb":{"home":"Home","useCases":"Use Cases","abaSessionNotes":"For ABA Therapists"},"metadata":{"title":"Speech-to-Text Tools for ABA Session Notes | OpenWhispr","description":"Dictate ABA session-note details with OpenWhispr, then review and paste the text into your existing record system. Choose Cloud for speed or local processing.","keywords":["speech-to-text for aba session notes","aba session notes software","rbt session note dictation","bcba session notes software","aba dictation software","aba progress notes dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text software","operatingSystem":"macOS, Windows, Linux","description":"Open-source speech-to-text software for dictating ABA session-note details. Review and paste the text into your existing record system; choose Cloud or local processing.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation software for ABA therapists & RBTs","title":"Dictate ABA session notes while the details are fresh.","body":"Dictate ABC observations, interventions, client response and goal progress while the session is still fresh. Review the text, then paste it into your existing record system. Choose Cloud for speed or local processing for sensitive and offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Therapist and a child at a low table with wooden building blocks and an open laptop"}},"trustBar":{"label":"Use dictated text in:","names":["your ABA record system","session-note fields","secure notes","email","any text field"]},"pain":{"label":"What slows a session note","title":"The session is still fresh. Typing the note is the delay.","subtitle":"You need clean dictation between clients, a practical mode for sensitive behavioral data, and a plan that fits your caseload.","cards":[{"title":"A stranger transcribing behavioral data.","body":"Traditional dictation services send the recording to a third party, so someone you've never met hears details about a child's behavior. Choose local processing and OpenWhispr transcribes on your own device instead, so no one else is in the loop."},{"title":"Sensitive client data needs a deliberate path.","body":"Choose local transcription when the session covers sensitive behavioral or health details, or you're between locations without signal. Your device processes the recording, so documentation continues wherever the session happens."},{"title":"Typing is minutes you don't have between clients.","body":"Typing an ABC note eats into your break, or pushes you late into the next session. OpenWhispr drafts it from your dictation while it's fresh, so you review and paste instead of type."},{"title":"The quickest route should still be dependable.","body":"When speed and consistent performance matter most, Cloud avoids a local model download and transcribes with audio discarded immediately after transcription."}]},"features":{"label":"What you get","title":"Faster notes. Choose the mode that fits the session.","subtitle":"Get the note out while the session is fresh. Use Cloud for fast, consistent transcription or local processing for sensitive client data.","items":[{"image":{"label":"Therapist and a child high-fiving across a table with a laptop and picture cards"},"title":"Cloud transcription without a local model download","body":"Dictate an ABC note, progress update or session summary into your mic. Cloud gives you top accuracy and speed with consistent performance across hardware, then discards the transcription audio immediately."},{"image":{"label":"Therapist speaking while working at a laptop in a private office"},"title":"Local processing for sensitive client data","body":"Choose a local model when a session touches details about a client's behavior or health you'd rather keep off a server, or you're documenting offline between locations. You review the drafted note before it goes into the client's file."},{"image":{"label":"Therapist pointing at picture cards beside a child, one hand resting on a laptop"},"title":"Dictation wherever your cursor is","body":"Press a hotkey and dictate straight into CentralReach, Rethink or your EHR. Clean, punctuated text lands where you are already working."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Client data sent to us","value":"None","footnote":"in local mode, transcription never leaves your device"},{"label":"Cost per RBT or BCBA","value":"$$0","footnote":"unlimited local dictation, the whole caseload, forever"},{"label":"Free Cloud allowance","value":"2,000 words","footnote":"per week, alongside unlimited local dictation"}]},"howItWorks":{"label":"How it works","title":"A session note drafted before the next client walks in.","subtitle":"Start free, choose Cloud for the quickest high-performance setup or local for sensitive client data, then get the note drafted before the next session.","image":{"label":"Therapist speaking at a laptop in a therapy room lined with toy shelves"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Choose Cloud for no local model download or local for sensitive client data."},{"title":"Dictate the note.","body":"Speak the ABC data, goal progress and session summary, and it's transcribed in whichever mode you chose: on your device, or via Cloud."},{"title":"Review, paste, delete.","body":"Edit the draft, paste it into CentralReach or your EHR, and delete the audio. You control all of it."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons ABA therapists choose OpenWhispr","subtitle":"Choosing a dictation tool is a confidentiality decision before it's anything else. Here are the facts, and your compliance team can verify every one in the repo.","items":[{"title":"In local mode, audio never leaves your device","body":"Choose local processing and there's no upload step: dictation and drafts are handled entirely on your machine."},{"title":"Free for the whole caseload","body":"Unlimited local dictation at $0, every RBT and BCBA, forever. No per-seat licence to renew."},{"title":"In local mode, nothing goes to a third-party transcriber","body":"Choose local processing and no outsourced service ever hears the file. The dictation stays between you and your machine."},{"title":"Works fully offline","body":"A client's home, the car, a hallway with no wifi. No signal required, so it never stalls mid-dictation."},{"title":"Dictate into CentralReach or any EHR","body":"A system-wide hotkey drops clean text wherever your cursor is: CentralReach, Rethink, WebABA, an email."},{"title":"Optional AI formatting for your draft","body":"Create a custom Note Action to format dictated details into an ABC-style draft, then verify every field against the session record before filing it."},{"title":"You control storage and deletion","body":"Local transcription processes dictation on your machine. Sync, Cloud transcription and connected AI providers use separate data paths, so choose them according to your practice's requirements."},{"title":"Open source, auditable by anyone","body":"Your compliance or IT team can read every line on GitHub. No faith required."},{"title":"Handles ABA terminology","body":"Add ABC, MO, SIB, FBA, program names and practice-specific acronyms to the custom dictionary for better transcription."},{"title":"99 languages, even offline","body":"The local Whisper model alone covers 99 languages, so you can document sessions in the language your team actually speaks with no internet required."}]},"faq":{"label":"Before you download","title":"The questions ABA therapists ask first.","subtitle":"Straight answers on confidentiality, cost and accuracy. If yours isn't here, ask us in Discord.","items":[{"question":"Does the audio ever leave my device?","answer":"With a local transcription model, the recording stays on your device. With OpenWhispr Cloud, audio is sent only to transcribe it, then immediately discarded. Choose the mode that fits the session and your workflow."},{"question":"Is this appropriate for confidential client behavioral data?","answer":"Confidentiality and any HIPAA obligations are your responsibility, and no tool discharges them for you. Local processing is the deliberate choice when a session covers sensitive behavioral or health details. Cloud transcription discards audio immediately after transcription. This is not compliance advice; follow your practice's and payer's requirements."},{"question":"Can it handle ABA terminology and acronyms?","answer":"Yes. Local Whisper models cover a wide range of accents and up to 99 languages (other local models, like Parakeet and Nemotron, cover fewer). Cloud transcription covers 100+ languages. The custom dictionary lets you add ABC, MO, SIB, FBA, client names and the program terms your practice uses so they come out right."},{"question":"Will it work between sessions with no signal?","answer":"Yes. Local models run entirely offline. Dictate in the car, the hallway or a client's home with no wifi, and file the note once you're back online. Choose local before you need that offline path."},{"question":"Does it draft the note, or replace my ABA record system?","answer":"It can transcribe your dictation, and a custom Note Action can format a draft. It does not collect ABA data, manage treatment plans, capture signatures or guarantee payer-compliant documentation. Review the draft and complete the note in your existing system."},{"question":"Is it actually free?","answer":"Yes. Free gives every RBT and BCBA useful local dictation plus a limited Cloud allowance. Upgrade to Pro for unlimited Cloud transcription, more meeting-recording time, sync, mobile, and API/MCP access. Business adds unlimited meeting recordings, Agent mode, and chat over data."}]},"reviews":{"label":"What users say","title":"Built by people who don't want to be recorded either.","subtitle":"From our Discord, GitHub and inboxes, the same instinct that matters when a client's records are on the line.","score":{"value":"0","label":"Confidential client files on a server","body":"In local mode the audio never leaves your device. Nothing is stored on a server, and nothing is sent to a transcriber."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for ABA therapists","tiers":[{"id":"free","name":"Free","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":false},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","badge":"Most Popular","cta":{"label":"Start with Business"},"featured":true}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"RBTs and BCBAs who want a useful free start with local dictation and limited Cloud transcription"}]},{"parts":[{"text":"RBTs and BCBAs who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"ABA practices documenting at volume with Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Useful Free access, including local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some conversations are confidential. We built for those.","paragraphs":["OpenWhispr started with a simple objection: the notes that describe a client's most sensitive moments shouldn't have to pass through a stranger's server to become documentation.","We build in public. The repository has thousands of stars and hundreds of forks, and anyone at your practice can inspect the code. In local mode, dictation is processed on your machine.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · Cloud or local","title":"Dictate the details. Review the ABA session note.","body":"Download OpenWhispr, dictate while the session is fresh, and review the text before pasting it into your existing record system.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free on Mac, Windows & Linux. Community support is available in Discord."}}},"useCaseTherapists":{"breadcrumb":{"home":"Home","useCases":"Use Cases","therapists":"For Therapists"},"metadata":{"title":"Free Therapist Dictation Software - OpenWhispr","description":"OpenWhispr is free therapist dictation software for session notes. Choose Cloud for speed or local processing for sensitive and offline work.","keywords":["therapist dictation software","therapy dictation software","session note dictation","mental health dictation software","offline therapy dictation","clinical note dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Therapist dictation software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source dictation software for therapists. Choose Cloud for top accuracy and speed without a local model download, or local for sensitive and offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation software for therapists","title":"Dictate the session. Get the note written up.","body":"Turn spoken session notes into clean text while the details are still clear. Choose Cloud for no-download speed or local processing for sensitive and offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Therapist listening to a client seated on a sofa across a wood-panelled office, a laptop open on the side table between them"}},"trustBar":{"label":"Dictate into:","names":["SimplePractice","TherapyNotes","Jane","TheraNest","Sessions Health","Microsoft Word"]},"pain":{"label":"What slows a session note","title":"You remember the session now. The keyboard comes later.","subtitle":"Capture the clinical detail while it is fresh, then choose the transcription path that fits the session and your setting.","cards":[{"title":"A full day leaves notes until evening.","body":"Back-to-back sessions leave little room to type between clients. Dictate the note while the session is fresh, then review the clean text before you save it."},{"title":"Client conversations need a deliberate path.","body":"Choose local transcription for highly sensitive sessions or offline work. Your device processes the audio, so you keep control of where the transcription happens."},{"title":"Names and terms need to come out right.","body":"Add client names, medications, modalities, and practice terms to the custom dictionary so OpenWhispr has the spellings you use."},{"title":"The quick setup should stay quick.","body":"When speed matters most, Cloud avoids a local model download and discards transcription audio immediately after transcription."}]},"features":{"label":"What you get","title":"Session notes while the session is still clear.","subtitle":"Use Cloud for fast transcription without a model download. For sensitive notes or an offline day, local processing keeps transcription on your device.","items":[{"image":{"label":"Therapist speaking while seated at an open laptop at a table in a warm office with a bookshelf and armchair"},"title":"Cloud transcription without a local model download","body":"Press your hotkey and dictate the session note. Cloud transcribes without a local model download, then discards the transcription audio immediately."},{"image":{"label":"Therapist gesturing while talking with a client, both in armchairs, a notebook on his lap and a laptop on the side table"},"title":"Local processing for sensitive or offline work","body":"Choose a local model when a highly sensitive session or an offline day calls for device-only transcription. Review the text before it becomes part of the record."},{"image":{"label":"Counsellor working at a laptop along a window counter in a counseling centre, a closed folder beside them"},"title":"Dictation wherever your cursor is","body":"Use a hotkey to dictate into your practice system, a referral letter, an email, or Word. Clean, punctuated text lands where you are already working."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Client session audio sent to us","value":"None","footnote":"in local mode, transcription stays on your device"},{"label":"Cost per therapist","value":"$$0","footnote":"unlimited local dictation, every therapist, forever"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first dictated session note"}]},"howItWorks":{"label":"How it works","title":"A session note written up before the next client.","subtitle":"Start free, choose Cloud for the quickest high-performance setup or local for sensitive and offline work, then dictate while the session is fresh.","image":{"label":"Man speaking while typing at a laptop at a round table in a bright office lounge with plants"},"steps":[{"title":"Download for free.","body":"Mac, Windows, or Linux. Choose Cloud for no local model download or local for sensitive and offline work."},{"title":"Dictate the session note.","body":"Press your hotkey, speak the details you need to record, and OpenWhispr turns your voice into text."},{"title":"Review, save, delete.","body":"Edit the text, save it in your practice system, and delete the audio when you are done. You stay in control of the record."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons therapists choose OpenWhispr","subtitle":"A dictation tool touches sensitive conversations. Here are the product facts, so you can choose the setup that fits your practice.","items":[{"title":"Local audio stays on your device","body":"Choose local transcription when you want the audio processed on your own device, with no internet connection required."},{"title":"Free local dictation for the whole practice","body":"Unlimited local models are included at $0. Start with every clinician who needs to dictate, without a per-seat transcription limit."},{"title":"Works without Wi-Fi","body":"Local models run offline, so a weak clinic connection or a day away from the office does not stop dictation."},{"title":"A custom dictionary for practice language","body":"Add client names, medications, modalities, and acronyms so the words your practice uses have the spellings you expect."},{"title":"Dictate into the record you already use","body":"A system-wide hotkey puts text where your cursor is, whether that is your practice system, Word, or an email."},{"title":"Cloud when setup speed matters","body":"OpenWhispr Cloud delivers top accuracy and speed without asking you to download a local transcription model."},{"title":"Cloud audio is not kept","body":"When you use OpenWhispr Cloud, audio is sent only to transcribe it and is immediately discarded afterwards."},{"title":"100+ languages","body":"Choose from speech-to-text support for more than 100 languages across every plan."},{"title":"Unlimited Cloud transcription when you need it","body":"Pro adds unlimited Cloud transcription, sync, mobile, and API/MCP access. Business adds Agent mode and chat over data."},{"title":"Open source, auditable by anyone","body":"The code is public on GitHub, so you can inspect how OpenWhispr works instead of taking a privacy slogan on faith."}]},"faq":{"label":"Before you download","title":"The questions therapists ask first.","subtitle":"Straight answers on privacy, cost, and how the transcription choices work. If yours is not here, ask us in Discord.","items":[{"question":"Does session audio ever leave my device?","answer":"With a local transcription model, the recording stays on your device. With OpenWhispr Cloud, audio is sent only to transcribe it, then immediately discarded. Choose the mode that fits the session and your workflow."},{"question":"Can I use this for confidential client notes?","answer":"Confidentiality is your professional responsibility, and no tool removes that responsibility. Local processing is the deliberate choice when a highly sensitive session calls for device-only work. Cloud transcription discards audio immediately after transcription. Follow the standards and requirements that apply to your practice."},{"question":"Can it recognize client names and clinical terminology?","answer":"Yes. Add names, medications, modalities, and practice terms to the custom dictionary so OpenWhispr uses the spellings you need. It also supports speech-to-text in more than 100 languages."},{"question":"Will it work when I do not have Wi-Fi?","answer":"Yes. Local transcription models run entirely on your device, without an internet connection. Choose local before you need that offline path."},{"question":"Does it write a note or just transcribe?","answer":"OpenWhispr turns your spoken session note into clean text. You can then review, edit, and save it in the record system you use."},{"question":"Is the free plan actually useful?","answer":"Yes. Free includes unlimited local dictation, a custom dictionary, and a limited Cloud allowance. Upgrade to Pro for unlimited Cloud transcription, more meeting-recording time, sync, mobile, and API/MCP access. Business adds unlimited meeting recordings, Agent mode, and chat over data."}]},"reviews":{"label":"What users say","title":"Built by people who do not want every conversation recorded.","subtitle":"From our Discord, GitHub, and inboxes, the same instinct that matters in a confidential practice.","score":{"value":"0","label":"Client session audio on a server","body":"In local mode the audio never leaves your device. Nothing is stored on a server, and nothing is sent to a transcriber."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for therapists","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Therapists who want a useful free start with local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Therapists who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Practices documenting at volume with Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Useful Free access, including local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Enterprise includes team features and admin controls, SSO, SAML, SCIM, audit logs, retention controls, and dedicated support."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some conversations are private. We built for those.","paragraphs":["OpenWhispr gives you a deliberate local option when you want speech transcribed on your own device.","We build in public. The code is on GitHub, where anyone can inspect how OpenWhispr works. When you choose local dictation, the transcription stays on your device.","Need help? Join the OpenWhispr community on Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Your session note can start with your voice.","body":"Download OpenWhispr and dictate the details while they are fresh. Choose the transcription path that fits your session and your setting.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free local dictation on Mac, Windows & Linux."}}},"useCaseStudentsWithDisabilities":{"breadcrumb":{"home":"Home","useCases":"Use Cases","studentsWithDisabilities":"For Students with Disabilities"},"metadata":{"title":"Speech-to-Text for Students with Disabilities - OpenWhispr","description":"OpenWhispr is free, open-source speech-to-text software for students with disabilities. Dictate coursework in desktop apps, online or offline.","keywords":["speech-to-text software for students with disabilities","dictation software for students with disabilities","assistive technology for students","student speech-to-text","disability accommodation software","offline dictation software"]},"schema":{"name":"OpenWhispr","applicationCategory":"Assistive speech-to-text software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source speech-to-text software for students with disabilities. Dictate coursework with Cloud transcription or local processing for offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text software for students with disabilities","title":"Say the answer. Get it onto the page.","body":"Dictate notes, essays and answers into the apps you already use. Choose Cloud for fast setup or local processing for offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Student in a wheelchair speaking at a library table with an open laptop and notebook"}},"trustBar":{"label":"Dictate into:","names":["Microsoft Word","Google Docs","Canvas","Blackboard","Notion","OneNote"]},"pain":{"label":"When writing gets in the way","title":"You know the answer. Getting it onto the page is the hard part.","subtitle":"Use your voice for notes, essays and short answers. Then review the text in the tools your course already uses.","cards":[{"title":"The idea is there. The sentence will not come out.","body":"When typing takes more effort than thinking, coursework becomes a fight with the keyboard. Speak the sentence, then edit clean text on the page."},{"title":"An accommodation should work in the actual assignment.","body":"Dictate straight into Word, Google Docs, Canvas or the answer field in front of you. No copying between a recorder and your coursework."},{"title":"A slow laptop should not decide whether you can write.","body":"Choose Cloud when you do not want to download a local model. Audio is sent only to transcribe it, then immediately discarded."},{"title":"Wi-Fi is not guaranteed.","body":"Choose local processing when you need to work offline. The transcription runs on your device, so you can keep writing when the connection drops."}]},"features":{"label":"What you get","title":"Speech-to-text where your coursework happens.","subtitle":"Turn spoken answers into editable text. Pick the mode that works for where and how you study.","items":[{"image":{"label":"Student wearing dark glasses speaking at a laptop in a quiet study room"},"title":"Start speaking without a local model download","body":"Choose Cloud for fast setup without a local model download."},{"image":{"label":"Student with a wrist brace speaking at a laptop in a science classroom"},"title":"Work offline with local processing","body":"Download a local model and dictate on your device without internet. It is a practical choice for study spaces with unreliable Wi-Fi."},{"image":{"label":"Student in a wheelchair speaking at a workshop desk while pressing a large accessible switch"},"title":"Put text where your cursor is","body":"Press a hotkey, speak, and place punctuated text in the document, learning platform or notes app you are already using."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Cost to start using local speech-to-text","value":"$$0","footnote":"unlimited local models on every plan"},{"label":"Cloud transcription on Free","value":"2,000","footnote":"words each week, plus unlimited local speech-to-text"},{"label":"Languages supported","value":"100+","footnote":"speech-to-text on every plan"}]},"howItWorks":{"label":"How it works","title":"Speak the answer. Review the text. Submit the work.","subtitle":"Start free, choose Cloud for fast setup or local processing for offline work, then dictate into the assignment in front of you.","image":{"label":"Student speaking at a desk while pressing a large green accessible button beside a monitor"},"steps":[{"title":"Download for free.","body":"Use OpenWhispr on Mac, Windows or Linux. Choose Cloud for no local model download, or local for offline work."},{"title":"Dictate the answer.","body":"Put your cursor in the document, learning platform or notes app, then speak your sentence."},{"title":"Review and submit.","body":"Edit the text in your own words, check the assignment, and hand it in when it is ready."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Built for the work in front of you","title":"10 practical reasons to choose OpenWhispr","subtitle":"A speech-to-text tool has to work where the assignment happens. These are the practical features you can check before you rely on it.","items":[{"title":"Free local speech-to-text","body":"Local AI models are unlimited and free on every plan, so you can dictate without a subscription clock."},{"title":"Works without Wi-Fi","body":"Local models run entirely on your device. Keep working in a library, on a commute, or anywhere the connection is unreliable."},{"title":"Dictate into the app already open","body":"Use a hotkey to place text at your cursor in documents, notes, email, and browser-based coursework."},{"title":"Cloud when fast setup matters","body":"Cloud transcription offers accuracy and speed without a local model download, with audio immediately discarded after transcription."},{"title":"A custom dictionary for your course","body":"Add names, technical terms and acronyms so the terms you use in class have a better chance of appearing correctly."},{"title":"More than 100 languages","body":"OpenWhispr supports speech-to-text in more than 100 languages on every plan."},{"title":"Editable text, not a locked recording","body":"Your speech becomes text in the document you are writing. Review it, revise it, and make it your own before you submit."},{"title":"Use it across your study tools","body":"Dictate into Word, Google Docs, OneNote, Notion, and browser-based learning platforms from the same desktop app."},{"title":"Open source","body":"The project is MIT-licensed, so students and accessibility teams can inspect how the software works."},{"title":"Upgrade only when the workflow needs it","body":"Free includes local models. Pro adds unlimited OpenWhispr Cloud transcription, sync, mobile, and API and MCP access."}]},"faq":{"label":"Before you download","title":"Questions students ask before trying speech-to-text.","subtitle":"Straight answers on accommodations, offline work, accuracy and cost.","items":[{"question":"Can I use this as speech-to-text software for coursework?","answer":"Yes. OpenWhispr turns speech into editable text at your cursor, so you can dictate notes, drafts and answers into the document or learning platform you are using."},{"question":"Does it work without internet?","answer":"Yes, with a local AI model. Local speech-to-text runs on your device without internet. Cloud transcription needs a connection and is the easy option when you want speed with no local model download."},{"question":"Is OpenWhispr free for students?","answer":"Yes. Free includes unlimited local AI models, 2,000 words a week of OpenWhispr Cloud transcription, 5 hours of meeting recordings a month, and unlimited Cloud transcription with your own API keys."},{"question":"Can it handle course vocabulary and names?","answer":"Use the custom dictionary to add names, acronyms and subject-specific terms. Review the final text before submitting, especially for specialised vocabulary."},{"question":"Does Cloud keep my audio?","answer":"Cloud audio is sent only to transcribe it, then immediately discarded. Local processing keeps the transcription on your device. Choose the mode that fits how and where you study."},{"question":"What do I get by upgrading from Free?","answer":"Pro adds unlimited OpenWhispr Cloud transcription, 20 hours of meeting recordings a month, cross-device sync, personal API access, MCP integration, and the iPhone and iPad app. Business adds unlimited meeting recordings, Agent mode, chat over your data, and priority support."}]},"reviews":{"label":"What people say","title":"Speech-to-text that lets the words get out first.","subtitle":"Three comments from OpenWhispr users.","score":{"value":"$$0","label":"Cost of local speech-to-text","body":"Unlimited local AI models are included on every plan."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for students with disabilities","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Students who want free, unlimited local speech-to-text and a useful amount of Cloud transcription"}]},{"parts":[{"text":"Students who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Students who record more meetings and want Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Free local speech-to-text plus limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud with no local model to download, then immediately discards it.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Assistive technology should not make you trade away control.","paragraphs":["OpenWhispr gives students with disabilities a practical way to turn speech into editable text, in the coursework tools they already use.","OpenWhispr is MIT-licensed and its source code is public on GitHub. Local models run on your device, and Cloud audio is immediately discarded after transcription.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team and community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Your next sentence can start out loud.","body":"Download OpenWhispr, dictate the work in front of you, then review it in your own words.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free forever on Mac, Windows and Linux. Get help from the community in Discord."}}},"useCaseKids":{"breadcrumb":{"home":"Home","useCases":"Use Cases","kids":"For Kids"},"metadata":{"title":"Speech-to-Text Software for Teens 13+ - OpenWhispr","description":"OpenWhispr is open-source speech-to-text software for parents and educators helping teens aged 13 and over turn spoken ideas into editable text.","keywords":["speech to text software for kids","speech to text for teens","dictation software for teens","speech to text for parents","speech to text for educators","offline speech to text for teens"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text software for parents and educators of teens","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source speech-to-text software for parents and educators helping teens aged 13 and over turn spoken ideas into editable text. Choose Cloud for speed or local for offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text software for kids","title":"Say the story. Get it on the page.","body":"Help a teen aged 13 and over speak a story, journal entry or note, then let them review and revise the editable text. Choose Cloud for quick setup and fast transcription, or local when working offline.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"A teenager speaking and gesturing at a laptop on a round table in a sunlit library"}},"trustBar":{"label":"Common writing tools:","names":["Google Docs","Microsoft Word","Apple Pages","Microsoft OneNote","Google Slides","Canva"]},"pain":{"label":"When typing slows a teen's first draft","title":"The idea is ready. Help them get it onto the page.","subtitle":"Set up the tool, let a teen aged 13 and over say the words while they are fresh, then give them editable text to review and revise.","cards":[{"title":"The idea is there. The keyboard gets in the way.","body":"A story opening, a journal thought or a list of ideas can be easy to say and slow to type. OpenWhispr turns spoken words into editable text for the teen to review."},{"title":"An offline writing session needs a practical option.","body":"Choose a local model for an offline writing session. The device processes the audio, so the teen can keep working without an internet connection."},{"title":"Speech-to-text is the start, not the finished work.","body":"Help the teen say the opening out loud. The editable text gives them something to reread, change and build on."},{"title":"Cloud is the quick, no-download option.","body":"Cloud avoids a local model download and gives fast transcription. The transcription audio is discarded immediately afterward."}]},"features":{"label":"What you get","title":"Help a teen get the words down. Then let them make them theirs.","subtitle":"Use Cloud for a fast, no-download start, or local processing for an offline writing session. Both leave the teen with editable text to check and improve.","items":[{"image":{"label":"A teenager working at a laptop at a wooden dining table in a bright kitchen"},"title":"Cloud transcription without a local model download","body":"Help a teen speak an opening, a list of ideas or a personal note into the microphone. Cloud is the no-download option for fast transcription, then discards the transcription audio immediately."},{"image":{"label":"A teenager pointing at a laptop screen at a dining table in the evening, with city lights outside the window"},"title":"Local processing for offline work","body":"Choose a local model when working offline. The device processes the audio, then the teen can reread and edit the words."},{"image":{"label":"A teenager speaking at a laptop in a school workshop, with a robotics kit on the bench"},"title":"Set a hotkey to start dictation","body":"Set the hotkey once, then help the teen say what they want to write and review the text before they keep going."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Audio sent to us in local mode","value":"None","footnote":"in local mode, transcription never leaves your device"},{"label":"Cost to get started","value":"$$0","footnote":"unlimited local dictation, free forever"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first dictated sentence"}]},"howItWorks":{"label":"How it works","title":"From a spoken idea to editable text.","subtitle":"Start free, choose Cloud for the quickest no-download setup or local for offline work, then help a teen aged 13 and over turn a spoken idea into text.","image":{"label":"A teenager speaking at a laptop beside an open notebook on a kitchen counter"},"steps":[{"title":"Set it up once.","body":"Download on Mac, Windows or Linux. Choose Cloud for no local model download or local for offline work."},{"title":"Help them say the idea.","body":"The teen speaks the words they want to get down. OpenWhispr turns their voice into editable text."},{"title":"Let them review and revise.","body":"Read the words together, fix what needs fixing, then let the teen keep writing in the document."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 ways OpenWhispr helps parents and educators","subtitle":"Speech-to-text should help a teen aged 13 and over get an idea onto the page, then leave them with words they can still shape themselves.","items":[{"title":"A local mode for offline work","body":"Set up a local model and the audio stays on the device. No internet connection is required for transcription."},{"title":"Free local dictation","body":"Try unlimited local AI models on the Free plan, with no monthly charge."},{"title":"One hotkey to start dictation","body":"Set a hotkey to start dictation while the teen is writing. Review the resulting text before they keep going."},{"title":"No Wi-Fi required in local mode","body":"Choose a local model when the connection drops. Transcription continues on the device."},{"title":"Start with the sentence they can say","body":"Help the teen speak the opening line, the tricky description or the next thought. Then let them edit the text in their own voice."},{"title":"Editable words to reread straight away","body":"OpenWhispr turns what the teen says into text, so they can reread the words before deciding what to write next."},{"title":"Cloud audio is not kept","body":"Cloud audio is sent only to transcribe it, then immediately discarded."},{"title":"Open source, so you can inspect it","body":"The app is MIT-licensed and its code is available on GitHub."},{"title":"A custom dictionary for names and terms","body":"Add names, places and words that matter to the teen so OpenWhispr has a better chance of getting them right."},{"title":"100+ languages","body":"Choose from 100+ supported languages for multilingual speech-to-text."}]},"faq":{"label":"Before you download","title":"Questions parents and educators consider for teens aged 13+.","subtitle":"Straight answers on age requirements, setup, working offline and helping a teen review editable text.","items":[{"question":"What is the age requirement?","answer":"OpenWhispr is intended for users aged 13 and over. If you are under 18, you confirm that your parent or legal guardian has read these Terms and agrees to them on your behalf."},{"question":"What is the difference between Cloud and local?","answer":"Cloud is the quick option: it gives top accuracy and speed without downloading a local model. Local runs on the device and works offline. Choose the setup that fits the writing session."},{"question":"Does OpenWhispr keep Cloud audio?","answer":"No. OpenWhispr Cloud receives audio only to transcribe it, then immediately discards the transcription audio. Local transcription keeps the audio on your device."},{"question":"Will it work without Wi-Fi?","answer":"Yes, when you choose a local transcription model. Local models run entirely on the device, so the teen can keep dictating without an internet connection."},{"question":"What if it gets a name or word wrong?","answer":"Review the text with the teen before they keep writing. You can add names, places and other useful words to OpenWhispr's custom dictionary to help with future transcripts."},{"question":"What does the Free plan include?","answer":"Free includes unlimited local AI models, 2,000 words of OpenWhispr Cloud transcription each week, 5 hours of meeting recordings each month, 100+ languages and a custom dictionary. Pro adds unlimited OpenWhispr Cloud transcription, sync and mobile."}]},"reviews":{"label":"What users say","title":"For parents and educators helping teens get words onto the page.","subtitle":"These are generic OpenWhispr reviews. The words, names and handles are unchanged.","score":{"value":"$$0","label":"to start with local dictation","body":"The Free plan includes unlimited local AI models. Start with the words you want to get onto the page."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for parents and educators supporting teens","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Parents and educators supporting teens aged 13+ who want a useful free start with local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Parents and educators supporting teens aged 13+ who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Teams recording at volume with Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"A useful free start with local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access for adults managing the setup"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some ideas are easier to say. Help a teen get them into editable text.","paragraphs":["For parents and educators helping teens aged 13 and over, OpenWhispr turns a spoken idea into editable text. The teen can then review, change and shape the words.","We build in public. The app is MIT-licensed and its code is available on GitHub. Local transcription runs on your device, and the code is there for anyone to inspect.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"OpenWhispr team and community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Help a teen get the idea onto the page.","body":"Download OpenWhispr, help a teen aged 13 and over say the first draft out loud, then let them make the editable words their own.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseDentists":{"breadcrumb":{"home":"Home","useCases":"Use Cases","dentists":"For Dentists"},"metadata":{"title":"Dental Dictation Software for Clinical Notes - OpenWhispr","description":"OpenWhispr turns spoken dental notes into editable text. Choose Cloud for no-download speed or local processing for sensitive and offline work.","keywords":["dental dictation software","dentist dictation software","dental voice dictation","dental note dictation","dental transcription software","dictation software for dentists"]},"schema":{"name":"OpenWhispr","applicationCategory":"Dental dictation software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source dental dictation software for clinical notes. Use Cloud without a local model download, or local processing for sensitive and offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dental dictation software for dentists","title":"Finish the note before the next patient.","body":"Place your cursor in the patient record and dictate the clinical note as soon as the appointment ends. Choose Cloud for no-download speed or local processing for sensitive and offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"A dentist in green scrubs stands at a counter, speaking while working at a laptop in a treatment room overlooking the sea."}},"trustBar":{"label":"Dictate into:","names":["Patient charts","Clinical notes","Procedure notes","Treatment plans","Referral letters","Microsoft Word"]},"pain":{"label":"What slows clinical notes","title":"The appointment is done. The chart still needs writing.","subtitle":"You need the details while they are fresh, a mode that fits the setting, and text ready to review before you sign.","cards":[{"title":"Notes written after the patient has gone.","body":"A full schedule leaves charting for the gap between patients or the end of the day. Dictate while the treatment details are still clear, then review the text before it enters the record."},{"title":"Procedure names that need checking.","body":"Tooth numbers, materials and procedure names are easy to mistype under time pressure. Add your practice vocabulary to the custom dictionary, then review the dictated note."},{"title":"A local path for sensitive or offline work.","body":"For device-only processing or without a dependable connection, use a local transcription model. Your device processes the recording, so you can keep working where the signal does not reach."},{"title":"A quick setup should still be a choice.","body":"When speed matters most, Cloud avoids a local model download and discards transcription audio immediately after transcription."}]},"features":{"label":"What you get","title":"Clinical notes, ready to review before the next patient.","subtitle":"Cloud transcribes without a model download. A local model is available for sensitive or offline work.","items":[{"image":{"label":"A dentist in navy scrubs sits at a laptop, speaking while holding a pen over a paper chart in an empty treatment room."},"title":"Cloud transcription without a local model download","body":"Dictate an exam, treatment or follow-up note into your microphone. Cloud transcribes without a local model download, then discards transcription audio immediately."},{"image":{"label":"A dentist on a stool beside a laptop talks with a patient sitting upright in the dental chair."},"title":"Local processing for sensitive and offline work","body":"Choose a local model when device-only processing or an offline day is the right fit. Review the editable text before you add it to the patient chart."},{"image":{"label":"A dentist at a laptop talks with an older patient reclined in the chair wearing a dental bib."},"title":"Dictation wherever your cursor is","body":"Press a hotkey and dictate into a dental chart, email or document. Clean, punctuated text lands where you are already working."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Sensitive audio sent to us","value":"None","footnote":"in local mode, transcription never leaves your device"},{"label":"Local dictation","value":"$$0","footnote":"unlimited local AI models on every plan"},{"label":"Languages supported","value":"100+","footnote":"on every plan"}]},"howItWorks":{"label":"How it works","title":"A clinical note transcribed before the next patient.","subtitle":"Start free. Cloud transcribes without a local model download, while a local model is available for sensitive and offline work. Review the text before the next appointment.","image":{"label":"A dentist in a white coat works at a laptop at her office desk, a folder beside her and the treatment room visible behind."},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Use Cloud without a local model download, or use a local model for sensitive and offline work."},{"title":"Dictate the note.","body":"Speak the clinical note, and OpenWhispr transcribes it into editable text."},{"title":"Review and add to the chart.","body":"Check the text, paste it into your dental software, and record it using your practice process."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Reasons to choose","title":"10 reasons OpenWhispr fits dental dictation","subtitle":"A dental dictation workflow needs speed, review and a practical choice of processing mode. Here are the product facts.","items":[{"title":"Unlimited local models at no cost","body":"Every plan includes unlimited local AI models, so local speech-to-text stays free."},{"title":"Cloud audio is never stored","body":"OpenWhispr Cloud receives audio to transcribe it, then immediately discards it."},{"title":"Local transcription works offline","body":"Local models run entirely on your device with no internet required."},{"title":"A custom dictionary for practice vocabulary","body":"Add names, jargon and acronyms so the transcript can better reflect the words your practice uses."},{"title":"Dictate into the active field","body":"Use the desktop app to dictate where your cursor is, whether that is a chart field, email or document."},{"title":"100+ languages supported","body":"Every plan supports transcription in more than 100 languages."},{"title":"Start with 2,000 Cloud words each week","body":"The Free plan includes 2,000 words of OpenWhispr Cloud transcription per week."},{"title":"Use your own API keys","body":"Every plan supports your own OpenAI, Anthropic, Gemini or Groq API keys for unlimited cloud transcription."},{"title":"Upgrade for unlimited Cloud transcription","body":"Pro includes unlimited OpenWhispr Cloud transcription, plus sync, mobile and API/MCP access."},{"title":"Open source under MIT","body":"OpenWhispr is open source, so your technical team can inspect the software."}]},"faq":{"label":"Before you download","title":"The questions dentists ask first.","subtitle":"Straight answers on clinical notes, processing modes, cost and workflow.","items":[{"question":"Does audio ever leave my device?","answer":"With a local transcription model, the recording stays on your device. With OpenWhispr Cloud, audio is sent only to transcribe it, then immediately discarded. Choose the mode that fits your workflow."},{"question":"Can I use this for dental clinical notes?","answer":"OpenWhispr turns your spoken words into editable text. Dictate the note while the appointment is fresh, then review it and enter it into your dental record according to your practice process."},{"question":"Can it handle dental terminology?","answer":"Use the custom dictionary to teach OpenWhispr the names, jargon and acronyms your practice uses. Review the transcript before it becomes part of the clinical record."},{"question":"Will it work without internet?","answer":"Yes. Local models run entirely on your device and do not need an internet connection. Use a local model when you need that offline path."},{"question":"Does it write the clinical record for me?","answer":"It transcribes spoken words into editable text. You remain responsible for reviewing the note and recording it according to your clinical and practice requirements."},{"question":"What do I get by upgrading?","answer":"Free includes unlimited local models and a limited Cloud allowance. Pro adds unlimited OpenWhispr Cloud transcription, more meeting-recording time, sync, mobile and API/MCP access. Business adds unlimited meeting recordings, Agent mode and chat over data."}]},"reviews":{"label":"What users say","title":"Voice input that earns a place in the workflow.","subtitle":"Real OpenWhispr user reviews.","score":{"value":"0","label":"Audio retained after Cloud transcription","body":"Cloud audio is sent to transcribe it, then immediately discarded. Local transcription stays on your device."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Start at $0.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for dentists","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"Unlimited local AI models","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Clinicians who want local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Clinicians who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Teams that need Agent mode, chat over data, and unlimited meeting recordings"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Useful Free access, including local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud without local models to download, then the audio is never stored.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Dictation software for the work between appointments.","paragraphs":["OpenWhispr helps dentists turn spoken clinical notes into editable text, so the details are ready to review while the appointment is still fresh.","It is built for people who want practical control over their workflow, whether you work alone or with a whole dental team.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"OpenWhispr founder photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Dictate the note. Review the text. Move to the next patient.","body":"Download OpenWhispr for free and turn spoken clinical notes into editable text with Cloud or local processing.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Download for Mac, Windows or Linux. If you need help, ask us in Discord."}}},"useCaseHearingImpaired":{"breadcrumb":{"home":"Home","useCases":"Use Cases","hearingImpaired":"For Deaf and Hard-of-Hearing People"},"metadata":{"title":"Free Speech-to-Text for Hearing Impaired - OpenWhispr","description":"OpenWhispr turns recorded speech into text you can read. Use it free with local models, or choose Cloud for fast transcription with no model download.","keywords":["best speech-to-text app for hearing impaired free","free speech-to-text app for hearing impaired","speech-to-text for deaf people","speech-to-text for hard of hearing","free audio transcription app","offline speech-to-text app"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text application","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source speech-to-text software for turning recorded speech into readable text. Choose Cloud for top accuracy and speed without a local model download, or local for private and offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text app for hearing-impaired people","title":"Record the conversation. Read it back as text.","body":"Use Meeting Notes to record conversations and turn them into a transcript you can read, review and keep. Choose Cloud for fast transcription without a model download, or local processing for private and offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Two students sitting together at an outdoor campus table, looking at a laptop screen; one wears a hearing aid"}},"trustBar":{"label":"Save text in:","names":["Microsoft Word","Google Docs","Apple Notes","Notion","Slack","Email"]},"pain":{"label":"When speech is hard to follow","title":"The words disappear before you can ask for them again.","subtitle":"You need text you can return to, a mode that fits the moment, and a free way to start.","cards":[{"title":"The speaker has already moved on.","body":"A missed name, address or next step is hard to reconstruct from memory. OpenWhispr turns recorded speech into text you can reread before the details are gone."},{"title":"You need a private or offline option.","body":"Choose local transcription when the conversation is private or you have no connection. Your device processes the audio, so you can keep working without sending it away."},{"title":"Missing a sentence is tiring.","body":"Instead of piecing a detail together from memory, search and reread the transcript. Keep the text where you can return to it."},{"title":"You want the quickest setup first.","body":"When speed matters most, Cloud avoids a local model download and discards transcription audio immediately after transcription."}]},"features":{"label":"What you get","title":"Text you can read back. A mode that fits the moment.","subtitle":"Use Cloud for fast transcription or local processing for private and offline situations.","items":[{"image":{"label":"Three colleagues talking around a meeting table in a glass-walled office, an open laptop between them; one wears a hearing aid"},"title":"Cloud transcription without a local model download","body":"Add recorded speech and get readable text quickly. Cloud gives you top accuracy and speed without a local model download, then discards the transcription audio immediately."},{"image":{"label":"Two women in conversation across a bright studio table with an open laptop between them; one wears a cochlear implant"},"title":"Local processing for private or offline moments","body":"Choose a local model when you need device-only processing or no connection. Read and review the text on your own device."},{"image":{"label":"A woman wearing a cochlear implant reading on a tablet at a lecture-hall desk, an open notebook and pen beside her"},"title":"Keep the transcript with your notes","body":"OpenWhispr Meeting Notes keeps the conversation as a transcript you can revisit. Ask the Agent to search across your meetings and bring back the details you need. Copy useful text into a note, document, email or chat when you need it elsewhere."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Audio sent away in local mode","value":"None","footnote":"local transcription runs on your device"},{"label":"Cost to start","value":"$$0","footnote":"unlimited local models, forever"},{"label":"Setup time","value":"No fixed time","footnote":"choose the setup that fits you"}]},"howItWorks":{"label":"How it works","title":"A written record before the details fade.","subtitle":"Start free, choose Cloud for a quick high-performance setup or local for private and offline work, then read the words back when you need them.","image":{"label":"Three people leaning over a laptop at a wooden table in a community space, one writing in a notebook"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Start free with unlimited local models that work offline, or choose Cloud when you do not want a model download."},{"title":"Start Meeting Notes.","body":"Record the conversation in Meeting Notes and turn it into a transcript you can read."},{"title":"Read, save, revisit.","body":"Review the transcript, save the useful parts in your notes, and return to the text when you need a detail."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 practical reasons to try OpenWhispr","subtitle":"Speech-to-text should be practical, private when you need it, and free to try. Here are the product facts.","items":[{"title":"Use local transcription when privacy matters","body":"Local AI models run entirely on your device with no internet required. Use them when the conversation calls for device-only processing."},{"title":"Start free with unlimited local models","body":"Free includes unlimited local AI models, plus 2,000 Cloud transcription words each week."},{"title":"Read back the words instead of replaying everything","body":"A transcript gives you a written record to scan, search and save when a detail was missed the first time."},{"title":"Keep going without a connection","body":"Local models run offline. Use them on a commute, in a waiting room, or anywhere reliable internet is not available."},{"title":"Keep useful text in the tools you already use","body":"Copy a transcript into a note, document, email or chat and leave yourself a record you can return to."},{"title":"Choose Cloud when you need speed","body":"Cloud transcription delivers top accuracy and speed without a local model download."},{"title":"Cloud transcription audio is immediately discarded","body":"When you choose OpenWhispr Cloud, audio is sent only to transcribe it, then never stored."},{"title":"Open source means the project is inspectable","body":"OpenWhispr is open source under the MIT license, so anyone can inspect the project rather than take its word for it."},{"title":"Teach it the names you need to get right","body":"Add names, places and recurring terms to the custom dictionary for sharper transcripts."},{"title":"Work across 100+ languages","body":"OpenWhispr supports 100+ languages on every plan, so the spoken language does not have to be English."}]},"faq":{"label":"Before you download","title":"The questions to ask before you download.","subtitle":"Straight answers on privacy, cost and what speech-to-text can do. If yours is not here, ask us in Discord.","items":[{"question":"Does the audio ever leave my device?","answer":"With a local transcription model, the recording stays on your device. With OpenWhispr Cloud, audio is sent only to transcribe it, then immediately discarded. Choose the mode that fits the conversation and your needs."},{"question":"Is this a replacement for hearing aids or assistive listening devices?","answer":"No. OpenWhispr is speech-to-text software for turning recorded speech into readable text. It does not replace hearing aids, cochlear implants, captions provided by a service, or professional accessibility support."},{"question":"Can it handle names, places and unfamiliar words?","answer":"The custom dictionary lets you add names, places, jargon and acronyms for sharper transcripts. OpenWhispr also supports 100+ languages."},{"question":"Will it work with no signal?","answer":"Yes. Local models run entirely offline. Choose local before you need it, then transcribe on your device without an internet connection."},{"question":"Does it provide live captions?","answer":"This page covers turning recorded speech into text you can read and revisit. It makes no claim that OpenWhispr provides live captions for a class, meeting or medical appointment."},{"question":"Is it actually free?","answer":"Yes. Free includes unlimited local AI models, 2,000 Cloud transcription words a week, and five hours of meeting recordings a month. Upgrade to Pro for unlimited Cloud transcription, more meeting-recording time, sync, mobile, and API/MCP access. Business adds unlimited meeting recordings, Agent mode, and chat over data."}]},"reviews":{"label":"What users say","title":"Built for people who want words they can read back.","subtitle":"These users came for a quicker, more dependable way to turn speech into text.","score":{"value":"0","label":"Stored Cloud transcription audio","body":"In local mode the audio stays on your device. With OpenWhispr Cloud, audio is sent only to transcribe it, then immediately discarded."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for hearing-impaired people","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"People who want a useful free start with local transcription and limited Cloud transcription"}]},{"parts":[{"text":"People who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"People transcribing at volume with Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Useful Free access, including local transcription and limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some spoken words need to become text you can keep.","paragraphs":["OpenWhispr started with a simple idea: spoken words should be easy to turn into text you can read, search and keep.","OpenWhispr is open source under the MIT license. Local AI models run on your device with no internet required.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team and community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Keep the spoken words you need in text.","body":"Download OpenWhispr, start Meeting Notes for the next conversation, then read and revisit the transcript when you need the details. Start free on your computer.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free on Mac, Windows & Linux. Community support in Discord."},"byline":{"authorLabel":"Written by","author":"Joshua Padoa","aiDisclosure":"Drafted with AI assistance","reviewerLabel":"Reviewed and fact checked by","reviewer":"Wesley van der Hoop","updatedLabel":"Last reviewed","dateModified":"2026-08-27"}}},"useCaseDyslexia":{"breadcrumb":{"home":"Home","useCases":"Use Cases","dyslexia":"For Dyslexia"},"metadata":{"title":"Best Speech-to-Text App for Dyslexia - OpenWhispr","description":"OpenWhispr is a free, open-source speech-to-text app for dyslexia. Speak naturally, get clean text, and choose local or Cloud transcription.","keywords":["speech to text app for dyslexia","best speech to text app for dyslexia","dyslexia speech to text","voice typing for dyslexia","dictation software for dyslexia","free speech to text for dyslexia"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source speech-to-text software for people with dyslexia. Choose Cloud for speed without a local model download, or local for device-only and offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text software for people with dyslexia","title":"Focus on what you want to say, not how to spell it.","body":"When spelling or typing interrupts your flow, speak instead. OpenWhispr turns your voice into editable text wherever your cursor is, so you can get the idea down and review it afterward.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Man speaking at his laptop in a design studio, sketches pinned on the wall behind him"}},"trustBar":{"label":"Write in:","names":["Microsoft Word","Google Docs","Notion","Gmail","Canvas","Blackboard"]},"pain":{"label":"When typing breaks the flow","title":"You know what you want to say. Getting it typed is the hard part.","subtitle":"Start with your voice, keep the words moving, and stay with the app in front of you.","cards":[{"title":"The thought arrives faster than the sentence.","body":"When spelling or typing interrupts the thought, the blank page takes over. OpenWhispr turns what you say into text, so you can start with the idea instead."},{"title":"You should not have to change apps to write.","body":"A new writing tool can add another hurdle. Use a hotkey and dictate into the text field you already have open, then keep working from there."},{"title":"A misspelled word can stop the next sentence.","body":"Use the custom dictionary for names, course terms, and words you use often. It helps guide the transcription toward the words you mean."},{"title":"The quick option and the private option should both be useful.","body":"Cloud starts without a local model download. Local keeps transcription on your device and works offline. Pick the path that lets you get the words down."}]},"features":{"label":"What you get","title":"Start with your voice. Get the words down.","subtitle":"Say the word, get editable text, and keep the idea moving. Choose Cloud for no model download, or local transcription for on-device and offline writing.","items":[{"image":{"label":"Student speaking at her laptop in a library, a closed book resting under her hands"},"title":"Speak instead of starting with the keyboard","body":"Hold your hotkey and speak when the keyboard would slow you down. OpenWhispr gives you editable text ready to paste where you are writing, then you can review and correct it."},{"image":{"label":"Woman sitting on a couch in a bright office lounge, speaking at a laptop on a low table"},"title":"Cloud transcription when you want to start fast","body":"Cloud gives you top accuracy and speed without a local model download. The transcription audio is discarded immediately after transcription."},{"image":{"label":"Man speaking at his laptop on a moving train, countryside blurring past the window"},"title":"Local transcription when you need it on your device","body":"Local models run entirely on your device and work without an internet connection. Pick this path for device-only processing or a day with unreliable Wi-Fi."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Audio sent to us in local mode","value":"None","footnote":"local transcription stays on your device"},{"label":"Cost to start","value":"$$0","footnote":"unlimited local dictation, forever"},{"label":"Choose the way you write","value":"Cloud or local","footnote":"start with Cloud or download a local model for offline writing"}]},"howItWorks":{"label":"How it works","title":"Get the thought down before it disappears.","subtitle":"Download free, choose the setup that fits, then say the next sentence before the thought disappears.","image":{"label":"Woman at a kitchen table speaking while working at her laptop, printed pages and reading glasses beside her"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Start with Cloud if you want no model download. Choose local transcription for on-device and offline writing."},{"title":"Speak the sentence.","body":"Press your hotkey and say what you mean. OpenWhispr turns your speech into text."},{"title":"Read and edit.","body":"Keep the words that sound like you, change the ones that do not, and move on to the next idea."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Keep the thought moving","title":"10 product facts for speech-to-text and dyslexia","subtitle":"Say the word instead of stopping at the keyboard. Then review the text and keep the idea moving.","items":[{"title":"Keep moving past the hard word","body":"Use a hotkey to turn the next thought into text. The first pass gives you something to review instead of a blank page."},{"title":"Unlimited local dictation is free","body":"Run local speech-to-text on every plan at $0. There is no minute meter running while you get your words down."},{"title":"Cloud is quick to start","body":"Choose Cloud for top accuracy and speed without a local model download. It is the quickest way to start dictating."},{"title":"Local works without Wi-Fi","body":"Choose a local model when the internet is unreliable or you want device-only processing. Your speech stays on the computer doing the work."},{"title":"Use the apps you already know","body":"Dictate into the text field in front of you, then stay in Word, Docs, email, or the learning platform you already use."},{"title":"Make names and key terms easier to catch","body":"Add names, course terms, and recurring words to the custom dictionary. OpenWhispr sends them to transcription as hints, giving it more context for the words you mean."},{"title":"Cloud audio is not kept","body":"When you use OpenWhispr Cloud, audio is sent only to transcribe it and is immediately discarded after transcription."},{"title":"Open source means you can inspect it","body":"OpenWhispr is open source. You can look at how the app handles your speech instead of taking the claim on faith."},{"title":"Keep writing in more than one language","body":"OpenWhispr supports 100+ languages, so the speech-to-text option can fit the language you need to use."},{"title":"Upgrade only when Cloud and workflow tools help","body":"Free includes local models and a limited Cloud allowance. Pro adds unlimited Cloud transcription, sync, mobile, and API or MCP access."}]},"faq":{"label":"Before you download","title":"Questions about speech-to-text for dyslexia.","subtitle":"Straight answers before you try speech-to-text: writing, privacy, cost, and Cloud or local transcription.","items":[{"question":"What changes when spelling slows me down?","answer":"Speak the sentence when typing or spelling interrupts your flow. OpenWhispr gives you editable text to review and correct afterward. It does not diagnose, treat, or guarantee outcomes for dyslexia."},{"question":"Should I choose Cloud or local transcription?","answer":"Choose Cloud when you want top accuracy and speed without downloading a local model. Choose a local model when you want device-only processing or need to write offline. Cloud audio is sent only to transcribe it, then immediately discarded."},{"question":"How can I give recurring names and terms a better chance of coming through?","answer":"Use the custom dictionary for names, course terms, acronyms, and recurring words. OpenWhispr sends them to transcription as hints, giving it more context for the words you mean."},{"question":"Will it work when I have no internet?","answer":"Yes, with a local model. Local transcription runs entirely on your device after the model download, so you can dictate when the Wi-Fi is unreliable or unavailable."},{"question":"Can I use it in Word, Google Docs, or email?","answer":"Yes. Press a system-wide hotkey and dictate into the text field where you are already working. The text lands at your cursor, so you can keep using the app you know."},{"question":"Is it actually free?","answer":"Yes. Free includes unlimited local transcription, local AI models, and 2,000 Cloud words a week. Upgrade to Pro for unlimited Cloud transcription, more meeting-recording time, sync, mobile, and API or MCP access. Business adds unlimited meeting recordings, Agent mode, and chat over data."}]},"reviews":{"label":"What users say","title":"For people who would rather talk than fight the keyboard.","subtitle":"These are general OpenWhispr reviews, not testimonials about dyslexia.","score":{"value":"0","label":"Audio stored on our servers in local mode","body":"With local transcription, audio stays on your device. With Cloud, audio is sent only to transcribe it, then immediately discarded."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for people with dyslexia","tiers":[{"id":"free","name":"Free","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"}},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","badge":"Most Popular","cta":{"label":"Start with Business"},"featured":true}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"People who want unlimited local transcription and a free start with Cloud"}]},{"parts":[{"text":"People who want unlimited Cloud transcription plus mobile, sync, and API or MCP access"}]},{"parts":[{"text":"People recording at volume with Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local transcription plus a limited Cloud allowance"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Your ideas should not get stuck at the keyboard.","paragraphs":["OpenWhispr started with a simple idea: getting words onto the page should not depend on getting every keystroke right the first time.","We build in public. Every release and every line of code is on GitHub, where people can inspect how local transcription works instead of taking it on faith.","Community support is available in Discord."],"image":{"label":"OpenWhispr team and community"}},"finalCta":{"tagline":"Free · open source · speech-to-text","title":"Say the next sentence instead of stopping at the keyboard.","body":"Download OpenWhispr, speak the next thought while it is fresh, then paste, review, and edit the text before continuing.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free forever on Mac, Windows, and Linux."}}},"useCaseRsi":{"breadcrumb":{"home":"Home","useCases":"Use Cases","rsi":"For People with RSI"},"metadata":{"title":"Voice to Text Apps for RSI - OpenWhispr","description":"OpenWhispr turns speech into editable text in your current app, helping people with RSI reduce sustained keyboard and mouse input.","keywords":["voice to text apps for rsi","voice to text for repetitive strain injury","rsi dictation software","speech to text for wrist pain","hands free typing app","accessibility voice typing"]},"schema":{"name":"OpenWhispr","applicationCategory":"Voice-to-text accessibility software","operatingSystem":"macOS, Windows, Linux","description":"Open-source voice-to-text software that lets people with RSI dictate editable text into the current app, then review and continue with less sustained manual input.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Voice to text apps for people with RSI","title":"Give your hands a break. Speak instead.","body":"Reduce the keyboard and mouse work that can become uncomfortable with RSI. Speak naturally to turn your words into text for messages, forms and notes. Choose fast Cloud transcription or local processing for offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Man in a wrist brace speaking at a laptop in a home office"}},"trustBar":{"label":"Use alongside:","names":["Gmail","Google Docs","Slack","Microsoft Word","Notion","ChatGPT"]},"pain":{"label":"When typing is the hard part","title":"The work is manageable. The keyboard and mouse time adds up.","subtitle":"A long email, form or document can turn into a stop-start job when your hands need a break. Speak instead, then review what lands on the page.","cards":[{"title":"A short reply becomes a long typing session.","body":"In tap mode, press the hotkey once to start and again to stop, then review the transcribed reply before you send it."},{"title":"A flare-up does not wait for a quieter week.","body":"When typing is difficult today, use your voice for the email, note or form in front of you. OpenWhispr gives you another way to get the words down."},{"title":"Cloud should be quick to start.","body":"Choose Cloud when you want fast transcription without downloading a local model. Audio is sent only to transcribe it, then immediately discarded."},{"title":"An internet drop should not end the task.","body":"Choose a local model when you need to work offline. Local transcription runs on your device, so you can keep dictating without a connection."}]},"features":{"label":"What you get","title":"Use your voice for the draft. Keep manual input for review.","subtitle":"Dictate into the app you already have open, then review and refine the returned text instead of entering every word manually.","items":[{"image":{"label":"Man speaking at a laptop beside an ergonomic keyboard"},"title":"Speak while you work","body":"Press a hotkey and dictate the email, document, chat or form in front of you. OpenWhispr pastes transcribed text into the active app when automatic pasting is available, or copies it to your clipboard."},{"image":{"label":"Woman in a wrist brace speaking at a laptop in an open studio office"},"title":"Cloud transcription without a local model download","body":"Start speaking without waiting for a local model download. Cloud transcribes with top accuracy and speed, then immediately discards the audio."},{"image":{"label":"Woman resting her hands on a cushion while working at a laptop at home"},"title":"A local option for offline work","body":"Choose a local model when you need device-only processing or no internet connection. Your computer processes the transcription while you stay in control of the workflow."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Audio sent away in local mode","value":"None","footnote":"local transcription runs on your device"},{"label":"Cost to start","value":"$$0","footnote":"unlimited local AI models, forever"},{"label":"Steps to your first dictation","value":"3","footnote":"download, dictate, review"}]},"howItWorks":{"label":"How it works","title":"A reply in the box with less sustained manual input.","subtitle":"Choose the interaction that suits the task, dictate into the app already open, then review the returned text before continuing.","image":{"label":"Woman in a wrist brace speaking at a laptop in an art studio"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Start with unlimited local AI models, or use the included Cloud allowance."},{"title":"Choose tap or hold, then speak.","body":"Use tap mode for separate start and stop presses, or hold the hotkey while speaking. Dictate the email, form answer or note in the app already open."},{"title":"Review and send.","body":"Read the text, make any edits you want, then carry on with the task."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Built for the next sentence","title":"10 reasons to try OpenWhispr with RSI","subtitle":"OpenWhispr does not treat RSI. It gives you a practical way to dictate the draft, then use the keyboard or mouse to review and edit the returned text.","items":[{"title":"A hotkey starts dictation","body":"Use a keyboard shortcut to start speaking without hunting for a voice button in every app."},{"title":"Review before you send","body":"Read the transcribed text and make any edits you want before you send or save it."},{"title":"Cloud gets you speaking sooner","body":"Choose Cloud when you want top accuracy and speed without first downloading a local model."},{"title":"Local transcription works offline","body":"Run a local AI model entirely on your device when you do not have an internet connection."},{"title":"Cloud audio is immediately discarded","body":"OpenWhispr Cloud receives audio only to transcribe it, then never stores it."},{"title":"The dictionary learns your words","body":"Add names, acronyms and specialist terms to the custom dictionary for sharper transcripts."},{"title":"100+ languages are supported","body":"Every plan supports 100+ languages."},{"title":"Local models are unlimited and free","body":"Every plan includes unlimited local AI models, so there is no per-minute local dictation bill."},{"title":"Bring your own API keys","body":"Use your own OpenAI, Anthropic, Gemini or Groq keys for unlimited cloud transcription with no OpenWhispr limits."},{"title":"Upgrade when the workflow needs more","body":"Pro adds unlimited OpenWhispr Cloud transcription, sync, mobile, and API/MCP access. Business adds Agent mode and chat over data."}]},"faq":{"label":"Before you download","title":"The questions people with RSI ask first.","subtitle":"Straight answers on voice input, privacy and cost.","items":[{"question":"Is OpenWhispr a treatment for RSI?","answer":"No. OpenWhispr is speech-to-text software, not a diagnosis or treatment. It gives you a way to use your voice for text input when that suits your workflow."},{"question":"Can I dictate into the apps I already use?","answer":"Yes. Press a hotkey and dictate the email, document, chat or form in front of you. OpenWhispr pastes transcribed text into the active app when automatic pasting is available. On Wayland without a paste tool, it is copied to your clipboard for Ctrl+V."},{"question":"Should I choose Cloud or local transcription?","answer":"Choose Cloud when you want top accuracy and speed without downloading a local model. Choose local when you need device-only processing or want to work offline. Both are deliberate options for different situations."},{"question":"Does Cloud keep my audio?","answer":"No. OpenWhispr Cloud receives audio only to transcribe it, then immediately discards it. Local transcription runs on your device and does not need an internet connection."},{"question":"Can it handle names, jargon, and drafting?","answer":"Use the custom dictionary for names, acronyms and specialist terms. OpenWhispr turns speech into text. You can also connect your own AI key to turn dictated text into a draft, then review it yourself."},{"question":"Is it free, and what do paid plans add?","answer":"Free includes unlimited local AI models, 2,000 Cloud words each week, and five hours of meeting recordings each month. Pro adds unlimited Cloud transcription, 20 meeting-recording hours, sync, mobile, and API/MCP access. Business adds unlimited meeting recordings, Agent mode, and chat over data."}]},"reviews":{"label":"What users say","title":"For people who would rather talk than type.","subtitle":"What users told us in their own words.","score":{"value":"$$0","label":"To start dictating","body":"Free includes unlimited local AI models. Choose a paid plan when Cloud transcription or workflow features become useful."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for people with RSI","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"People with RSI who want unlimited local dictation and a useful Cloud allowance"}]},{"parts":[{"text":"People with RSI who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"People who need unlimited meeting recordings, Agent mode, and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local AI models plus a limited Cloud allowance"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Prolonged keyboard and mouse work should not be the only way to get words out.","paragraphs":["OpenWhispr started with a simple objection: someone whose hands need a break should not have to stop getting words onto the page.","OpenWhispr is open source under the MIT license, so the code is available for anyone to inspect. Local AI models run entirely on your device, and you can choose them when that is the right fit for your workflow.","Free includes Community support. Pro includes Email support, and Business includes Priority support."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"When sustained manual input adds friction, let your voice take the next turn.","body":"Download OpenWhispr and dictate the next email, form or note. Review the words, make your edits, and keep the task moving.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Start with three steps: download, dictate, review. Free forever on Mac, Windows & Linux."},"byline":{"authorLabel":"Written by","author":"Joshua Padoa","aiDisclosure":"Drafted with AI assistance","reviewerLabel":"Reviewed and fact checked by","reviewer":"Wesley van der Hoop","updatedLabel":"Last reviewed","dateModified":"2026-08-27"}}},"useCaseIpad":{"breadcrumb":{"home":"Home","useCases":"Use Cases","iPad":"For iPad Users"},"metadata":{"title":"Free Speech to Text on iPad? Check the Plan - OpenWhispr","description":"Capture notes and first drafts by voice on iPad, then continue them on desktop. OpenWhispr’s iPad app and sync are included with Pro.","keywords":["best free speech to text app for ipad","free speech to text app for ipad","ipad speech to text app","ipad dictation app","voice to text for ipad","speech to text ipad"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text app for iPad users","operatingSystem":"macOS, Windows, Linux, iOS","description":"Dictate editable text on iPad and continue synced OpenWhispr notes on desktop. The iPhone & iPad app is included with Pro; Free starts on desktop.","price":"8","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text app for iPad users","title":"Say it on iPad. Finish it on your desktop.","body":"Start the follow-up while the meeting is still fresh. Pro includes the iPhone & iPad app and syncs notes and agent conversations across devices; Free starts on desktop with unlimited local AI models.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"A man speaking to an iPad he is holding at a library table"}},"trustBar":{"label":"Your everyday iPad apps:","names":["Apple Notes","Pages","Microsoft Word","Google Docs","Notion","Obsidian"]},"pain":{"label":"The catch behind the free search","title":"Capture the work while the context is still fresh.","subtitle":"A meeting follow-up, note or rough first draft should not wait until you are back at a full keyboard.","cards":[{"title":"Free should mean useful, not a three-minute trial.","body":"Free includes unlimited local AI models on macOS, Windows and Linux. It also includes 2,000 OpenWhispr Cloud transcription words each week and five hours of meeting recordings each month."},{"title":"The iPad app is not free. It is a Pro feature.","body":"OpenWhispr does have an iPhone & iPad app, but it is included with Pro. That is the answer before you spend time installing anything."},{"title":"Follow up before you are back at your desk.","body":"Start the note while the meeting is still fresh, then pick it up on desktop."},{"title":"Cloud speed needs a clear data path.","body":"OpenWhispr Cloud avoids a local model download and is built for accuracy and speed. Audio is sent only to transcribe it, then never stored."}]},"features":{"label":"A mobile-to-desktop writing workflow","title":"Speak the draft on iPad. Keep working on desktop.","subtitle":"The mobile app turns speech into editable text on iPad; OpenWhispr Notes provide the supported path for carrying that work to desktop.","items":[{"image":{"label":"A woman tapping an iPad at a kitchen table with a laptop open beside her"},"title":"A useful Free plan on desktop","body":"Use unlimited local AI models on macOS, Windows or Linux. You also get 2,000 OpenWhispr Cloud transcription words per week, five meeting-recording hours per month, 100+ languages and a custom dictionary."},{"image":{"label":"A woman speaking toward an iPad in a keyboard case on a sunny rooftop terrace"},"title":"The iPhone & iPad app with Pro","body":"Use the OpenWhispr keyboard to dictate in any text field. When the work needs to continue inside OpenWhispr, start it in an OpenWhispr Note and pick up that synced note on desktop."},{"image":{"label":"A man in a suit speaking to a propped-up iPad in a hotel lobby"},"title":"Unlimited Cloud transcription with Pro","body":"Choose Pro when you want unlimited OpenWhispr Cloud transcription without downloading a local model. Cloud audio is transcribed, then immediately discarded."}]},"stats":{"ariaLabel":"Plan facts for iPad users","items":[{"label":"Free desktop start","value":"$$0","footnote":"unlimited local AI models on macOS, Windows and Linux"},{"label":"iPhone & iPad app","value":"Pro","footnote":"mobile access is included with the Pro plan"},{"label":"Free Cloud allowance","value":"2,000","footnote":"OpenWhispr Cloud transcription words each week"}]},"howItWorks":{"label":"How it works","title":"Capture it on iPad. Finish it on desktop.","subtitle":"Use an iPad text field for immediate editable copy, or an OpenWhispr Note for work that continues inside OpenWhispr on desktop.","image":{"label":"A woman speaking toward a propped-up iPad at an outdoor cafe table with a closed laptop beside her"},"steps":[{"title":"Open the place you want to write.","body":"Use the OpenWhispr keyboard in an email, document or note. Start an OpenWhispr Note when cross-device continuation is part of the job."},{"title":"Speak the follow-up or first draft.","body":"OpenWhispr returns editable text so you can correct names, details and structure while the context is fresh."},{"title":"Continue the supported work on desktop.","body":"Open the synced OpenWhispr Note on desktop and keep editing. OpenWhispr sync covers notes and agent conversations; text inserted into another app is not added to that sync."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No hidden plan comparison","title":"10 reasons iPad users choose OpenWhispr","subtitle":"Looking for free speech-to-text on iPad deserves a straight answer. These are the limits, upgrades and data paths as the plans define them.","items":[{"title":"The iPad app is clearly a Pro feature","body":"OpenWhispr includes an iPhone & iPad app with Pro. Free is desktop-only, so there is no bait-and-switch in the plan comparison."},{"title":"Free is still a real desktop plan","body":"Free includes unlimited local AI models on macOS, Windows and Linux, plus a limited OpenWhispr Cloud allowance."},{"title":"Pro unlocks unlimited OpenWhispr Cloud transcription","body":"Use Pro when the 2,000 free Cloud words per week are not enough and you want unlimited OpenWhispr Cloud transcription."},{"title":"Pro includes sync across devices","body":"Mobile access comes with device sync, so Pro is built for people who move between their computer and iPhone or iPad."},{"title":"Cloud does not need a local model download","body":"OpenWhispr Cloud is the faster setup when accuracy and speed matter and you do not want to download local models first."},{"title":"Cloud audio is never stored","body":"Audio is sent to OpenWhispr Cloud only to transcribe it, then immediately discarded."},{"title":"Bring your own API keys on any plan","body":"Use your own OpenAI, Anthropic, Gemini or Groq keys for unlimited Cloud transcription without OpenWhispr limits."},{"title":"A custom dictionary is included","body":"Teach OpenWhispr your names, jargon and acronyms for sharper transcripts, whether you start Free or use Pro."},{"title":"100+ languages are supported","body":"Every plan supports 100+ languages, so the plan decision is about access and workflow, not a reduced language list."},{"title":"The upgrade cost is stated upfront","body":"Pro is $8 per month or $80 per year. It adds the iPhone & iPad app, unlimited OpenWhispr Cloud transcription and device sync."}]},"faq":{"label":"Before you download","title":"What iPad users need to know first.","subtitle":"The direct answer on price, mobile access, Cloud transcription and the useful free starting point.","items":[{"question":"Is OpenWhispr a free speech-to-text app for iPad?","answer":"Not on iPad. The Free plan is desktop-only. OpenWhispr's iPhone & iPad app is included with Pro, which also includes unlimited OpenWhispr Cloud transcription and sync across devices."},{"question":"What does the Free plan include?","answer":"Free includes unlimited local AI models on macOS, Windows and Linux, 2,000 OpenWhispr Cloud transcription words per week, five hours of meeting recordings per month, 100+ languages, a custom dictionary and community support."},{"question":"What does Pro add for an iPad user?","answer":"Pro adds the iPhone & iPad app, unlimited OpenWhispr Cloud transcription, 20 hours of meeting recordings per month, sync across devices, personal API access, MCP integration in all your apps and email support."},{"question":"What happens to audio sent to OpenWhispr Cloud?","answer":"Cloud audio is sent only to transcribe it, then immediately discarded."},{"question":"Can I use my own API keys instead?","answer":"Yes. Every plan supports your own OpenAI, Anthropic, Gemini or Groq keys for unlimited Cloud transcription without OpenWhispr limits."},{"question":"When should I choose Business?","answer":"Business adds unlimited meeting recordings, Agent mode, chat over your data and priority support. It is for people who need more than Pro's mobile, sync and unlimited Cloud transcription workflow."}]},"reviews":{"label":"What users say","title":"Built for people who would rather speak than type.","subtitle":"Three unedited reviews from OpenWhispr users.","score":{"value":"1","label":"clear boundary","body":"The plan for iPad access is Pro."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free starts on desktop.","Pro carries OpenWhispr from iPad to desktop."],"caption":"Compare OpenWhispr plans for iPad users","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"People who want a useful free desktop start with local AI models and limited Cloud transcription"}]},{"parts":[{"text":"People who capture notes and drafts on iPad, then continue them in OpenWhispr on desktop"}]},{"parts":[{"text":"People who also need unlimited meeting recordings, Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Useful Free desktop access, including unlimited local AI models and limited Cloud transcription"}]},{"parts":[{"text":"The iPhone & iPad app, sync for OpenWhispr notes and agent conversations, and unlimited OpenWhispr Cloud transcription"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device. It works without an internet connection and is unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"A free plan should tell you exactly where it ends.","paragraphs":["OpenWhispr gives you unlimited local AI models on desktop for free. The iPhone & iPad app is part of Pro, and that distinction belongs in the first answer, not the fine print.","We build in public. The code is open source under the MIT license, so you can inspect the product and choose the plan and data path that fit your work.","And if something breaks, ask in Discord."],"image":{"label":"OpenWhispr team and community"}},"finalCta":{"tagline":"Voice on iPad · editable text · continue on desktop","title":"Start the work on iPad. Finish it on desktop.","body":"Capture the follow-up, note or rough first draft while it is fresh. Use an OpenWhispr Note when you want that work to sync to desktop; iPad access and sync are included with Pro.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free includes unlimited local AI models on macOS, Windows and Linux. The iPhone & iPad app is included with Pro."},"byline":{"authorLabel":"Written by","author":"Joshua Padoa","aiDisclosure":"Drafted with AI assistance","reviewerLabel":"Reviewed and fact checked by","reviewer":"Wesley van der Hoop","updatedLabel":"Last reviewed","dateModified":"2026-08-27"}}},"useCaseArthritis":{"breadcrumb":{"home":"Home","useCases":"Use Cases","arthritis":"For Arthritis"},"metadata":{"title":"Free Dictation App for Arthritis - OpenWhispr","description":"OpenWhispr gives people with arthritis free local dictation, or Cloud transcription for speed without a model download.","keywords":["best dictation apps for arthritis","dictation software for arthritis","voice typing for arthritis","speech to text for arthritis","arthritis voice recognition software","hands-free typing for arthritis"]},"schema":{"name":"OpenWhispr","applicationCategory":"Dictation software for arthritis","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source dictation software for people with arthritis. Choose Cloud for accuracy and speed without a model download, or local processing for offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation software for arthritis","title":"Arthritis makes typing harder. Say the reply. Get text to review.","body":"When you are comparing dictation apps for arthritis, check how much keyboard work the app leaves you. OpenWhispr offers free local dictation for offline work, or Cloud transcription for top accuracy and speed without a model download.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Older woman with grey hair speaking at an open laptop, hands resting flat on the desk, in a bright home study"}},"trustBar":{"label":"Dictate into:","names":["Microsoft Word","Google Docs","Gmail","Slack","Notion","Microsoft Teams"]},"pain":{"label":"When the keyboard hurts","title":"When an email takes more out of you than it should.","subtitle":"An email, a form, a document, a message. Arthritis can make every keystroke a decision. Dictation lets your voice do the typing.","cards":[{"title":"A short email takes too long.","body":"When typing hurts, even a few sentences can turn into a stop-start job. Speak the reply, then paste the transcription into the app you already use."},{"title":"Holding a key is uncomfortable.","body":"Use toggle mode: press once to start dictating and once to stop. You do not need to keep a key held down while you speak."},{"title":"Corrections mean more typing.","body":"A custom dictionary lets you add names, places and terms you use often, so there is less to fix after you dictate."},{"title":"A bad connection should not end the sentence.","body":"Local models run on your device without internet. Dictate at home, in a waiting room or wherever the connection drops out."}]},"features":{"label":"What changes","title":"Speak the words. Keep your hands for the rest.","subtitle":"Choose the route that fits the moment: Cloud for fast, accurate transcription without a model download, or local processing when you need to work offline.","items":[{"image":{"label":"Older man in a green polo shirt speaking at a desktop monitor, one hand resting on a large round desk button, in a bright open-plan office"},"title":"Start and stop without holding a key","body":"Set dictation to toggle mode when press-and-hold is not comfortable. One press starts listening. The next press stops it."},{"image":{"label":"Older woman in a blue patterned blouse speaking at an open laptop, forearms resting on a cushion, at a community centre table"},"title":"Put words where you already work","body":"Dictate into a document, email, chat or web form. OpenWhispr copies the transcription so you can paste it where you are already working."},{"image":{"label":"Woman in a green shirt speaking into a phone held near her mouth, laptop and open notebook on a wooden table in a flower shop"},"title":"Choose speed or offline work","body":"Cloud gives you top accuracy and speed with no local model to download. Local models keep working on your device when internet is not available."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Words you need to type to dictate","value":"None","footnote":"speak into the app and review the text when you are ready"},{"label":"Cost for local dictation","value":"$$0","footnote":"unlimited local AI models on every plan"},{"label":"Languages supported","value":"100+","footnote":"available on every plan"}]},"howItWorks":{"label":"How it works","title":"Download it. Speak the email. Send it.","subtitle":"Start with the free desktop app. Choose a local model for offline work, or use Cloud when you want accuracy and speed without downloading a model.","image":{"label":"Woman in a light blue top speaking at a laptop, one hand on folded fabric beside a sewing machine in a sunlit sewing room"},"steps":[{"title":"Download for free.","body":"Install OpenWhispr on Mac, Windows or Linux. Pick a local model during setup, or begin with the included Cloud allowance."},{"title":"Dictate instead of typing.","body":"Choose push-to-talk or toggle mode, then speak into the document, email or message you already have open."},{"title":"Review and send.","body":"Make any edits you want, then send the text as usual. Your keyboard is there when you need it, not for every word."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"The practical reasons","title":"10 practical reasons to try OpenWhispr for arthritis","subtitle":"The point is not another place to write. It is fewer keystrokes in the apps you already use.","items":[{"title":"Toggle mode does not ask you to hold a key","body":"Press once to start dictating and once to stop. Use it when push-to-talk is not comfortable."},{"title":"It works in the apps you already have open","body":"Dictate into documents, email, chat and web forms without moving your work into a separate editor."},{"title":"Local dictation is free","body":"Every plan includes unlimited local AI models. There is no subscription required to keep using local dictation."},{"title":"Cloud is ready when setup needs to be simple","body":"Cloud transcription is built for top accuracy and speed, with no local model download. Free includes 2,000 words each week."},{"title":"Local models work without internet","body":"Run transcription on your device when Wi-Fi is unreliable or unavailable."},{"title":"A custom dictionary cuts repeat corrections","body":"Add names, places, acronyms and terms you use often to help produce sharper transcripts."},{"title":"100+ languages supported","body":"Dictate in the language you use at home or at work."},{"title":"Cloud audio is discarded after transcription","body":"Audio sent to OpenWhispr Cloud is used to transcribe, then immediately discarded."},{"title":"The app is open source","body":"The code is available under the MIT licence for anyone to inspect."},{"title":"Paid plans remove Cloud limits and add workflow tools","body":"Pro adds unlimited Cloud transcription, sync, the iPhone and iPad app, and API/MCP access. Business adds unlimited meeting recordings, Agent mode and chat over your data."}]},"faq":{"label":"Before you download","title":"Questions people with arthritis ask first.","subtitle":"Straight answers about typing less, setup, cost and privacy.","items":[{"question":"Will I need to hold down a key while I dictate?","answer":"No. OpenWhispr supports toggle mode: press once to start dictating and once to stop. You can choose that when holding a key is uncomfortable."},{"question":"Can I dictate into the apps I already use?","answer":"Yes. Put your cursor in a document, email, chat or web form, then dictate. OpenWhispr copies the transcription so you can paste it wherever you can type."},{"question":"Should I choose Cloud transcription or a local model?","answer":"Choose Cloud when you want top accuracy and speed without downloading a model. Choose a local model when you need to work without internet."},{"question":"Does audio leave my device?","answer":"With a local model, transcription runs on your device. With OpenWhispr Cloud, audio is sent only to transcribe it, then immediately discarded."},{"question":"Can it recognise names and words I use often?","answer":"Yes. Add names, acronyms and other recurring terms to the custom dictionary to help produce sharper transcripts."},{"question":"Is it free, or do I need a paid plan?","answer":"Free includes unlimited local AI models, 2,000 Cloud transcription words each week and 5 hours of meeting recordings each month. Pro adds unlimited Cloud transcription, sync, the iPhone and iPad app, and API/MCP access. Business adds unlimited meeting recordings, Agent mode and chat over your data."}]},"reviews":{"label":"What users say","title":"Built for people who would rather speak than keep typing.","subtitle":"Existing quotes carried forward from the reference page.","score":{"value":"$$0","label":"Cost to try local dictation","body":"Every plan includes unlimited local AI models. You can see whether voice input suits your routine before paying for more Cloud transcription or workflow features."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for arthritis dictation","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"People who want free local dictation and a limited Cloud allowance"}]},{"parts":[{"text":"People who want unlimited Cloud transcription, plus sync, the iPhone and iPad app, and API/MCP access"}]},{"parts":[{"text":"People and teams who want unlimited meeting recordings, Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local dictation, plus 2,000 Cloud words each week"}]},{"parts":[{"text":"Unlimited Cloud transcription without managing local models or your own API keys"}]},{"parts":[{"text":"Unlimited meeting recordings plus advanced workflow features"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Typing should not be the price of getting words out.","paragraphs":["OpenWhispr started with a simple objection: people who need to type less should not have to give up the apps and documents they already rely on.","OpenWhispr is open source, so you can inspect the code before you install it. Local models run on your device. Cloud audio is transcribed, then immediately discarded.","Free includes community support. Pro adds email support, and Business adds priority support."],"image":{"label":"Team and community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Let your voice carry more of the typing.","body":"Download OpenWhispr, choose Cloud transcription or a local model, and dictate the next email or document.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free on macOS, Windows and Linux. Paid plans add unlimited Cloud transcription, the iPhone and iPad app, sync and workflow tools."}}},"useCaseIphone":{"breadcrumb":{"home":"Home","useCases":"Use Cases","iphone":"For iPhone"},"metadata":{"title":"Dictation App for iPhone: Speak Notes Fast - OpenWhispr","description":"OpenWhispr is a dictation app for iPhone. Speak longer messages instead of thumb-typing, refine the text by touch, and send it from the app you are using.","keywords":["dictation app for iphone","iphone dictation app","voice to text iphone","iphone speech to text","dictation app ios","voice notes iphone"]},"schema":{"name":"OpenWhispr","applicationCategory":"Dictation app for iPhone","operatingSystem":"macOS, Windows, Linux, iOS","description":"OpenWhispr is a dictation app for iPhone with free, unlimited on-device transcription. Pro adds unlimited Cloud transcription, 20 meeting-recording hours per month, and sync across devices.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation app for iPhone","title":"Say more. Type less on your phone.","body":"Your phone is where conversations and ideas already happen. OpenWhispr turns speech into text on iPhone, so you can capture longer thoughts without typing every word, then refine and send them in the app you are already using.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman walking through a city plaza, speaking into the phone in her hand"}},"trustBar":{"label":"Use OpenWhispr on:","names":["iPhone","iPad","Mac","Windows","Linux","OpenWhispr Cloud"]},"pain":{"label":"Why thumb-typing gets in the way","title":"You have more to say than your thumbs want to type.","subtitle":"Long messages, replies, and updates are often easier to say. The hard part is getting those words into the app where the conversation is already happening.","cards":[{"title":"The conversation is already on your phone.","body":"Messages, email, and work threads happen on the device in your hand. Moving to a laptop just to type a longer response breaks the flow."},{"title":"A voice note still makes someone else listen.","body":"Communication moves faster when speech becomes text you can review, refine, and send in the same app."},{"title":"Switching devices should not mean starting over.","body":"If a message turns into longer work, Pro sync lets you continue an OpenWhispr note on the device that suits the job."},{"title":"Mobile access starts free.","body":"OpenWhispr is free to install on iPhone and iPad. Unlimited on-device transcription needs no account and has no usage limit; Pro is optional."}]},"features":{"label":"What you get on iPhone","title":"Speak in the app where you already write.","subtitle":"Use OpenWhispr on iPhone and iPad for free with unlimited on-device transcription, or add Pro for unlimited Cloud transcription, 20 meeting-recording hours per month, and sync across devices.","items":[{"image":{"label":"Man sitting on stone steps outside, speaking into his phone with a backpack beside him"},"title":"Direct insertion across iPhone apps","body":"Use the OpenWhispr keyboard to turn speech into text directly in the iPhone apps where you write."},{"image":{"label":"Clinician in a white coat sitting outdoors, reading her phone with a laptop in the bag beside her"},"title":"Sync across devices","body":"When a message turns into longer work, Pro sync lets you continue an OpenWhispr note on the device that suits the job."},{"image":{"label":"Traveller in an airport lounge armchair, speaking into his phone by the window"},"title":"Unlimited Cloud transcription with Pro","body":"Pro includes unlimited OpenWhispr Cloud transcription. Cloud is the fast, no-local-model-download option, and transcription audio is never stored."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"iPhone and iPad app","value":"Free","footnote":"free to install"},{"label":"On-device transcription limit","value":"None","footnote":"no account or usage limit"},{"label":"US App Store Pro price","value":"$$9.99","footnote":"$$99.99/year; website checkout pricing differs"}]},"howItWorks":{"label":"How it works","title":"Speak the long version. Edit the details.","subtitle":"When something is easier to say than type, use OpenWhispr on iPhone to turn it into text. Review the wording, then continue across your synced devices.","image":{"label":"Woman standing in an empty sunlit apartment holding keys, speaking into her phone"},"steps":[{"title":"Open the app where you want to write.","body":"In a message, email, document, or other iPhone text field, switch to the OpenWhispr keyboard."},{"title":"Say the long version.","body":"Tap the microphone and speak in context instead of typing every word with your thumbs."},{"title":"Speak. Refine by touch. Continue.","body":"Turn speech into text in the iPhone app where you write. For an OpenWhispr note, review the wording by touch, then continue across synced devices."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"The practical reasons","title":"10 things to know before choosing OpenWhispr on iPhone","subtitle":"Start with the plan and device that fit the work. These are the current product facts behind the mobile option.","items":[{"title":"The iPhone app is free to install","body":"Use unlimited on-device transcription without an account or usage limit."},{"title":"Your devices can stay in sync","body":"Pro keeps OpenWhispr notes available across devices when a message turns into longer work."},{"title":"Cloud transcription is unlimited on Pro","body":"Pro includes unlimited OpenWhispr Cloud transcription."},{"title":"Cloud does not need a local model download","body":"OpenWhispr Cloud is the no-local-model-download option when speed and simple setup matter."},{"title":"Cloud transcription audio is never stored","body":"Audio sent to OpenWhispr Cloud is transcribed, then immediately discarded."},{"title":"On-device transcription stays free on iPhone","body":"Private mode runs on your iPhone; on-device transcription is unlimited and does not require an account."},{"title":"Custom dictionary is included","body":"Add names, jargon, and acronyms to improve the words you use most."},{"title":"100+ languages are supported","body":"OpenWhispr supports 100+ languages on every plan."},{"title":"App Store pricing is channel-specific","body":"In the US App Store, Pro is $9.99 a month or $99.99 a year; website checkout pricing is separate."},{"title":"The upgrade has a defined job","body":"Pro adds unlimited Cloud transcription, 20 hours of meeting recording per month, and sync across devices."}]},"faq":{"label":"Before you choose a plan","title":"The questions iPhone users ask first.","subtitle":"Straight answers on free iPhone access, on-device and Cloud transcription, and the optional Pro upgrade.","items":[{"question":"Is there an OpenWhispr app for iPhone?","answer":"Yes. The iPhone and iPad app is free to install, and unlimited on-device transcription works without an account or usage limit."},{"question":"Do I need Pro to use OpenWhispr on iPhone?","answer":"No. Pro is optional. It adds unlimited Cloud transcription, 20 hours of meeting recording per month, and sync across devices."},{"question":"What does the free plan include?","answer":"Free includes 2,000 words a week of OpenWhispr Cloud transcription, five hours of meeting recordings a month, unlimited on-device transcription without an account or usage limit, 100+ languages, a custom dictionary, and zero data retention."},{"question":"Does OpenWhispr Cloud keep my audio?","answer":"No. Audio sent to OpenWhispr Cloud is used to transcribe, then immediately discarded. Cloud sync and transcription-audio retention are separate choices."},{"question":"When should I use on-device transcription?","answer":"Use on-device transcription on iPhone when you want offline, device-only processing. It is free, unlimited, and does not require an account."},{"question":"Why upgrade to Business?","answer":"Business adds unlimited meeting recordings, Agent mode, chat over your data, and priority support on top of Pro."}]},"reviews":{"label":"What users say","title":"For people who would rather speak than stop moving.","subtitle":"Three unedited quotes on local models, travel, and getting clean text back.","score":{"value":"Free","label":"Unlimited on-device transcription","body":"The iPhone and iPad app is free to install. Pro is optional and adds unlimited Cloud transcription, 20 meeting-recording hours per month, and sync across devices."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Website pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr website plans for iPhone users","tiers":[{"id":"free","name":"Free","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"iPhone users who want unlimited on-device transcription without an account"}]},{"parts":[{"text":"iPhone and iPad users who want unlimited Cloud transcription and sync"}]},{"parts":[{"text":"Teams that need unlimited meeting recordings, Agent mode, and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited on-device iPhone transcription plus limited OpenWhispr Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription, 20 meeting hours, and sync"}]},{"parts":[{"text":"Unlimited meeting recordings, Agent mode, and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Easy: install the iPhone app, download a model, and use on-device transcription without an account"}]},{"parts":[{"text":"Optional upgrade for unlimited cloud speech-to-text and sync"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Mobile app","body":" free on-device dictation on iPhone and iPad"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile app","body":" free on-device dictation on iPhone and iPad"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile app","body":" free on-device dictation on iPhone and iPad"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Talk to us about Enterprise features for your team."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Having more to say should not mean more thumb-typing.","paragraphs":["OpenWhispr is built for the gap between having something to say and finding a keyboard. The iPhone and iPad app is free to install, with unlimited on-device transcription and optional Pro for Cloud transcription and sync.","We build in public. The desktop code is on GitHub, so you can inspect the local models, Cloud option, and the choices behind the plans before you make one.","Free includes community support."],"image":{"label":"OpenWhispr team and community photo"}},"finalCta":{"tagline":"Voice first · touch to refine","title":"Say it on your phone. Send it as text.","body":"Use the OpenWhispr keyboard in the iPhone app where you are already writing. Speak the long version, refine the text by touch, then send it or continue in a synced OpenWhispr note.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Unlimited on-device transcription is free without an account. US App Store Pro is $9.99 a month or $99.99 a year; website checkout pricing is shown above."}}},"careersHub":{"metadata":{"title":"Careers at OpenWhispr: Help Build the Voice Interface for AI","description":"Join a small team building open-source voice AI. We're hiring a Founding Engineering Lead, a Founding Engineer and a Social & Brand Growth Creator, in person in San Francisco or Chicago.","keywords":["OpenWhispr careers","founding engineer jobs","founding engineering lead jobs","AI startup jobs San Francisco","AI startup jobs Chicago","voice AI jobs"]},"breadcrumb":{"home":"Home","careers":"Careers"},"hero":{"eyebrow":"Careers","title":"Help build the voice interface for AI","subtitle":"OpenWhispr turns what you say into action. It's open source, it's been downloaded about 150,000 times, and it's built for people who want a choice: run it locally, bring your own keys, or use our cloud. We're hiring people who want to own big problems end to end."},"roles":{"title":"Open roles","subtitle":"All roles are in person, in San Francisco or Chicago. Visa sponsorship is not available.","viewRole":"View role","items":{"foundingEngineeringLead":{"title":"Founding Engineering Lead","blurb":"A player-coach role. Lead a small team of engineers, set the technical direction, and stay in the code every day.","location":"San Francisco or Chicago","type":"Full-time","compensation":"$$160K–$200K + meaningful equity"},"foundingEngineer":{"title":"Founding Engineer","blurb":"Own big parts of the product, from first idea to production, and help set the engineering bar.","location":"San Francisco or Chicago","type":"Full-time","compensation":"$$140K–$180K + meaningful equity"},"brandGrowthCreator":{"title":"Social & Brand Growth Creator","blurb":"Be the face, voice and storyteller behind OpenWhispr. Build our brand story in public on TikTok and Reels.","location":"San Francisco or Chicago","type":"Full-time","compensation":"$$60K–$120K + meaningful equity"}}},"principles":{"label":"How we work","title":"Our operating principles","subtitle":"These aren't wall posters. They're how we make decisions every day.","items":[{"title":"Bias for action","body":"If something's broken, fix it. If you have an idea, try it. We don't wait for issue reports or perfect specs."},{"title":"Proactive communication","body":"Tell people before they have to ask. That goes for teammates and for users: if we see a problem coming, we reach out first, and if we get a bill wrong, we refund it before anyone complains."},{"title":"No AI slop","body":"AI gives us leverage, not an excuse. Everything we ship gets reviewed, tested and refined, and anything we share with each other is already clear."},{"title":"Context is the bottleneck","body":"We run a lot of workstreams in parallel. The core skill is managing context: giving it, keeping it, and switching between streams without dropping quality."},{"title":"Do the thinking for the user","body":"Simplification is the strategy. Every screen, sentence and default should take work off the user's plate. Opinionated by design, and still built for the tinkerer."},{"title":"Grateful for every user","body":"People caring enough to complain is a privilege. We use the product every day, and we treat taste as something you train on purpose."}]},"closing":{"title":"Don't see your role?","body":"If you think you'd make us better, write to us anyway and show us something you've built.","ctaLabel":"Email us"}},"careerFoundingEngineer":{"metadata":{"title":"Founding Engineer at OpenWhispr","description":"Own big parts of an open-source voice AI product, from first idea to production. In person in San Francisco or Chicago. $140K to $180K plus meaningful equity. Visa sponsorship not available.","keywords":["founding engineer","OpenWhispr jobs","voice AI engineer","AI-native engineer","open source startup jobs","startup engineering jobs San Francisco","startup engineering jobs Chicago"]},"breadcrumb":{"home":"Home","careers":"Careers","role":"Founding Engineer"},"posting":{"hero":{"eyebrow":"Careers","title":"Founding Engineer","summary":"Own big parts of the product, from first idea to production, and help decide what engineering looks like at OpenWhispr.","meta":{"location":"San Francisco or Chicago","employmentType":"Full-time","workplace":"In person","visa":"No visa sponsorship","compensation":"$$140K–$180K + meaningful equity"}},"intro":["Hi, I'm Gabe, the founder of OpenWhispr.","I think voice is about to become one of the main ways people work with AI. Not talking to a chatbot, but saying what you actually mean and having software turn it into action. For that to work, people need to trust it with what they say. So we build in the open, and we give people a real choice: run everything locally on your own machine, bring your own API keys, or use our cloud when you want the best accuracy and speed. We're not anti-cloud. We just think the choice should be yours.","The other thing we believe is that a great consumer product has to be opinionated in its design, and that we should still build for the tinkerer. You can edit your prompts, pick your models, and customise your own voice pipeline. Holding both of those at once is most of the challenge, and most of the fun.","I built the first version on my own. It's open source, it's been downloaded about 150,000 times, and the people who use it tell us loudly when something breaks. That's a privilege. Most products never get anyone to care that much. Now I'm hiring the engineers who'll own big parts of it with me."],"sections":[{"title":"What you'd work on","paragraphs":["You take large, fuzzy problems and drive them all the way to production. No tickets, and no one writing the spec for you. You'll use the product every day, talk to the people who use it, and fix what you find before anyone files an issue.","The engineering problems are genuinely interesting. On our cloud, it's how we get the best accuracy across every language with the fastest inference. On device, it's how we make local speech-to-text and local language models run well on whatever Mac, Windows or Linux machine someone has. In between, it's how we let people bring their own keys without the product getting complicated. And underneath all of it, it's shaving latency out of the voice pipeline while the experience stays simple and beautiful.","Right now that looks like offline meeting recording and speaker identification, the local model pipeline, speech to action inside whatever app you're in, and the calendar and email integrations."]},{"title":"How we work","paragraphs":["We have a strong bias for action. If something's broken, fix it. If a user is about to hit a problem, tell them before they find it. If we get a bill wrong, refund it before anyone complains. The same goes for each other: share what you know before someone has to ask, and flag a blocker the moment it appears.","We lead with context. You get the full picture, then the ownership and agency to go and deliver it with the team. Everyone runs several AI-assisted workstreams at once, and the skill that matters most is keeping context and quality across all of them. No AI slop. Everything gets reviewed, tested and refined before it ships, and anything you share with the team is already clear by the time it lands.","Getting to 90% is easy now. We spend our energy on the last 10%, and we do the thinking for the user so they don't have to."]},{"title":"You'll do well here if","bullets":[{"body":"You've shipped something real that people actually used: a company, a product, a side project with users. You can tell us what you started, what broke, and what you did about it."},{"body":"You already run parallel AI workstreams and can walk us through exactly how you keep quality high while you do it. Not “I use Cursor.” The actual system."},{"body":"You have taste. You notice when something works but doesn't feel right, and you'd rather simplify than add."},{"body":"You communicate before people have to ask, and you write things down."},{"body":"You've got around five years of real engineering behind you. Less is fine if the evidence is unusual."}]},{"title":"Nice to have","paragraphs":["None of these are required, but they help."],"bullets":[{"body":"You've worked at a VC-backed startup before: a YC company, a Series A or B, a16z Speedrun, that kind of environment. If not, you at least know you want a high-ownership, high-velocity role, and you can say why."},{"body":"You've worked on voice in some form: voice AI, speech-to-text, audio pipelines, real-time systems."},{"body":"Desktop apps, open-source work people actually use, or AI agents, evals and model infrastructure."}]},{"title":"The honest bits","paragraphs":["This is in person in San Francisco or Chicago. Not remote. If you're not in one of those cities you'd need to move, and we can't sponsor visas.","The pace is real and our users complain loudly. If that sounds energising rather than exhausting, keep reading."]},{"title":"First 90 days","paragraphs":["In your first 30 days you'll ship a real product surface end to end. By 90 you'll own a major part of the product, and people will be asking you how it should work."]},{"title":"Comp and time off","paragraphs":["$$140K to $180K plus meaningful equity, and unlimited PTO. We'll make sure you actually take it."]},{"title":"How it goes","paragraphs":["A call with me. Then a deep dive on something you've built. Then a day working together in person on a real problem. We move quickly, and we'll tell you where you stand."]}],"apply":{"title":"Apply","body":"Send a few lines to careers@openwhispr.com about the most impressive thing you've built, ideally something nobody asked you to build, plus links. GitHub, demos and weird side projects beat a polished resume.","ctaLabel":"Apply by email","note":"No cover letter needed."}}},"careerFoundingEngineeringLead":{"metadata":{"title":"Founding Engineering Lead at OpenWhispr","description":"A player-coach role: lead a small team of engineers, set the technical direction, and stay in the code every day. In person in San Francisco or Chicago. $160K to $200K plus meaningful equity. Visa sponsorship not available.","keywords":["founding engineering lead","engineering lead jobs","OpenWhispr jobs","voice AI engineer","hands-on engineering leader","open source startup jobs","startup engineering jobs San Francisco","startup engineering jobs Chicago"]},"breadcrumb":{"home":"Home","careers":"Careers","role":"Founding Engineering Lead"},"posting":{"hero":{"eyebrow":"Careers","title":"Founding Engineering Lead","summary":"A player-coach role. Lead a small team of engineers, set the technical direction, and stay in the code every day.","meta":{"location":"San Francisco or Chicago","employmentType":"Full-time","workplace":"In person","visa":"No visa sponsorship","compensation":"$$160K–$200K + meaningful equity"}},"intro":["Hi, I'm Gabe, the founder of OpenWhispr.","I think voice is about to become one of the main ways people work with AI. Not talking to a chatbot, but saying what you actually mean and having software turn it into action. For that to work, people need to trust it with what they say. So we build in the open, and we give people a real choice: run everything locally on your own machine, bring your own API keys, or use our cloud when you want the best accuracy and speed. We're not anti-cloud. We just think the choice should be yours.","The other thing we believe is that a great consumer product has to be opinionated in its design, and that we should still build for the tinkerer. You can edit your prompts, pick your models, and customise your own voice pipeline. Holding both of those at once is most of the challenge, and most of the fun.","I built the first version on my own. It's open source, it's been downloaded about 150,000 times, and the people who use it tell us loudly when something breaks. That's a privilege. Most products never get anyone to care that much. Now I'm hiring the engineers who'll own big parts of it with me, and I need someone to lead them."],"sections":[{"title":"What you'd work on","paragraphs":["This is a player-coach role. You'll spend most of your time in the code, and you'll lead the engineers around you: setting direction, unblocking, reviewing, raising the bar.","You'll own the architecture and the tradeoffs, keep the system simple as it grows, and define how we build: how we review, how we share context, how we keep AI leverage from turning into slop. You'll help me hire the engineers who come after you, and you'll report to me.","The engineering problems are genuinely interesting. On our cloud, it's how we get the best accuracy across every language with the fastest inference. On device, it's how we make local speech-to-text and local language models run well on whatever Mac, Windows or Linux machine someone has. In between, it's how we let people bring their own keys without the product getting complicated. And underneath all of it, it's shaving latency out of the voice pipeline while the experience stays simple and beautiful.","Right now that looks like offline meeting recording and speaker identification, the local model pipeline, speech to action inside whatever app you're in, and the calendar and email integrations."]},{"title":"How we work","paragraphs":["We have a strong bias for action. If something's broken, fix it. If a user is about to hit a problem, tell them before they find it. If we get a bill wrong, refund it before anyone complains. The same goes for each other: share what you know before someone has to ask, and flag a blocker the moment it appears.","We lead with context. You get the full picture, then the ownership and agency to go and deliver it with the team. Everyone runs several AI-assisted workstreams at once, and the skill that matters most is keeping context and quality across all of them. No AI slop. Everything gets reviewed, tested and refined before it ships, and anything you share with the team is already clear by the time it lands.","Getting to 90% is easy now. We spend our energy on the last 10%, and we do the thinking for the user so they don't have to."]},{"title":"You'll do well here if","bullets":[{"body":"You've led a small team of engineers before, set direction, reviewed their work and grown them, while still shipping meaningful code yourself. You took it on because it multiplied what the team could ship, not because you wanted to stop building."},{"body":"You've shipped something real that people actually used: a company, a product, a side project with users. You can tell us what you started, what broke, and what you did about it."},{"body":"You already run parallel AI workstreams and can walk us through exactly how you keep quality high while you do it. Not “I use Cursor.” The actual system."},{"body":"You have taste. You notice when something works but doesn't feel right, and you'd rather simplify than add."},{"body":"You communicate before people have to ask, and you write things down."},{"body":"You've got around five years of real engineering behind you. Less is fine if the evidence is unusual."}]},{"title":"Nice to have","paragraphs":["None of these are required, but they help."],"bullets":[{"body":"You've worked at a VC-backed startup before: a YC company, a Series A or B, a16z Speedrun, that kind of environment. If not, you at least know you want a high-ownership, high-velocity role, and you can say why."},{"body":"You've worked on voice in some form: voice AI, speech-to-text, audio pipelines, real-time systems."},{"body":"Desktop apps, open-source work people actually use, or AI agents, evals and model infrastructure."}]},{"title":"The honest bits","paragraphs":["This is in person in San Francisco or Chicago. Not remote. If you're not in one of those cities you'd need to move, and we can't sponsor visas.","The pace is real and our users complain loudly. If that sounds energising rather than exhausting, keep reading."]},{"title":"First 90 days","paragraphs":["In your first 30 days you'll ship something real end to end and have a clear view of how we build. By 90 you'll own the technical direction and the practices the team runs on."]},{"title":"Comp and time off","paragraphs":["$$160K to $200K plus meaningful equity, and unlimited PTO. We'll make sure you actually take it."]},{"title":"How it goes","paragraphs":["A call with me. Then a deep dive on something you've built. Then a day working together in person on a real problem. We move quickly, and we'll tell you where you stand."]}],"apply":{"title":"Apply","body":"Send a few lines to careers@openwhispr.com about the most impressive thing you've built, ideally something nobody asked you to build, plus links. GitHub, demos and weird side projects beat a polished resume.","ctaLabel":"Apply by email","note":"No cover letter needed."}}},"careerBrandGrowthCreator":{"metadata":{"title":"Social & Brand Growth Creator at OpenWhispr","description":"Be the face, voice and storyteller behind OpenWhispr on TikTok and Instagram Reels, and build our brand story in public. Full-time, in person in San Francisco or Chicago. $60K to $120K plus meaningful equity.","keywords":["social media creator jobs","brand growth creator","content creator startup job","TikTok creator job San Francisco","Instagram Reels creator job Chicago","OpenWhispr jobs","build in public"]},"breadcrumb":{"home":"Home","careers":"Careers","role":"Social & Brand Growth Creator"},"posting":{"hero":{"eyebrow":"Careers","title":"Social & Brand Growth Creator","summary":"Be the face, voice and storyteller behind OpenWhispr, and build our brand story in public.","meta":{"location":"San Francisco or Chicago","employmentType":"Full-time","workplace":"In person","compensation":"$$60K–$120K + meaningful equity","visa":"No visa sponsorship"}},"intro":["Hi, I'm Gabe, the founder of OpenWhispr.","OpenWhispr turns what you say into action. It's an open-source voice product, it's been downloaded about 150,000 times, and the people who use it care enough to tell us loudly when something breaks. We build in the open, and we give people a real choice about how their voice gets processed. That's a story worth telling, and right now nobody is telling it.","If you've spent any time on Instagram or TikTok this year, you've probably seen what Kinso has done: two founders building their company in public, with the arguments, the prototype reviews, the investor calls and the small daily wins all on camera, and a brand people actually root for. That's the kind of narrative I want us to build. Not ads. A story."],"sections":[{"title":"About the role","paragraphs":["You'll be the face, voice and storyteller behind OpenWhispr on TikTok and Instagram Reels. You'll build our presence the way a brand team would, with strategy, consistency and real rigour about what performs, but framed around a personal story: your day-to-day here, the product being built around you, and your point of view on voice, AI and privacy.","The goal is a loyal, engaged following that drives awareness, affinity and, in the end, demand for OpenWhispr. You'll work directly with me and the team, shooting and editing short-form content every week."]},{"title":"What you'll do","bullets":[{"body":"Build and grow OpenWhispr's presence on TikTok and Instagram Reels, positioned as a real person on the team, not a brand account."},{"body":"Develop an on-camera persona, voice and content universe that feels authentically you, with OpenWhispr woven in naturally."},{"body":"Tell the story of building the company: the product decisions, the launches, the bugs, the users, the office. Some of it scripted, most of it not."},{"body":"Shoot and edit new creative every week, used across organic and paid channels."},{"body":"Study what's working: hooks, like-to-view ratio, retention. Then make the next one better."}]},{"title":"What we're looking for","bullets":[{"body":"You've grown a personal, creator-led or brand account before, with follower and engagement growth on TikTok or Reels that you can show us."},{"body":"You're comfortable and charismatic on camera. You know which moments to spotlight, how to hook someone in the first three seconds, and how to leave them wanting the next post."},{"body":"You have taste. You can make a hook that pulls in a wide audience while keeping the brand elevated."},{"body":"You can edit. CapCut, Premiere, Final Cut, After Effects, whatever you use, you know how to tell a story through the cut."},{"body":"You're TikTok and Instagram native. You spend hours a day noticing trends, hooks, CTAs and brand style."}]},{"title":"Nice to have","paragraphs":["None of these are required, but they help."],"bullets":[{"body":"You already film your own life most days."},{"body":"You've made graphics, animations or illustrations."},{"body":"You're the friend who always has a story and loves to overshare."},{"body":"You've worked at an early-stage startup, or you know you want to."}]},{"title":"The honest bits","paragraphs":["This is in person in San Francisco or Chicago. Not remote. If you're not in one of those cities you'd need to move, and we can't sponsor visas.","We're a small team that moves fast. You'll be building this from zero, and a lot of what you make in the first month will be about finding out what works for us."]},{"title":"Comp and time off","paragraphs":["$$60K to $120K plus meaningful equity, depending on experience, and unlimited PTO. We'll make sure you actually take it."]},{"title":"How it goes","paragraphs":["A call with me. Then a look at what you've made and grown. Then a day with us in person, shooting something real. We move quickly, and we'll tell you where you stand."]}],"apply":{"title":"Apply","body":"Make a video under 60 seconds telling us why you're the person to tell OpenWhispr's story. Sell us on yourself the way you'd sell a product to a viewer. Post it publicly or send us the file, and email the link to careers@openwhispr.com along with the accounts you've grown.","ctaLabel":"Apply by email","note":"Energy and editing over polish."}}},"useCaseLaptop":{"breadcrumb":{"home":"Home","useCases":"Use Cases","laptop":"For Laptop Users"},"metadata":{"title":"Speech to Text App for Laptop - OpenWhispr","description":"OpenWhispr turns laptop speech into text at your active cursor, with Cloud or local transcription and a free local option.","keywords":["speech to text app for laptop","laptop speech to text","voice typing app for laptop","dictation software for laptop","speech to text for Mac and Windows","offline laptop dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Laptop speech-to-text app","operatingSystem":"macOS, Windows, Linux","description":"Open-source laptop speech-to-text software for live dictation at the active cursor. Choose Cloud for speed without a local model download, or local for sensitive and offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech to text app for laptop users","title":"Speak at the cursor. Get your words written.","body":"Press one hotkey to dictate into the laptop app you already have open, then review and correct the text where it lands. Choose Cloud for quick setup or local for sensitive and offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman sitting on a green sofa with a laptop on her lap."}},"trustBar":{"label":"Write in:","names":["Gmail","Slack","Microsoft Word","Google Docs","Notion","Microsoft Teams"]},"pain":{"label":"What breaks laptop voice typing","title":"The thought is ready. The keyboard is the delay.","subtitle":"You need the text in the field you are using, a familiar way to start, and a quick way to correct it.","cards":[{"title":"The cursor is waiting in another app.","body":"A separate dictation window turns one sentence into a copy-and-paste job. OpenWhispr starts from a global hotkey and automatically pastes text into the app you are already using."},{"title":"Corrections pull you back to the keyboard.","body":"Names, punctuation and a changed sentence can turn voice typing into cleanup. Keep the draft at the cursor, review it there, and use your dictionary for words you repeat."},{"title":"Starting should not take you away from the draft.","body":"Set one global hotkey, then start each dictation from the email, message, note or document field already in front of you."},{"title":"Typing all day adds up.","body":"When your hands need a break or a thought is moving quickly, speak the first pass and spend your effort reviewing the words instead."}]},"features":{"label":"What you get","title":"Voice typing that stays in the work you are doing.","subtitle":"Start at the active cursor. Then choose the transcription path that makes sense for this laptop, this connection and this piece of work.","items":[{"image":{"label":"Woman pressing a laptop key at a wooden desk beside an open notebook."},"title":"Dictate where the cursor already is","body":"Press the global dictation hotkey, speak, and let automatic pasting put the text into the app you have open. This is live laptop writing, not a separate transcription queue."},{"image":{"label":"Woman gesturing beside a laptop and an architectural model at a studio desk."},"title":"Choose Cloud for setup speed, local for offline work","body":"Cloud offers top accuracy and speed without a local model download, then discards transcription audio. Local models run on the device when sensitive or offline work calls for that choice."},{"image":{"label":"Man gesturing toward a laptop at a library desk with books and headphones."},"title":"Keep review and correction in the loop","body":"Read the text where it lands and refine it before you send or save. Voice Assistant can use a dedicated hotkey to edit highlighted text in place, without a wake word."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Direct integrations needed to start","value":"None","footnote":"a global hotkey and automatic pasting return text to the focused cursor"},{"label":"Cost to start voice typing","value":"$$0","footnote":"unlimited local AI models are included on every plan"},{"label":"From download to first dictation","value":"5 min","footnote":"then keep writing at the active cursor"}]},"howItWorks":{"label":"How it works","title":"Text in the field before you reach for the keyboard.","subtitle":"Start free, set up the dictation path you want, then speak into the active text field and review the result before you move on.","image":{"label":"Woman using a laptop beside papers and equipment cases backstage."},"steps":[{"title":"Download for free.","body":"Use OpenWhispr on macOS, Windows or Linux. Open a text field you actually use, then choose the hotkey that starts your dictation."},{"title":"Dictate at the cursor.","body":"Open the email, message, note or document you want to write. Press the hotkey, speak, and let the text paste into that field."},{"title":"Review, correct, continue.","body":"Read the draft where it lands, make the edit you need, and keep working in the same app."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons laptop users choose OpenWhispr","subtitle":"A speech-to-text app has to earn its place in the writing loop. These are the choices that make it useful after the first dictation.","items":[{"title":"Start from one hotkey","body":"Use a global hotkey instead of finding a separate dictation window whenever a sentence is ready."},{"title":"Keep the text in the app you opened","body":"Automatic pasting puts dictated text at the focused cursor, so your email, chat, note or document stays the place you review."},{"title":"Write long thoughts before they disappear","body":"Speak a rough first draft when the thought is moving, then shape the words on screen instead of starting with a blank keyboard."},{"title":"Handle short replies without breaking focus","body":"A quick response still begins at the cursor. Dictate it, read it in place, make the small correction, then send."},{"title":"Teach it the names you use","body":"Add names, jargon and acronyms to the custom dictionary, so recurring words need less cleanup in the text you review."},{"title":"Use Cloud when you want to start quickly","body":"Cloud transcription gives you top accuracy and speed without a local model download."},{"title":"Keep local work on the laptop","body":"Local transcription runs entirely on your device with no internet required, for an offline day or work you want to process locally."},{"title":"Know what happens to Cloud audio","body":"Cloud audio is sent only to transcribe it, then immediately discarded. It is a speed choice, not a storage choice."},{"title":"Make a selected edit without retyping","body":"With Voice Assistant's dedicated hotkey, highlighted text can be edited in place. It does not turn your laptop into a hands-free operating system."},{"title":"Start free, then pay for more of the workflow","body":"Free includes unlimited local models and a Cloud allowance. Pro adds unlimited OpenWhispr Cloud transcription, sync, mobile and API/MCP access."}]},"faq":{"label":"Before you download","title":"The questions laptop users ask first.","subtitle":"Straight answers on where text lands, how you correct it, and which transcription path fits the work in front of you.","items":[{"question":"Will it type into the app I am using?","answer":"OpenWhispr uses a global dictation hotkey and automatic pasting, so the text can land at your focused cursor in the laptop app you already have open. That is text insertion, not a direct integration with each app."},{"question":"Which laptop platforms can I use?","answer":"OpenWhispr is available for macOS, Windows and Linux. Start with the app and test it in the text fields where you actually write."},{"question":"What if I need to correct a name or sentence?","answer":"Review the text where it lands. Add names, jargon and acronyms to the custom dictionary for recurring terms. With Voice Assistant's dedicated hotkey, highlighted text can be edited in place."},{"question":"Should I choose Cloud or local transcription?","answer":"Choose Cloud for top accuracy and speed without a local model download. Audio is sent only to transcribe, then immediately discarded. Choose local when sensitive or offline work calls for device-only processing."},{"question":"Is this for live voice typing or recorded audio?","answer":"This page is for live dictation into an active laptop text field. OpenWhispr also supports meeting transcription and importing audio or video, but those are separate workflows."},{"question":"What is free, and what changes on a paid plan?","answer":"Free includes unlimited local AI models, 2,000 Cloud words a week and five hours of meeting recordings a month. Pro is $8 a month or $80 a year for unlimited Cloud transcription, 20 meeting-recording hours, sync, mobile and API/MCP access. Business is $16 a month or $160 a year and adds unlimited meeting recordings, chat over your data and priority support."}]},"reviews":{"label":"What users say","title":"Built for people who would rather speak than type.","subtitle":"Three users describing the practical shift from keyboard-first work to voice input.","score":{"value":"1","label":"Global hotkey to start dictation","body":"Press the hotkey, speak, and let automatic pasting return the text to the focused cursor."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for laptop voice typing","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Laptop users who want unlimited local models and a limited Cloud allowance"}]},{"parts":[{"text":"Laptop users who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"People who also need unlimited meeting recordings, chat over data, and priority support"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local AI models, plus 2,000 OpenWhispr Cloud words a week"}]},{"parts":[{"text":"Unlimited Cloud transcription, 20 meeting-recording hours, sync, mobile, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings, chat over your data, and priority support"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"A laptop dictation app you can inspect.","paragraphs":["OpenWhispr is an open-source desktop app for turning voice into text, notes and actions. It runs on macOS, Windows and Linux.","The writing loop is simple: press a hotkey, speak, see the text at the cursor, then review it in the app where the work already lives.","This page is for live laptop dictation at the active cursor. Meeting recordings and imported audio are separate workflows."],"image":{"label":"OpenWhispr desktop app on a laptop beside a document"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Leave the keyboard for the words that are easier to say.","body":"Download OpenWhispr, put the cursor where you want to write, and start the next draft with your voice.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. Free forever on macOS, Windows and Linux."}}},"useCaseMacbook":{"breadcrumb":{"home":"Home","useCases":"Use Cases","macbook":"For MacBook"},"metadata":{"title":"Dictation App For MacBook, No 30-Second Cutoff - OpenWhispr","description":"OpenWhispr is free, open-source dictation software for MacBook. Hold the hotkey, talk through the pause, and it keeps listening where macOS stops.","keywords":["dictation app for macbook","macbook dictation software","voice to text macbook","macos dictation alternative","speech to text app for mac","dictate on macbook offline","macbook voice typing app"]},"schema":{"name":"OpenWhispr","applicationCategory":"MacBook dictation software","operatingSystem":"macOS, Windows, Linux","description":"Open-source dictation software for MacBook. Local transcription runs on your Apple Silicon chip for offline and battery work, or choose Cloud for top accuracy and speed.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation app for MacBook","title":"Hold the Globe key. Get the whole thought down.","body":"Talk through any pause, nothing to fix afterward. It transcribes on your MacBook's own chip, so it keeps working on battery with no signal.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Person sitting on a floor cushion and using a laptop on a wooden stool"}},"trustBar":{"label":"Dictate into:","names":["Mail","Messages","Notes","Slack","Word","Google Docs"]},"pain":{"label":"What breaks macOS dictation on a MacBook","title":"You paused to think. It stopped listening.","subtitle":"Away from a desk, on battery, on the mic already built into the lid, the built-in tool keeps cutting the thought off before it lands.","cards":[{"title":"It stops after 30 seconds of silence.","body":"macOS Dictation ends automatically when it hears no speech for 30 seconds, Apple's own documented limit. Every pause to think is a chance it quits on you mid-sentence. OpenWhispr's hotkey stays recording for as long as you hold or toggle it. No silence timer to outrun."},{"title":"Names come back mangled.","body":"One MacBook owner who dictates because typing hurts their hands found macOS's own Dictation turning a name they were writing about into a different name entirely, and called the result unusable. OpenWhispr's custom dictionary learns the names, clients and jargon you actually say, so they come out right."},{"title":"Punctuation is a spoken chore.","body":"Say the word 'comma' out loud or it doesn't appear, outside a handful of supported languages. OpenWhispr transcribes full sentences, not commands, so you talk like you talk and the punctuation is already there."},{"title":"A plane, a train, a café with no signal.","body":"A MacBook runs all day on battery, on a built-in mic, in rooms you don't control: a flight, a shared desk, a hotel room with no wifi. Local transcription runs on your MacBook's own Apple Silicon chip, so none of that stops you dictating."}]},"features":{"label":"What you get","title":"Dictation that keeps up with a laptop that moves.","subtitle":"Run it fully offline on your MacBook's own chip, or switch to Cloud when you're at a desk and want the fastest, most consistent transcription.","items":[{"image":{"label":"Person speaking toward a laptop beside an external monitor at a home desk"},"title":"Local transcription that needs no signal","body":"Whisper runs GPU-accelerated on your MacBook's own Apple Silicon chip, with Parakeet alongside it, both entirely on-device. Dictate on a flight, in the car, anywhere the wifi icon has a line through it."},{"image":{"label":"Person speaking beside an open laptop at a shared café table"},"title":"Text that's already finished","body":"No spoken punctuation commands, and nothing for you to clean up afterward. A custom dictionary primes Whisper and OpenWhispr Cloud with the names and jargon you actually use, so they land right the first time."},{"image":{"label":"Person speaking with a finger on a laptop keyboard at a workshop table"},"title":"One hotkey, talk as long as you need","body":"Hold or tap the same Globe key macOS already uses. Nothing here cuts you off after 30 seconds, so a long pause to think doesn't cost you the sentence."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Words lost to a silence cutoff","value":"None","footnote":"hold the hotkey and keep talking, there's no timer"},{"label":"Cost to start","value":"$$0","footnote":"unlimited local dictation on your MacBook, forever"},{"label":"Setup takes","value":"5 min","footnote":"download, grant Accessibility access, start dictating"}]},"howItWorks":{"label":"How it works","title":"From a held key to finished text, before you've moved your hands.","subtitle":"Download, choose Cloud for the quickest setup or local for offline and battery work, then dictate straight into whatever's already open.","image":{"label":"Person speaking while using a laptop at a round table in an office lounge"},"steps":[{"title":"Download for free.","body":"macOS on Apple Silicon or Intel, plus Windows and Linux. Choose Cloud for no model download or local for offline and battery work."},{"title":"Hold the Globe key and talk.","body":"Speak through as long a pause as you need. Nothing stops you at 30 seconds."},{"title":"It's already in the field.","body":"Finished, punctuated text lands where your cursor was. Review it and send."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons MacBook users choose OpenWhispr","subtitle":"Each one is a fact about what the app does on your machine, not an adjective.","items":[{"title":"Never hits a 30-second wall","body":"Hold or tap the hotkey and OpenWhispr keeps recording until you stop it. No silence timer decides when your thought is over."},{"title":"Works offline, on battery","body":"Local transcription runs GPU-accelerated on your MacBook's own Apple Silicon chip. No signal required, so a flight or a dead wifi router doesn't stop you."},{"title":"Free, unlimited, forever","body":"Local dictation costs $0 on every plan, with no word cap and no trial clock. Upgrade if you want unlimited Cloud, Agent mode, sync or mobile."},{"title":"Learns the names spelling gets wrong","body":"A custom dictionary you edit yourself teaches OpenWhispr the client names, product names and jargon your MacBook's autocorrect never learned."},{"title":"Drops text into whatever's already open","body":"One system-wide hotkey pastes into Mail, Messages, Notes, Slack, Word or any other text field, once you grant the Accessibility permission macOS itself asks for."},{"title":"Cloud audio is discarded the moment it's transcribed","body":"Choose Cloud for speed and OpenWhispr sends audio only to transcribe it, then never stores it. Nothing sits on a server afterward."},{"title":"Open source, not a black box","body":"Every line is on GitHub under MIT. If you want to know exactly what happens to your voice, you can read the code instead of a privacy policy."},{"title":"Rewrite the last sentence without touching the keyboard","body":"A separate Voice Assistant hotkey edits your highlighted text in place, no wake word, no cleanup pass. Say what to change, and the selection updates."},{"title":"100+ languages, not just English","body":"macOS auto-inserts punctuation in only a handful of languages, and a browser's voice typing works in English alone. OpenWhispr dictates 100+ languages without either ceiling."},{"title":"Full sentences, not voice commands","body":"No command vocabulary to learn, no 'new paragraph' or 'exclamation point' to remember. Say it the way you'd say it out loud, and the punctuation and capitalization arrive on their own."}]},"faq":{"label":"Before you download","title":"The questions MacBook users ask first.","subtitle":"Straight answers on privacy, cost and whether this actually beats what's already built in. If yours isn't here, ask us in Discord. We reply within 24 hours.","items":[{"question":"Does the audio ever leave my MacBook?","answer":"With a local model, no. Recording and transcription happen on your own Apple Silicon chip. With OpenWhispr Cloud, audio is sent only to transcribe it, then immediately discarded. Pick whichever mode fits where you're working."},{"question":"Will it cut me off mid-sentence like macOS Dictation does?","answer":"No. macOS's own Dictation stops automatically after 30 seconds of silence. OpenWhispr's hotkey keeps recording for as long as you hold or toggle it, so a long pause to think doesn't end the session."},{"question":"Will it actually get my names and jargon right?","answer":"Better than a generic model, because you can teach it. The custom dictionary lets you add the client names, product names and technical terms you use, so OpenWhispr stops guessing at them."},{"question":"Does it work on a flight or somewhere with no wifi?","answer":"Yes, in local mode. Transcription runs entirely on your MacBook's own chip with no network call, so it works exactly the same at 35,000 feet as it does at a desk. Choose local before you lose signal, not after."},{"question":"Is it actually free, or free with a catch?","answer":"Free gives you unlimited local dictation plus a limited Cloud allowance, no card required. Upgrade to Pro for unlimited Cloud transcription, Agent mode, sync across devices, the iPhone and iPad app, and API/MCP access. Business adds unlimited meeting recordings, chat over your data, and priority support."},{"question":"Does it just transcribe, or can it clean up what I said?","answer":"Both. Dictation gets your words down as finished, punctuated text. A separate Voice Assistant hotkey can also edit text you've already written, in place, using a local or cloud AI model, with no wake word needed."}]},"reviews":{"label":"What users say","title":"Built by people who don't want to be recorded either.","subtitle":"From our Discord, GitHub and inboxes, the same instinct that matters on a laptop you carry everywhere.","score":{"value":"0","label":"Voice recordings kept on a server","body":"In local mode the audio never leaves your MacBook. Nothing is stored on a server, and nothing is sent anywhere to be reviewed."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for MacBook users","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"MacBook users who want a useful free start with local dictation and limited Cloud transcription"}]},{"parts":[{"text":"MacBook users who want unlimited Cloud transcription plus sync, the iPhone app, and API/MCP access"}]},{"parts":[{"text":"Anyone dictating at volume across a team, with unlimited meeting recordings and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Useful Free access, including local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus Agent mode, sync, mobile, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus chat over your data and priority support"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"We built it for a laptop that leaves the desk.","paragraphs":["OpenWhispr started with a simple objection: a thought you already have in your head, mid-sentence on a laptop away from your desk, shouldn't get cut off by a 30-second silence timer.","We build in public. Every release and every line of code is on GitHub, where 5,000+ engineers have starred, forked and audited the project. When we say the dictation stays local, that's not a promise in a privacy policy. It's a property of the code, and you can verify it yourself.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Your next sentence doesn't stop at 30 seconds.","body":"Download OpenWhispr and dictate straight into whatever's open on your MacBook. On battery, offline, on your own machine.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCasePc":{"breadcrumb":{"home":"Home","useCases":"Use Cases","pc":"For PC users"},"metadata":{"title":"Speech to Text Software for PC - OpenWhispr","description":"OpenWhispr is speech to text software for Windows PCs. Hold a hotkey to dictate into the app you use, with Cloud or local processing.","keywords":["speech to text software for PC","speech to text software for Windows","PC dictation software","voice typing software for PC","Windows speech to text app","dictation software for PC"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text software for PC","operatingSystem":"Windows","description":"Open-source speech-to-text software for Windows PCs. Use a hotkey to dictate into the app you are using, with Cloud transcription or local processing.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech to text software for PC","title":"Hold the hotkey. Get your words in the box.","body":"Put the cursor in an email, document, note, or browser field. With auto-paste on, OpenWhispr puts the transcription there. Choose Cloud for fast transcription without a local download, or local models for offline and sensitive work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Man speaking and gesturing toward a desktop monitor at a wooden desk"}},"trustBar":{"label":"Dictate into:","names":["Microsoft Word","Outlook","OneNote","Google Docs","PowerPoint","Notepad"]},"pain":{"label":"When voice typing stops being enough","title":"The words are ready. The typing is the slow part.","subtitle":"A PC writing tool should get spoken words into the place you already work, without making every sentence a new workflow.","cards":[{"title":"The cursor is waiting. You are still typing.","body":"Put the cursor in the reply, document, note, or browser field first. With auto-paste enabled, OpenWhispr returns the transcription there after you speak."},{"title":"Your voice tool lives in only one app.","body":"OpenWhispr uses a global hotkey for dictation instead of an add-in for one editor. Use it where you write, then review the text in that same place."},{"title":"A name or acronym comes back wrong.","body":"Add the spellings you use to the custom dictionary. It gives transcription and cleanup a better guide for names, jargon, and abbreviations."},{"title":"The network is not always part of the job.","body":"Choose a local transcription model when you need to work offline or keep processing on the PC. Choose Cloud when speed and no model download matter more."}]},"features":{"label":"Speech-to-text that fits the PC","title":"Speak where the work already is.","subtitle":"Review the words in the field before you send or save.","items":[{"image":{"label":"Woman gesturing toward a desktop monitor with a wall calendar behind her"},"title":"Use a hotkey at the current cursor","body":"Open the app you need, put the cursor in the text field, and use the global dictation hotkey. With auto-paste on, the transcription goes back to that field."},{"image":{"label":"Woman holding a small microphone beside a desktop monitor at a home office desk"},"title":"Cloud when you want to skip the model download","body":"OpenWhispr Cloud is the fast, high-accuracy path with no local model to download. It sends audio only to transcribe it, then immediately discards that audio."},{"image":{"label":"Man holding handwritten notes and a pencil while looking at a desktop monitor"},"title":"Review text where it will be used","body":"With auto-paste enabled, the transcription returns to the field where it will be used. Check names, wording, and punctuation before you send or save."}]},"stats":{"ariaLabel":"Key facts for PC speech-to-text","items":[{"label":"Extra app between your voice and the text field","value":"None","footnote":"with auto-paste enabled, dictation returns to the cursor in the app you are using"},{"label":"Unlimited local dictation","value":"$$0","footnote":"local transcription models are included on every plan, including Free"},{"label":"Free Cloud transcription","value":"2,000/week","footnote":"words included each week, separate from unlimited local transcription"}]},"howItWorks":{"label":"From download to first draft","title":"Three steps from a hotkey to a finished field.","subtitle":"Choose Cloud for no-download transcription or install a local model for on-device, offline work.","image":{"label":"Man stretching his shoulder while seated at a desk beside a desktop monitor"},"steps":[{"title":"Download it free.","body":"Install OpenWhispr on your Windows PC. Start with OpenWhispr Cloud or download a local transcription model."},{"title":"Put the cursor where you write.","body":"Open an email, document, note, presentation, or browser field. Use the global hotkey and say what you want to write."},{"title":"Review the text, then finish the job.","body":"With auto-paste enabled, the transcription returns to that field. Check names and wording, then send, save, or keep writing."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Why PC users choose it","title":"10 reasons PC users choose OpenWhispr","subtitle":"It is not another document editor. It is speech-to-text that starts with the cursor you already have and gives you a real choice about how the audio is processed.","items":[{"title":"Dictate into the field you opened","body":"Use the global hotkey from the email, document, note, or browser field you are already working in. With auto-paste enabled, the transcription returns to that cursor."},{"title":"Keep auto-paste under your control","body":"Auto-paste and keeping a transcription in the clipboard are separate Settings choices. Turn auto-paste off when you want to decide where text goes, and enable clipboard retention when you want to keep a copy there."},{"title":"Start with useful Free access","body":"Free includes unlimited local AI models and 2,000 OpenWhispr Cloud words each week. Test real PC writing work without a card or a local word cap."},{"title":"Skip the local download when speed matters","body":"OpenWhispr Cloud gives you the high-accuracy, fast path without downloading a local transcription model. The audio is sent only to transcribe it, then immediately discarded."},{"title":"Keep offline work on the PC","body":"Choose a local transcription model for work without an internet connection or work that needs device-only processing. Local models are included on every plan."},{"title":"Give names and jargon a better chance","body":"Add client names, product names, acronyms, and specialist terms to the custom dictionary. OpenWhispr uses those spellings to guide transcription and cleanup."},{"title":"Do not confuse a transcript with finished writing","body":"Speech-to-text gets the first draft into the correct field. You still see the text where it will be used, so checking it is part of the same writing pass."},{"title":"Use the Windows path without a permission detour","body":"Windows auto-paste does not need the accessibility permission step required on macOS. You can decide whether auto-paste is on in Settings."},{"title":"Work in more than one language","body":"OpenWhispr supports 100+ languages on every plan. Keep the writing destination on your PC while the spoken language changes with the work."},{"title":"Upgrade when your Cloud writing grows","body":"Pro adds unlimited OpenWhispr Cloud transcription, Agent mode, sync, mobile, and API or MCP access. Business keeps those PC writing features and adds priority support."}]},"faq":{"label":"Before you install it","title":"The questions PC users ask before they install speech-to-text software.","subtitle":"Straight answers on Windows, app coverage, privacy, correction, offline work, and paid plans.","items":[{"question":"Does OpenWhispr work on a Windows PC?","answer":"Yes. OpenWhispr is available for Windows, macOS, and Linux. The Free plan is available on desktop, including Windows."},{"question":"Can I dictate into the apps I already use?","answer":"Open the app and put the cursor in the text field you want to fill. OpenWhispr uses a global hotkey. With auto-paste enabled, it puts the transcription at that cursor. It is not an add-in for Word, Outlook, Google Docs, or any other editor."},{"question":"Should I choose Cloud or local speech-to-text?","answer":"Choose OpenWhispr Cloud when you want fast, high-accuracy transcription with no local model download. Choose a local model for offline or device-only processing. Cloud sends audio only to transcribe it, then immediately discards it."},{"question":"What if it gets a name or acronym wrong?","answer":"Add the spelling to the custom dictionary. It guides transcription and cleanup for names, jargon, and acronyms. Review the text before you send or save it, especially where wording matters."},{"question":"Can I use it without an internet connection?","answer":"Yes, after you download a local transcription model. Local models run on your device without internet and are unlimited on every plan, including Free."},{"question":"What do I get by upgrading from Free?","answer":"Free includes unlimited local models plus 2,000 OpenWhispr Cloud words each week. Pro adds unlimited Cloud transcription, Agent mode, 20 hours of meeting recordings each month, sync, mobile, and API or MCP access. Business adds unlimited meeting recordings and chat over your data."}]},"reviews":{"label":"What people tell us","title":"Better than typing. Still your words to review.","subtitle":"From our Discord, GitHub and inboxes. Read these for the PC dictation workflow, not as a promise that every spoken word needs no correction.","score":{"value":"At the cursor","label":"With auto-paste enabled","body":"Start in the text field that needs the words. With auto-paste on, the transcription returns there for review instead of living in a separate editor."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free to start.","Upgrade when the workflow grows."],"caption":"Compare OpenWhispr plans for PC speech-to-text","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"PC users who want unlimited local dictation and 2,000 Cloud words each week"}]},{"parts":[{"text":"PC users who want unlimited Cloud transcription plus Agent mode, sync, mobile, and API or MCP access"}]},{"parts":[{"text":"Teams that also need unlimited meeting recordings and chat over their data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local dictation plus 2,000 OpenWhispr Cloud words each week"}]},{"parts":[{"text":"Unlimited OpenWhispr Cloud transcription plus Agent mode and connected workflow features"}]},{"parts":[{"text":"Everything in Pro plus unlimited meeting recordings and chat over your data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Choose local models, or use the Free Cloud allowance. You can also bring your own API keys."}]},{"parts":[{"text":"Use unlimited OpenWhispr Cloud transcription without a local model download."}]},{"parts":[{"text":"Everything in Pro, plus unlimited meeting recordings and chat over your data."}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" guides names and terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" guides names and terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Agent mode","body":" use AI actions on transcripts"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools and workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" guides names and terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Agent mode","body":" use AI actions on transcripts"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools and workflows"},{"title":"Chat over your data","body":" ask about files and transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Enterprise adds SSO, audit logs, retention controls, and dedicated support."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"Your PC already has the place to write. We built for getting words there.","paragraphs":["OpenWhispr started with a simple idea: people writing all day should not have to choose between typing every first draft and moving speech through a separate app.","The app uses a hotkey and optional auto-paste so you can dictate into the PC apps you already use, then decide what needs a correction.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"One of OpenWhispr’s founders"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Open the app. Put the cursor where you write.","body":"Download OpenWhispr on your Windows PC, choose Cloud or a local model, and dictate the next email, document, note, or browser field without changing where the work lives.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free to download on Windows, macOS and Linux."}}},"useCaseWindows11":{"breadcrumb":{"home":"Home","useCases":"Use Cases","windows11":"For Windows 11"},"metadata":{"title":"Free Speech-to-Text Software for Windows 11 - OpenWhispr","description":"OpenWhispr is free, open-source speech-to-text for Windows 11 that fixes punctuation automatically and works offline, no Copilot+ PC required.","keywords":["free speech to text software for windows 11","windows 11 dictation software","windows 11 voice typing alternative","offline dictation software windows","voice access alternative","speech to text without internet windows 11","free dictation software windows"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text software for Windows 11","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source, free speech-to-text software for Windows 11. Local transcription that runs with no internet connection and no Copilot+ hardware required.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Free speech-to-text software for Windows 11","title":"Speak into any app. Get it right, offline.","body":"A local model runs on the Windows 11 PC you already own, no internet or Copilot+ hardware required. Teach it your names and terms once, and stop correcting the same word twice.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman speaking and gesturing toward a laptop at a table with bookshelves behind her."}},"trustBar":{"label":"Dictate into any app on Windows 11, including:","names":["Microsoft Word","Outlook","Gmail","Slack","Google Docs","Chrome"]},"pain":{"label":"What Windows 11 already gave you, and where it stops","title":"Win+H got you this far. It won't get you the rest of the way.","subtitle":"You already tried the built-in tools. Here's exactly where each one runs out.","cards":[{"title":"Say \"question mark\" and get the words \"question mark\", typed out.","body":"Voice Access takes spoken punctuation literally instead of inserting the symbol, and a thinking pause can drop a stray period mid-sentence. OpenWhispr's cleanup converts spoken punctuation to the actual mark and treats a pause as a pause, not a full stop."},{"title":"You corrected the same name a hundred times. It's still wrong.","body":"Voice Access's custom vocabulary only improves the correction list, not what gets typed first. With OpenWhispr Cloud or the Whisper local model, your custom dictionary is sent to transcription itself, so your names, jargon, and terms come out right on the first pass."},{"title":"The good tier needs a new PC. The free tier needs Wi-Fi.","body":"Windows' voice typing runs on Microsoft's cloud and needs a connection, and the auto-punctuating fluid dictation tier is locked to Copilot+ hardware. OpenWhispr's local model runs entirely on the Windows 11 PC you already have, online or off."},{"title":"Win+H for this, Voice Access for that, Word's Dictate for the rest.","body":"Microsoft's own support answer is to switch tools depending on what you're doing. OpenWhispr is one hotkey that works the same way in Word, email, chat, or any app with a cursor."}]},"features":{"label":"From your voice to typed text","title":"Built for the moment Win+H gives up.","subtitle":"Choose the local model for offline, private dictation, or Cloud for the fastest transcription.","items":[{"image":{"label":"Man with glasses speaking toward a desktop monitor at a desk beside a bright window."},"title":"A local model, no internet or new PC required","body":"Download a local model once and OpenWhispr transcribes entirely on your PC, the one you already own. No round-trip to a server, no Copilot+ requirement, no signal needed for transcription."},{"image":{"label":"Person in a green sweater speaking toward a desktop monitor in a kitchen."},"title":"A dictionary that changes what gets typed, not just the correction list","body":"Add your name, your field's jargon, and the proper nouns Windows keeps missing. With OpenWhispr Cloud or the Whisper local model, that list is passed to transcription itself, so the right word shows up the first time."},{"image":{"label":"Man gesturing toward a desktop monitor with shelves of records behind him."},"title":"One hotkey, every app","body":"Word, Outlook, Slack, a web form: press the hotkey and clean, punctuated text pastes in wherever your cursor already is. Nothing to install per app, no add-in, no plugin to configure."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Internet needed to dictate","value":"None","footnote":"the local model transcribes entirely on your Windows 11 PC"},{"label":"Cost to dictate offline","value":"$$0","footnote":"unlimited local transcription in the Free plan"},{"label":"Punctuation cleanup","value":"Automatic","footnote":"converts spoken \"comma\" and \"question mark\" into the mark, by default"}]},"howItWorks":{"label":"Download to first sentence","title":"Three steps from download to typed text.","subtitle":"Choose Cloud transcription, or download a local model that needs no internet.","image":{"label":"Man speaking toward a desktop monitor at a wooden desk in a brick-walled room."},"steps":[{"title":"Download it free.","body":"Setup offers OpenWhispr Cloud or a local model. Pick local and it downloads once, then works with no internet."},{"title":"Add your names and terms.","body":"Put the proper nouns and jargon Windows keeps missing into the custom dictionary, once."},{"title":"Press the hotkey and speak.","body":"Clean, punctuated text pastes into whatever you're typing into: Word, email, chat, any app. Want the cleanup pass offline too? Set a local model for that in Settings; transcription already is."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Why Windows 11 users switch","title":"10 reasons Windows 11 users choose OpenWhispr over the built-in tools","subtitle":"Every one of these is a specific gap in Win+H or Voice Access, checkable in the app itself.","items":[{"title":"Works with the Wi-Fi off","body":"A local model needs no connection to transcribe, unlike Windows' voice typing, which runs on Microsoft's cloud and stops working the moment your connection does."},{"title":"No Copilot+ PC required","body":"Windows' best free auto-punctuation tier is locked to Copilot+ hardware. OpenWhispr's local dictation runs on the ordinary Windows 11 PC you already own, no hardware upgrade required."},{"title":"One hotkey, every app, not just Word","body":"Word's Dictate only works in Word, and Win+H behaves differently app to app. OpenWhispr's hotkey works the same way everywhere, so there's one habit to build instead of three."},{"title":"A dictionary that changes the first pass","body":"Voice Access's custom vocabulary only reorders the correction suggestions. With OpenWhispr Cloud or the Whisper local model, your custom dictionary is sent to transcription itself, so a term you've taught it shows up correctly the first time, not the fifth."},{"title":"Punctuation from how you speak, not a command you say right","body":"Say \"question mark\" in Windows and you may get the literal words typed out. OpenWhispr's cleanup converts spoken punctuation into the mark and treats a thinking pause as a pause, not a period."},{"title":"Self-corrections dropped automatically","body":"Say \"send it by Thursday, no wait, Friday\" and OpenWhispr keeps only \"Send it by Friday.\" Windows' tools transcribe the retry verbatim and leave you to fix it by hand. Need to fix something after it's already typed? Agent mode (Pro) edits it in place, no re-highlighting."},{"title":"Free for the whole machine, no per-minute meter","body":"Unlimited local dictation at $0, plus 2,000 Cloud words a week. No trial clock, no per-minute charge, no forced upgrade to keep dictating."},{"title":"Cloud audio is transcribed, then gone","body":"Choose OpenWhispr Cloud for the highest accuracy and speed, with no local model to download. Audio is sent only to transcribe it, then immediately discarded, never stored."},{"title":"Open source, so you can check it yourself","body":"How OpenWhispr's local dictation works isn't a promise, it's code. Read exactly what happens to your audio on GitHub before you trust it with anything."},{"title":"More languages than Voice Access supports","body":"Voice Access covers 15 languages and dialects, and Windows' best free auto-punctuation only works in English. OpenWhispr supports 100+ languages on every plan."}]},"faq":{"label":"Before you install it","title":"Six questions Windows 11 users ask before switching from Win+H.","subtitle":"Straight answers on internet, hardware, accuracy, and cost. More questions? Ask in Discord.","items":[{"question":"How is this different from Windows' own Win+H voice typing?","answer":"Win+H runs on Microsoft's cloud speech service and needs an internet connection, and the auto-punctuating fluid dictation tier is exclusive to Copilot+ PCs. OpenWhispr's local model transcribes entirely on the Windows 11 PC you already have, works offline, and adds a custom dictionary for your own vocabulary."},{"question":"Do I need a Copilot+ PC?","answer":"No. OpenWhispr's local dictation is unlimited and free on an ordinary Windows 11 PC. Copilot+ is Microsoft's own hardware requirement for their fluid dictation tier, not something OpenWhispr needs."},{"question":"Will it get my name and technical terms right?","answer":"Add them to the custom dictionary once. With OpenWhispr Cloud or the Whisper local model, that list is passed to transcription itself, so the right word shows up on the first pass instead of only appearing in a correction menu."},{"question":"Does it work without an internet connection?","answer":"Yes, for transcription: download a local model once during setup and speech-to-text runs entirely on your device afterward, no connection required. The punctuation cleanup pass uses a quick cloud call by default; set a local model for that too in Settings if you want the whole pipeline offline."},{"question":"Is it actually free, or a trial?","answer":"Free includes unlimited local dictation and local AI models, plus 2,000 Cloud words a week, with no time limit. Pro adds unlimited Cloud transcription, Agent mode, 20 hours of meeting recordings, sync, mobile, and API/MCP access."},{"question":"Does it work in every app, or just Word?","answer":"Every app. One system-wide hotkey pastes clean text wherever your cursor already is: Word, Outlook, a browser, Slack, a chat box. There's no add-in to install into any of them."}]},"reviews":{"label":"What people tell us","title":"Built by people who wanted their own machine to do this.","subtitle":"From our Discord, GitHub and inboxes. They're not Windows 11 testimonials, but the no-WiFi, local-model story below will sound familiar.","score":{"value":"0","label":"Audio stored on our servers with local transcription","body":"With a local transcription model the audio never leaves your PC, so there's nothing on our side to lose, breach or hand over."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free to start.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for Windows 11","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Windows 11 users who want offline local dictation and 2,000 Cloud words a week"}]},{"parts":[{"text":"Windows 11 users who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Individuals and teams documenting at volume with unlimited meeting recordings and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited offline local dictation plus 2,000 Cloud words a week"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Enterprise adds SSO, audit logs, retention controls, and dedicated support."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"The built-in tool should have just worked.","paragraphs":["Free dictation software shouldn't need an internet connection or new hardware to be any good. That's the position OpenWhispr builds from.","We build in public. The app is open source, so you can inspect how it works before you install it, instead of taking our word for it.","Community support is available in Discord. Paid plans add email or priority support."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"$$0 · open source · punctuation fixed automatically","title":"Stop fighting Win+H. Start dictating.","body":"Download OpenWhispr, add your names and terms to the custom dictionary, and dictate into whatever you're already typing into.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free on Windows, macOS, and Linux. Paid plans add Cloud transcription, mobile, and workflow tools."}}},"useCaseWindows10":{"breadcrumb":{"home":"Home","useCases":"Use Cases","windows10":"For Windows 10"},"metadata":{"title":"Speech to Text Software for Windows 10 - OpenWhispr","description":"OpenWhispr is free, open-source speech to text software built for Windows 10, no Windows 11 or internet connection required.","keywords":["speech to text software for windows 10","windows 10 dictation software","voice to text windows 10","offline speech to text windows 10","free dictation software windows 10","dictation software no windows 11"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech to text software for Windows 10","operatingSystem":"macOS, Windows, Linux","description":"Open-source speech to text software for Windows 10 or later. Choose Cloud for speed with no model download, or local processing that works with no internet at all.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech to text software for Windows 10 PCs","title":"Press the hotkey. Get the reply, ready to send.","body":"Installs on Windows 10 or later, no Windows 11 required. Local models work fully offline, once downloaded.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman speaking and gesturing at a desktop computer beside plants and large windows."}},"trustBar":{"label":"Dictate into:","names":["Microsoft Word","Microsoft Outlook","Microsoft Edge","Microsoft Teams","Notepad"]},"pain":{"label":"What stalls dictation on Windows 10","title":"Windows 10 dictation stalled. Yours doesn't have to.","subtitle":"You need something that installs on the OS you actually have, keeps working without an internet connection, and gets your words right the first time.","cards":[{"title":"Win+H needs the internet, every time.","body":"Windows 10's built-in dictation requires a live connection to work at all, and it stops responding for some people after a Windows update, with no fix in sight. OpenWhispr's local models run entirely on your device, so a dropped connection never stops you mid-sentence."},{"title":"Voice access never came to your PC.","body":"Microsoft's newer voice tools are gated to Windows 11 version 22H2 and later; Windows 10 keeps the deprecated Windows Speech Recognition, and it isn't getting better. OpenWhispr installs on Windows 10 or later. No upgrade required."},{"title":"It typed \"background noise\" into an email.","body":"Generic recognition mishears background sound as words and turns punctuation into a spoken chore. OpenWhispr's custom dictionary learns your names and terms, and local speech models built for accuracy do the listening."},{"title":"The professional option you remember is gone.","body":"The consumer edition of Dragon was discontinued in 2023, with no replacement at that price. OpenWhispr is free, with unlimited local dictation for as long as you use it. No card required."}]},"features":{"label":"What you get","title":"Dictation that doesn't care which Windows you're on.","subtitle":"Cloud is the easy choice for an older CPU; local is the choice when you want zero reliance on a connection. Both run on the Windows 10 you have today.","items":[{"image":{"label":"Man speaking and gesturing at a desktop computer in a workshop."},"title":"Local processing that never needs a connection","body":"Download a Whisper or Parakeet model once, and every dictation after that runs on your own device. No internet, no server, and it keeps working through the day the Wi-Fi drops."},{"image":{"label":"Person speaking and gesturing at a desktop monitor at a kitchen table."},"title":"Cloud transcription with nothing to download","body":"Sign in and dictate. OpenWhispr Cloud transcribes on our servers for the fastest, most consistent results, with no model to download and no strain on an older machine."},{"image":{"label":"Man gesturing toward a laptop in a room with acoustic panels and a guitar."},"title":"Dictate wherever your cursor already is","body":"Press a hotkey and speak. The text is typed into Word, Outlook, Edge, Teams, Notepad or any app that accepts typed text. No per-app setup needed."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Internet needed in local mode","value":"None","footnote":"once the model is downloaded, dictation works fully offline"},{"label":"Cost for unlimited local dictation","value":"$$0","footnote":"free forever, no Windows 11 upgrade needed"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first dictated line on Windows 10"}]},"howItWorks":{"label":"How it works","title":"Installed on the Windows 10 you already have.","subtitle":"Start free. Three simple steps and you're dictating today.","image":{"label":"Man speaking and gesturing at a laptop beside a window and a mug."},"steps":[{"title":"Download for free.","body":"Windows 10 or later, 64-bit. No Windows 11, and nothing extra to grant beyond your microphone."},{"title":"Dictate the note.","body":"Press the hotkey and speak. Choose Cloud for speed, or a local model if you'd rather skip the connection entirely."},{"title":"Review, paste, done.","body":"Dictation cleanup tidies filler words, grammar and punctuation before it lands where your cursor is, so a quick reply is ready to send. Edit if you need to, and you're done."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons Windows 10 users choose OpenWhispr","subtitle":"Not a workaround for an OS Microsoft moved past. A dictation tool built to run on it.","items":[{"title":"Runs on Windows 10, not just Windows 11","body":"The minimum requirement is Windows 10 or later, 64-bit. No Windows 11 gate, no new hardware to buy."},{"title":"Works with no internet connection at all","body":"Local models process entirely on your device once downloaded, unlike Windows' built-in dictation, which needs a live connection every time."},{"title":"The published requirement has no expiry date","body":"Microsoft's own Office security updates on Windows 10 are set to run through 10 October 2028; Edge and WebView2 keep updating on Windows 10 22H2 until at least that same date. OpenWhispr's requirement is simply Windows 10 or later, with no end date attached to it."},{"title":"Free, with no card and no expiry","body":"Unlimited local dictation on the Free plan, forever. No trial clock, no card required to start."},{"title":"Dictate into whatever you're already using","body":"One hotkey types clean text into Word, Outlook, Edge, Teams, Notepad, or any app with a text field. Nothing to configure per app."},{"title":"Hold to talk or tap to talk, your choice","body":"Windows supports both activation modes for dictation: hold the key while you speak, or press once to start and again to stop. Built-in dictation forces an all-or-nothing switch between speaking and typing that breaks your rhythm; picking the mode that fits you means it doesn't."},{"title":"Teach it the words it keeps getting wrong","body":"A custom dictionary learns names, product codes and jargon, so the terms you use every day stop coming out mangled."},{"title":"GPU acceleration for local models, one click away","body":"NVIDIA, AMD and Intel GPUs get a one-click acceleration download from Settings. On older hardware, pick a smaller local model, or skip local entirely and use Cloud, which asks very little of your computer."},{"title":"Say what you want changed, not just what you said","body":"Select the text, press the voice agent hotkey (set it once in Settings), and say what to change: \"make this more formal\", \"fix the typos\". The rewritten version lands in place, not your raw words."},{"title":"Open source, so IT can check for themselves","body":"Every line is public on GitHub. Before it goes on a work PC, someone can read exactly what happens to the audio."}]},"faq":{"label":"Before you download","title":"The questions Windows 10 users ask first.","subtitle":"Straight answers on compatibility, cost and accuracy. If yours isn't here, ask us in Discord. We reply within 24 hours.","items":[{"question":"Does OpenWhispr actually run on Windows 10?","answer":"Yes. It installs on Windows 10 or later, 64-bit, no Windows 11 required."},{"question":"Do I need an internet connection to dictate?","answer":"Not if you choose a local model. Local transcription runs entirely on your device and works with no internet connection at all once the model is downloaded. OpenWhispr Cloud needs a connection, since your audio is sent to our servers to transcribe it."},{"question":"Will it run well on an older Windows 10 PC?","answer":"Yes. Pick a smaller local model for older hardware, or use the one-click GPU acceleration in Settings. Cloud is the alternative if you'd rather skip local models altogether."},{"question":"How is this different from Windows' built-in dictation?","answer":"Win+H needs an internet connection even for plain dictation, and the older Windows Speech Recognition's per-user training option isn't reliably there anymore either. OpenWhispr's custom dictionary learns the names and terms that keep coming out wrong, with no training menu to go hunting for."},{"question":"What does it cost?","answer":"Free, with unlimited local dictation and no card needed. Pro is $8/month for unlimited Cloud transcription, sync and mobile. Business is $16/user/month and adds unlimited meeting recordings and chat over your data."},{"question":"Can it learn my own words and terms?","answer":"Yes. The custom dictionary learns names, product codes and jargon you use often, on every plan."}]},"reviews":{"label":"What users say","title":"What people actually said, unedited.","subtitle":"From our Discord, GitHub and inboxes: people who kept an older machine going and wanted the dictation to keep up.","score":{"value":"0","label":"Files stored on a server","body":"In local mode the audio never leaves your device. Nothing is stored on a server, and nothing is sent to a transcriber."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for Windows 10 users","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Windows 10 users who want free, unlimited local dictation, no Windows 11 or internet required"}]},{"parts":[{"text":"Windows 10 users who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Teams keeping older Windows 10 machines productive, with unlimited meeting recordings and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Dragon Home, the affordable paid option, is gone from sale. OpenWhispr is free, and stays free"}]},{"parts":[{"text":"Cloud transcription that doesn't ask anything of an older CPU"}]},{"parts":[{"text":"Unlimited meeting recordings plus chat over your data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some machines aren't getting upgraded. We built for those too.","paragraphs":["OpenWhispr started with a simple objection: dictation software shouldn't stop working the day your OS falls out of support. Windows 10 is still Windows, so dictation, Cloud and local transcription work the same as they do everywhere else we ship.","We build in public. Every release and every line of code is on GitHub, where 5,000+ engineers have starred, forked and audited the project. When we say local dictation runs on your device, that's not a promise in a privacy policy. It's a property of the code, and anyone still running Windows 10 can verify it themselves.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Your words, typed on the Windows 10 PC you already have.","body":"Download OpenWhispr and your next dictation lands as clean text, offline if you want it. No Windows 11 required.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseUbuntu":{"breadcrumb":{"home":"Home","useCases":"Use Cases","ubuntu":"For Ubuntu Users"},"metadata":{"title":"Speech to Text Software for Ubuntu - OpenWhispr","description":"OpenWhispr turns speech into editable text at your Ubuntu cursor. Use Cloud for fast setup or local models for offline and sensitive work.","keywords":["speech to text software for ubuntu","ubuntu speech to text","ubuntu dictation software","voice typing ubuntu","linux dictation app","offline speech to text ubuntu"]},"schema":{"name":"OpenWhispr","applicationCategory":"Ubuntu speech-to-text software","operatingSystem":"Linux","description":"Open-source speech-to-text software for Ubuntu users. Use a global hotkey to turn speech into editable text for the field you have focused. Choose Cloud without a local model download, or local for offline and device-only transcription.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text software for Ubuntu users","title":"Press the hotkey. Get text at your cursor.","body":"Speak into the document, message or prompt you already have open. Choose Cloud for a fast start without a local model download, or local for offline and sensitive work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman speaking beside a desktop monitor with one hand on a keyboard in a bright room."}},"trustBar":{"label":"Dictate into:","names":["Documents","Messages","Prompts","Terminals","Browser forms","Text editors"]},"pain":{"label":"What makes voice typing feel unfinished","title":"You need text in this field, not a transcript in another tab.","subtitle":"A hotkey should turn one spoken paragraph into an editable draft where you are already writing. It should not turn that paragraph into a browser, script, or copy-paste detour.","cards":[{"title":"A browser tab leaves you copying.","body":"You are writing a reply, document, or prompt now. Browser transcription makes the words arrive somewhere else first. OpenWhispr is a desktop app for the moment the field in front of you needs a first draft."},{"title":"Source-code setup is not a shortcut.","body":"A usable Ubuntu dictation tool starts as a Linux desktop app. You should be choosing how to transcribe, not assembling a voice-typing workflow before you can write one sentence."},{"title":"Wayland auto-paste is conditional.","body":"OpenWhispr chooses a compositor-specific paste method instead of promising the same behavior on every desktop session. KDE with XWayland tries the RemoteDesktop portal first; GNOME and other Wayland compositors try Linux input access first, then use later fallbacks."},{"title":"A missed paste should not erase the sentence.","body":"When automatic paste is unavailable, OpenWhispr copies the completed transcription to the clipboard. Paste it with Ctrl+V, edit it, and carry on without dictating the paragraph again."}]},"features":{"label":"The whole voice-to-text loop","title":"Speak. See editable text. Keep writing.","subtitle":"The job is typing by voice in the field you already opened. Cloud and local models are two deliberate ways to do it, not separate products.","items":[{"image":{"label":"Woman speaking and gesturing beside an open laptop at a desk."},"title":"One hotkey for the field you are in","body":"Put the cursor where the sentence belongs, press the global hotkey, speak, and release. With auto-paste available, OpenWhispr puts the transcription in the active app. Otherwise it is ready in the clipboard."},{"image":{"label":"Woman speaking toward a desktop monitor while using a mouse at a wooden desk."},"title":"Set up automatic paste for your desktop session","body":"On Wayland, OpenWhispr chooses a compositor-specific paste method. On GNOME and KDE, approve the RemoteDesktop permission when asked. If automatic paste cannot reach the field, the completed transcription stays in the clipboard."},{"image":{"label":"Man holding a sheet of paper and speaking beside a laptop near a rain-streaked window."},"title":"Choose Cloud or local for the conditions","body":"Choose OpenWhispr Cloud for top accuracy and speed without a local model download, with audio discarded after transcription. Download Whisper or Parakeet when offline or device-only transcription is the better fit."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Local-mode audio uploaded","value":"None","footnote":"a downloaded model transcribes on your Ubuntu device"},{"label":"Cost for local dictation","value":"$$0","footnote":"unlimited local AI models on every plan"},{"label":"When automatic paste fails","value":"Ctrl+V","footnote":"the transcription is copied for a manual paste"}]},"howItWorks":{"label":"How it works on Ubuntu","title":"Start in the field you want to fill.","subtitle":"Install the Linux desktop app, choose Cloud or a local model, then use the hotkey in the document, reply, or prompt already in front of you.","image":{"label":"Man speaking toward a desktop monitor at a desk beside a window overlooking the sea."},"steps":[{"title":"Install and choose a transcription path.","body":"Download the Linux desktop app. Choose OpenWhispr Cloud to start without downloading a model, or download a local Whisper or Parakeet model for device-only transcription."},{"title":"Put the cursor where the sentence belongs.","body":"Open your document, message, or prompt. Put the cursor where the sentence belongs, press the global hotkey, speak, then release it."},{"title":"Review the text where you work.","body":"With auto-paste set up, OpenWhispr sends the text to the active app. Otherwise paste the completed transcription from the clipboard, then edit, send, or save it."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"The parts that make it usable","title":"10 reasons Ubuntu users choose OpenWhispr","subtitle":"The result should be editable text where you are writing. These are the product choices and Linux details that make that workflow practical.","items":[{"title":"Start at the cursor you already need","body":"Put the cursor in the document, reply, or prompt first. The global hotkey turns the spoken paragraph into text for the active app, so your draft starts where the next sentence belongs."},{"title":"Use a Linux desktop app, not a browser detour","body":"OpenWhispr ships Linux desktop targets, including a Debian package. Install the app, set a hotkey, and dictate from the program where the writing is already happening."},{"title":"Wayland auto-paste has a visible setup path","body":"OpenWhispr does not promise one Wayland paste path. KDE with XWayland tries the RemoteDesktop portal first; GNOME and other Wayland compositors try Linux input access first, then use later fallbacks. The Debian package recommends wl-clipboard for clipboard support."},{"title":"Clipboard mode is still a usable fallback","body":"If automatic paste is not available, the transcription is copied to the clipboard. One Ctrl+V puts the completed sentence in the focused field without repeating the dictation."},{"title":"Cloud skips the local-model download","body":"Choose OpenWhispr Cloud when you want top accuracy and speed without managing a local model. Cloud audio is sent only to transcribe it, then never stored."},{"title":"Local works offline and stays on the device","body":"Download a Whisper or Parakeet model once, then transcribe without an internet connection. The local path keeps the audio and transcription work on your Ubuntu machine."},{"title":"Speak one language, paste another","body":"Use the dedicated translation hotkey to dictate in one language and paste text in another. It still starts from the focused field, not a separate transcription queue."},{"title":"Keep project words out of the correction loop","body":"Add project names, technical terms, and acronyms to the custom dictionary. It helps OpenWhispr put the spelling you use into the field already open, so you edit instead of retype it."},{"title":"Choose the engine for the language you speak","body":"Before you dictate into the next field, choose an engine that supports the language you speak. The website's 100+ refers to the underlying Whisper model family; the app currently offers 60 selectable transcription languages plus auto-detect."},{"title":"The local path starts at $0","body":"Every plan includes unlimited local AI models. Start with device-only dictation at no cost, then use the plan comparison to decide whether Cloud's no-download convenience is worth an upgrade."}]},"faq":{"label":"Before you install","title":"Six Ubuntu questions worth answering first.","subtitle":"Straight answers on typing by voice, Wayland setup, offline work, and where your audio goes.","items":[{"question":"Can this replace the keyboard for a paragraph?","answer":"Put the cursor in the field you want to fill, press the hotkey, speak, and release. With automatic paste available, the transcription goes to the active app. Otherwise paste it from the clipboard and edit it there."},{"question":"Does automatic paste work on Ubuntu Wayland or X11?","answer":"Current releases provide Linux .deb, AppImage, RPM, and tar.gz downloads. On Wayland, OpenWhispr uses a compositor-specific paste path: KDE with XWayland tries the RemoteDesktop portal first; GNOME and other Wayland compositors try Linux input access first, then use later fallbacks. The Debian package recommends wl-clipboard for clipboard support. Test the hotkey-to-field flow on your own session. If automatic paste is unavailable, paste the transcription manually with Ctrl+V."},{"question":"Should I choose Cloud or a local model?","answer":"Choose OpenWhispr Cloud for top accuracy and speed without a local model download. Cloud audio is sent only to transcribe it, then never stored. Choose a local Whisper or Parakeet model when you need offline or device-only transcription."},{"question":"Can I dictate with no internet connection?","answer":"Yes. Download a local model first, then transcription runs on your Ubuntu device without an internet connection. Local models are unlimited on every plan."},{"question":"Can it handle my project names and acronyms?","answer":"Add them to the custom dictionary. It is included on every plan and is designed to improve recognition of names, jargon and acronyms in your transcripts."},{"question":"What does a paid plan add after the free local path?","answer":"Free includes unlimited local models and 2,000 OpenWhispr Cloud words each week. Pro is $8 monthly or $80 yearly for unlimited Cloud transcription, 20 monthly meeting-recording hours, sync, API and MCP access, Agent mode, and the iPhone and iPad app. Business is $16 monthly or $160 yearly and adds unlimited meeting recordings, chat over your data, and priority support."}]},"reviews":{"label":"What users say","title":"Built for people who need text in the field they are using.","subtitle":"Three users on why local models, voice input, and offline work changed their typing workflow.","score":{"value":"At your cursor","label":"Where live dictation belongs","body":"With automatic paste available, OpenWhispr puts the transcription in the active app. Otherwise, it remains in the clipboard for you to paste and review there."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for Ubuntu dictation","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Ubuntu users who want unlimited local dictation with a limited Cloud allowance"}]},{"parts":[{"text":"Ubuntu users who want unlimited Cloud dictation without a local model download"}]},{"parts":[{"text":"People who need paid desktop workflow features beyond unlimited Cloud dictation"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local dictation, plus 2,000 OpenWhispr Cloud words each week"}]},{"parts":[{"text":"Unlimited Cloud dictation, plus sync, API and MCP access"}]},{"parts":[{"text":"Unlimited Cloud dictation, team features, chat over your data and priority support"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with team features and admin, SSO, SAML and SCIM, audit logs, retention controls, and dedicated support."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built for the Ubuntu desktop","title":"The sentence belongs in your app, not in a transcription queue.","paragraphs":["OpenWhispr starts from a simple expectation: a hotkey should turn a spoken thought into editable text for the Ubuntu app where the cursor already is. The rest of the desktop workflow should not become somebody else's project.","We build in public. The desktop app is MIT licensed on GitHub. You can inspect the Linux downloads, local-model choices, and Wayland setup path before choosing how you want to transcribe.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"Free · open source · built for desktop dictation","title":"Put the next sentence at your Ubuntu cursor.","body":"Download OpenWhispr, choose Cloud or a local model, then speak into the document, message, or prompt already open. Review the text where it lands.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free local models on Linux. Cloud starts with 2,000 words each week."},"byline":{"authorLabel":"Written by","author":"Joshua Padoa","aiDisclosure":"Drafted with AI assistance","reviewerLabel":"Reviewed and fact checked by","reviewer":"Wesley van der Hoop","updatedLabel":"Last reviewed","dateModified":"2026-09-10"}}},"useCaseMicrosoftWord":{"breadcrumb":{"home":"Home","useCases":"Use Cases","microsoftWord":"For Microsoft Word Users"},"metadata":{"title":"Free Speech to Text Software for Microsoft Word - OpenWhispr","description":"OpenWhispr turns speech into editable text. With auto-paste, it can paste at an active, writable Word cursor or keep text in clipboard. No Word add-in.","keywords":["free speech to text software for microsoft word","speech to text for Word","free dictation software for Word","voice typing for Microsoft Word","Word dictation alternative","offline speech to text for Word"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text software for Microsoft Word users","operatingSystem":"macOS, Windows, Linux","description":"Desktop speech-to-text software that can paste generated text at an active writable cursor in Microsoft Word. It is not a native Word integration.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text software for Microsoft Word","title":"Set the cursor in Word. Get editable text.","body":"OpenWhispr is desktop dictation software, not a Word add-in. With auto-paste, it can paste generated text at an active, writable Word cursor.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman speaking and gesturing beside a laptop at a table in a bright studio."}},"trustBar":{"label":"Compare these dictation and review tools:","names":["Microsoft Word","Word Dictate","Windows voice typing","Voice Access","Microsoft Editor","Track Changes"]},"pain":{"label":"What slows a Word draft","title":"The cursor is waiting. The first draft is still stuck in your head.","subtitle":"The practical question is not whether Word has dictation. It is how spoken text reaches your document, and what still needs a review pass.","cards":[{"title":"Native Word Dictate is only one route.","body":"Word Dictate is Word's own feature. OpenWhispr is separate desktop software that can paste text at the active Word cursor when auto-paste is enabled. The two are not the same kind of tool."},{"title":"Leaving Word to move text back in.","body":"Copying a draft from another window creates another stop. OpenWhispr can paste generated text at the focused, writable Word cursor, so the text arrives in the document you already have open."},{"title":"Recognition errors turn into editing work.","body":"Names, terms and a rough microphone can still need correction. OpenWhispr gives you editable text and a custom dictionary. Word remains where you proofread and revise the result."},{"title":"Free should not hide the tradeoff.","body":"Local speech-to-text is unlimited and free. OpenWhispr Cloud includes 2,000 words a week on Free, then paid plans add unlimited Cloud transcription and other workflow features."}]},"features":{"label":"What OpenWhispr does","title":"Use speech to make the Word draft. Keep Word for Word work.","subtitle":"Choose Cloud for fast transcription without a local model download. Choose local processing for device-only or offline work. Then review the editable text in Word.","items":[{"image":{"label":"Man speaking beside a laptop at a desk surrounded by books and papers."},"title":"Generated text at the active Word cursor","body":"Put the cursor in a writable part of Word, use the desktop hotkey and speak. With auto-paste, OpenWhispr can paste generated text at that focused cursor. Otherwise, use the clipboard."},{"image":{"label":"Woman gesturing beside a laptop at a train window."},"title":"Cloud for speed, local for device-only work","body":"Cloud transcription is the fast option with no local model download and transcription audio discarded after processing. Local models run on your device and can work offline."},{"image":{"label":"Woman speaking beside a laptop and mug at a library table."},"title":"Word still owns formatting and review","body":"OpenWhispr does not control Word formatting, comments, Track Changes or Word commands. It supplies text. Use Word's own tools to format, comment, track and accept edits."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Word add-in required","value":"None","footnote":"OpenWhispr is a desktop app, not a native Word integration"},{"label":"Cost for unlimited local speech-to-text","value":"$$0","footnote":"on every desktop plan, with a local model"},{"label":"Setup estimate","value":"5 min","footnote":"from download to a first local dictation"}]},"howItWorks":{"label":"How it works","title":"A Word draft you can review before the next sentence slips away.","subtitle":"Choose a transcription path, start at the active Word cursor, then make Word's own review tools the final pass.","image":{"label":"Woman speaking and gesturing toward a laptop at a wooden table under a pendant lamp."},"steps":[{"title":"Download for free.","body":"OpenWhispr runs on macOS, Windows and Linux. Choose Cloud for no local model download or local for on-device and offline speech-to-text."},{"title":"Set the active Word cursor and speak.","body":"Click a writable place in Word, press the desktop hotkey and dictate. With auto-paste, generated text can be pasted at the active cursor. Otherwise, use the clipboard."},{"title":"Review the text in Word.","body":"Correct, format, comment and track edits with Word's own tools. OpenWhispr does not operate those Word features for you."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Before you choose a Word dictation method","title":"10 reasons Microsoft Word users choose OpenWhispr","subtitle":"This is speech-to-text for the document you already use, with clear boundaries around what happens at the cursor and what stays a Word job.","items":[{"title":"No Word add-in to install","body":"OpenWhispr is desktop dictation software. It does not add a ribbon, plugin or API connection to Word. It uses the active writable cursor when automatic pasting is enabled."},{"title":"One hotkey instead of a new drafting surface","body":"Start from the document you already have open. The desktop hotkey records your speech, then generated text can be pasted at the active Word cursor."},{"title":"Editable text, not a locked transcript","body":"With automatic pasting, text can reach your active cursor. Otherwise, use the clipboard. You can revise the draft in Word just as you would after typing it. Review remains important for names, specialist terms and meaning."},{"title":"Word retains its own controls","body":"Formatting, comments, Track Changes and native Word commands stay in Word. OpenWhispr does not control them or claim to edit Word as an integration."},{"title":"A custom dictionary for names and jargon","body":"Add names, terms and acronyms to OpenWhispr's custom dictionary as context hints for the speech model. It can improve recognition, but you still review the final Word text."},{"title":"Local speech-to-text works offline","body":"Choose a local speech model to process dictation on your device without an internet connection. That is useful when online speech input is not the right fit."},{"title":"Cloud when you want the fast setup","body":"OpenWhispr Cloud provides speed and accuracy without downloading a local model. Cloud transcription audio is sent only to transcribe, then immediately discarded."},{"title":"A clipboard option when you prefer it","body":"You can keep dictated text in your clipboard for manual placement. Automatic pasting is optional, which keeps the active-cursor workflow distinct from an app integration."},{"title":"Unlimited local dictation at no cost","body":"Local AI models are unlimited on every desktop plan, including Free. This makes it practical to test a real drafting and correction workflow before paying for Cloud features."},{"title":"Open source, so the mechanism is inspectable","body":"OpenWhispr is open source. You can inspect how the app handles dictation rather than treating a Word compatibility claim as a black box."}]},"faq":{"label":"Before you download","title":"The questions Word users ask first.","subtitle":"Straight answers about active-cursor text insertion, Word's own tools, privacy, local processing and what Free includes.","items":[{"question":"Is OpenWhispr a Microsoft Word integration?","answer":"No. OpenWhispr is a desktop dictation app, not a Word add-in, plugin, API integration or Word-specific feature. With auto-paste, it can paste generated text at the active, writable cursor. That cursor can be in Word. Otherwise, use the clipboard."},{"question":"How is this different from Word Dictate?","answer":"Word Dictate is Microsoft's native Word feature. OpenWhispr runs separately on your desktop. Its generated text can be pasted at an active Word cursor, but it does not become part of Word or gain native Word controls."},{"question":"Can OpenWhispr format Word text or use Track Changes?","answer":"No. It does not control Word formatting, comments, Track Changes or native Word commands. Use Word for those actions after the text arrives at the cursor."},{"question":"What does free speech-to-text include?","answer":"The Free plan includes unlimited local AI models and 2,000 words a week of OpenWhispr Cloud transcription. Local speech-to-text has no internet requirement. Pro adds unlimited OpenWhispr Cloud transcription, sync, mobile and API or MCP access."},{"question":"What happens to my audio?","answer":"With a local transcription model, processing stays on your device. With OpenWhispr Cloud, audio is sent only to transcribe it and is immediately discarded. Choose the path that fits the document and your working conditions."},{"question":"What if a name or specialist term is wrong?","answer":"Review the editable text in Word before you finish the document. You can add names, jargon and acronyms to OpenWhispr's custom dictionary as context hints, then correct anything the finished document still needs."}]},"reviews":{"label":"What users say","title":"Less typing. More control over where the text goes.","subtitle":"Three verbatim notes from people using OpenWhispr. They speak to the tool, not a Microsoft Word integration.","score":{"value":"1","label":"document cursor you choose","body":"With auto-paste, generated text can go to the active writable cursor. Word is one possible destination, not an integration surface. Otherwise, use the clipboard."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for Microsoft Word users","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Word users who want a useful free start with local speech-to-text and limited Cloud transcription"}]},{"parts":[{"text":"Word users who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Teams that also need unlimited meeting recordings and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Useful Free access, including unlimited local speech-to-text and limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings and chat over your data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Writing should not require leaving the document you already use.","paragraphs":["OpenWhispr started with a simple idea: say what you mean, then put the resulting text where you are already writing. In this workflow, that can be the active cursor in a Word document.","We build in public. Every release and every line of code is on GitHub, so anyone can inspect how the app handles dictation. When we say local dictation stays local, that is a property of the code you can verify.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"OpenWhispr team and community photo placeholder"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Keep Word as the workspace. Use speech for the first draft.","body":"Download OpenWhispr, put the cursor in Word and speak. With auto-paste, generated text can arrive at the active writable cursor. Otherwise, use the clipboard before your Word review.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Set up a hotkey, select a transcription path and test it in a Word document. Free local speech-to-text is available on macOS, Windows and Linux."},"byline":{"authorLabel":"Written by","author":"Joshua Padoa","aiDisclosure":"Drafted with AI assistance","reviewerLabel":"Reviewed and fact checked by","reviewer":"Wesley van der Hoop","updatedLabel":"Last reviewed","dateModified":"2026-09-10"}}},"useCaseGoogleDocs":{"breadcrumb":{"home":"Home","useCases":"Use Cases","googleDocs":"For Google Docs Users"},"metadata":{"title":"Google Docs App Speech To Text - OpenWhispr","description":"OpenWhispr is a free desktop speech-to-text app that pastes dictated text at your active Google Docs cursor. No Google Docs add-on required.","keywords":["google docs app speech to text","speech to text for google docs","google docs dictation app","voice to text google docs desktop","google docs speech to text app","dictate into google docs"]},"schema":{"name":"OpenWhispr","applicationCategory":"Desktop speech-to-text software for Google Docs users","operatingSystem":"macOS, Windows, Linux","description":"Open-source desktop speech-to-text software that turns speech into editable text at a focused cursor. It is not a Google Docs add-on or native integration. Free on macOS, Windows and Linux.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech to text for Google Docs users","title":"Press the hotkey. Get text in your Google Doc.","body":"On desktop, OpenWhispr turns your speech into editable text and pastes it at a focused Google Docs cursor. It is system-wide dictation, not a Google Docs add-on or native integration.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Man gesturing toward a laptop at an outdoor table overlooking city buildings."}},"trustBar":{"label":"Dictate into on desktop:","names":["Google Docs","Microsoft Word","Gmail","Notion","Slack","Obsidian"]},"pain":{"label":"The writing moment","title":"The cursor is ready. Your hands are not.","subtitle":"You want a paragraph in this document, not a transcript stranded in another app.","cards":[{"title":"A blank Google Doc and no easy way to type.","body":"OpenWhispr starts from a desktop hotkey. Speak the paragraph, then return to editable text at the focused document cursor."},{"title":"A recorder leaves you with a transfer job.","body":"The destination matters. Desktop dictation can paste text into the document in front of you, so you review the draft there instead of moving text from a separate transcript."},{"title":"The browser tool is not the same as mobile.","body":"Google Docs desktop voice typing and phone keyboard dictation are separate paths. This OpenWhispr page describes the desktop app, not a claim about dictating into Docs from mobile."},{"title":"An integration claim creates the wrong expectation.","body":"OpenWhispr does not need a Google Docs plugin to use the active cursor. It is a system-wide desktop app, not a native Google Docs integration."}]},"features":{"label":"Desktop dictation for the document in front of you","title":"Speak the draft. Review it in place.","subtitle":"Use a global hotkey while a writable document cursor is focused. Local transcription is unlimited on every plan.","items":[{"image":{"label":"Man wearing glasses gesturing toward a desktop monitor at a desk with papers."},"title":"A hotkey, then editable text at the cursor","body":"Focus the Google Docs document on desktop, hold your OpenWhispr hotkey, and speak. The app uses the focused editable cursor for its paste path."},{"image":{"label":"Woman wearing a blue headscarf gesturing toward a laptop at a bright office table."},"title":"Choose the processing route that fits the draft","body":"Local models run on your device and work offline. OpenWhispr Cloud offers top accuracy and speed without a local model download, then immediately discards transcription audio."},{"image":{"label":"Woman holding a pen over papers and gesturing beside a desktop monitor in an evening office."},"title":"A usable fallback when automatic paste cannot run","body":"If the desktop app cannot automatically paste, it copies the dictated text to the clipboard for a manual paste. That is a fallback, not a Google Docs integration."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Cloud transcription audio retained","value":"None","footnote":"Cloud audio is transcribed then immediately discarded. Local audio stays on your device."},{"label":"Cost for local desktop dictation","value":"$$0","footnote":"local dictation is unlimited on every plan, including Free"},{"label":"Desktop start","value":"Hotkey","footnote":"download, choose a model, focus a writable document"}]},"howItWorks":{"label":"From desktop hotkey to document","title":"Focus the doc. Speak the text. Review the result.","subtitle":"OpenWhispr uses system-wide desktop dictation. It does not install inside Google Docs or claim a native Docs connection.","image":{"label":"Man gesturing toward a desktop monitor at a wooden desk with a keyboard and mug."},"steps":[{"title":"Download for free.","body":"Install on macOS, Windows, or Linux. Local models are unlimited on every plan and can run offline."},{"title":"Focus Google Docs, then hold the hotkey.","body":"Keep a writable Google Docs cursor focused in your desktop browser. Speak the paragraph and OpenWhispr uses its paste path for that active cursor."},{"title":"Review the editable paragraph.","body":"Correct the document where you are already working. If automatic paste is unavailable, paste the copied text manually from the clipboard."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"The desktop route, clearly bounded","title":"10 reasons Google Docs users choose OpenWhispr","subtitle":"For people who want speech to become editable text in the active desktop document, without claiming a Google Docs add-on.","items":[{"title":"The cursor stays the destination.","body":"OpenWhispr is built to paste dictated text at a focused editable cursor. In Google Docs on desktop, that means the active document remains where you review the paragraph."},{"title":"No Docs add-on to install.","body":"The workflow relies on the desktop app's system-wide hotkey and paste behavior, not a Google Docs extension, sidebar, or native integration."},{"title":"Desktop and phone are different decisions.","body":"This is the desktop route for macOS, Windows, and Linux. Google Docs mobile voice paths and the paid OpenWhispr iPhone and iPad app are separate product contexts."},{"title":"Keep the document focused while you dictate.","body":"The desktop paste path is designed for a focused editable cursor. If automatic paste is unavailable, OpenWhispr copies the dictated text to the clipboard for a manual paste."},{"title":"Clipboard fallback preserves the text.","body":"When automatic pasting is unavailable, OpenWhispr writes the result to the clipboard so you can paste it manually into the document."},{"title":"Local works when the connection does not.","body":"Choose local models for on-device, offline transcription. You can keep drafting on the desktop even without an internet connection."},{"title":"Local desktop dictation is unlimited and free.","body":"Every plan includes unlimited local AI models. The Free plan's 2,000-word weekly limit is for OpenWhispr Cloud transcription, not local dictation."},{"title":"Cloud is the no-download route.","body":"OpenWhispr Cloud is the choice for top accuracy and speed without downloading local models. Cloud transcription audio is immediately discarded."},{"title":"A custom dictionary can help with document terms.","body":"Add recurring names, jargon, and acronyms to OpenWhispr's custom dictionary for sharper transcripts when your document uses specialized language."},{"title":"Pro adds cloud and cross-device workflow features.","body":"Pro includes unlimited OpenWhispr Cloud transcription, sync across devices, personal API access, MCP integration, and the iPhone and iPad app. None of those features claim a native Google Docs integration."}]},"faq":{"label":"Before you choose a route","title":"Questions about Google Docs speech to text.","subtitle":"Straight answers on desktop, mobile, privacy, cost, and the integration boundary.","items":[{"question":"Can I dictate directly into Google Docs with OpenWhispr?","answer":"On desktop, focus a writable cursor in the active Google Docs document, then use the OpenWhispr hotkey. The app's system-wide paste path is designed to put dictated text at that focused cursor. It is not a Google Docs add-on or native integration."},{"question":"What if automatic paste does not work?","answer":"OpenWhispr writes the dictated text to the clipboard. You can paste it manually into Google Docs. On macOS, automatic pasting requires Accessibility permission."},{"question":"Is this the same as Google Docs Voice typing?","answer":"No. Google Docs has its own desktop Voice typing feature. OpenWhispr is a separate system-wide desktop dictation app that can use a focused editable cursor in Google Docs."},{"question":"Does OpenWhispr dictate into Google Docs on my phone?","answer":"This page makes no claim that it does. The OpenWhispr iPhone and iPad app is included with Pro and above, but mobile dictation into Google Docs is not established by the desktop product sources used here."},{"question":"Can I use local speech to text offline?","answer":"Yes. Local AI models run entirely on your device without an internet connection and are unlimited on every plan. Google Docs itself may still need its usual connection to save and sync your document."},{"question":"What does Pro add for this workflow?","answer":"Free includes unlimited local models and 2,000 OpenWhispr Cloud words each week. Pro adds unlimited Cloud transcription, sync across devices, personal API access, MCP integration, and the iPhone and iPad app."}]},"reviews":{"label":"What people tell us","title":"A desktop app for words at the cursor.","subtitle":"These are general product reviews, not claims of a native Google Docs integration.","score":{"value":"Desktop","label":"System-wide dictation","body":"The relevant product behavior here is system-wide desktop dictation with paste at a focused editable cursor, plus a clipboard fallback."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for Google Docs users","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Desktop Google Docs users who want unlimited local dictation"}]},{"parts":[{"text":"People who want unlimited Cloud transcription and cross-device features"}]},{"parts":[{"text":"Teams that need unlimited meeting recordings and priority support"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local desktop dictation at no cost."}]},{"parts":[{"text":"Unlimited Cloud transcription plus sync, API, MCP, and mobile access"}]},{"parts":[{"text":"Everything in Pro plus unlimited meeting recordings and priority support"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"More: unlimited meeting recordings and chat over your data"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Contact us about Enterprise."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"The sentence belongs in the document you already opened.","paragraphs":["OpenWhispr started with a simple objection: getting your own sentences into the document faster should not mean moving a transcript between apps.","On desktop, its system-wide hotkey uses a focused editable cursor. That is useful in Google Docs without being a Docs add-on.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Open the doc. Press the hotkey. Start the paragraph.","body":"Download OpenWhispr for desktop and turn speech into editable text at a focused cursor. It is system-wide dictation, not a Google Docs add-on.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free to download on Mac, Windows and Linux."}}},"offerLanding":{"title":"你的创作者优惠","description":"领取 OpenWhispr Pro 前 3 个月八折优惠。","hero":"领取 3 个月八折优惠","nav":"领取折扣","bottom":"领取优惠","bottomTitle":"领取创作者优惠","offer":"前 3 个月享受八折优惠","inlinePrice":"前 3 个月每月 {discounted}，之后每月 {regular}。","priceUnavailable":"价格暂时不可用。"},"offerAccount":{"title":"创建账号或登录","priceReminder":"前 3 个月每月 {discounted}，<renewal>之后每月 {regular}。</renewal>","verificationMessage":"点击邮件中的验证链接，然后返回此处继续领取创作者优惠。","processing":"正在使用你的账号继续…","backToOffer":"返回优惠"},"offerCheck":{"title":{"checking":"正在核实优惠…","eligible":"你的创作者优惠","retryable":"无法确认你的优惠","ineligible":"此账号不符合条件","unavailable":"此优惠不可用","already_subscribed":"你已拥有 Pro","pending":"正在确认你的购买"},"creator":"你的创作者优惠","priceUnavailable":"价格暂时不可用。","signedIn":"当前登录账号","switchAccount":"切换账号","checkoutFailed":"无法打开结账页面。请重试，我们会再次核实优惠。","checkingBody":"正在核实此账号的优惠和付款安排。","product":"OpenWhispr Pro","plan":"个人套餐 · 按月 · {currency}","discount":"八折优惠","perMonth":"/ 月","terms":"前 3 次月付享优惠，之后为每月 {regular}。","today":"今天","trialEnds":"{date} · 试用结束","firstPayment":"首次付款","atCheckout":"结账时支付 {amount}","nextPayments":"接下来的 2 次月付","each":"每次 {amount}","renewal":"从第 4 次付款起","afterTrial":"折扣将在试用结束后开始。","tax":"价格以 {currency} 计。适用税费将在结账时显示。","opening":"正在打开安全结账页面…","retryCheckout":"重试结账","checkout":"继续结账","free":"继续使用 Free","download":"下载 OpenWhispr","retry":"重试","plans":"查看常规套餐","manage":"管理订阅","body":{"retryable":"优惠核实未完成。请重试以查看折扣价格。","ineligible":"此优惠仅适用于首次个人订阅。此账号不符合条件。","unavailable":"未找到此账号可用的创作者优惠。你仍可免费使用 OpenWhispr 或查看常规套餐。","already_subscribed":"你可以下载 OpenWhispr 并用此账号登录，或管理现有订阅。","pending":"结账已完成，正在确认访问权限。请下载 OpenWhispr 并用此账号登录。请勿重复购买。"},"returnPending":"现在即可下载。请用同一账号登录，订阅访问权限仍在核实中。请勿重复购买。","returnExpired":"此结账已过期，未产生任何扣款。你可以从优惠页面重新开始。","returnNotOwned":"此结账属于另一个账号。请用结账时使用的账号登录。","existingCheckout":"此账号已有另一个结账正在进行。请完成或关闭它，或稍后再试。","manageFailed":"无法打开订阅管理页面。请重试。"}},"now":"$undefined","timeZone":"UTC","children":"$L16"}] 17:I[89554,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"ThemeProvider"] 18:I[50103,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"GitHubStarsProvider"] 19:I[28473,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"ConsentProvider"] 1a:I[44149,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"SiteNavigation"] 1b:I[44149,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"SitePageStatus"] 1c:I[31494,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js","/_next/static/immutable/chunks/01pbao-w8op81.js"],"default"] 1d:I[44149,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"SiteFooter"] 1e:I[95338,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"default"] 1f:I[88667,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"AttributionCapture"] 20:I[24901,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"MetaPixel"] 21:I[21307,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndg</article></body></html>

1:"$Sreact.fragment" 2:I[39756,["/_next/static/immutable/chunks/0r3sg558voqp8.js"],"default"] 3:I[37457,["/_next/static/immutable/chunks/0r3sg558voqp8.js"],"default"] 4:I[22016,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js","/_next/static/immutable/chunks/1jjsyi7c0q8kj.js"],""] 7:I[97367,["/_next/static/immutable/chunks/0r3sg558voqp8.js"],"OutletBoundary"] 8:"$Sreact.suspense" b:I[97367,["/_next/static/immutable/chunks/0r3sg558voqp8.js"],"ViewportBoundary"] d:I[97367,["/_next/static/immutable/chunks/0r3sg558voqp8.js"],"MetadataBoundary"] f:I[68027,[],"default",1] :HL["/_next/static/immutable/chunks/0bupflxufswob.css","style"] :HL["/_next/static/immutable/chunks/0o4yun1hxa7er.css","style"] :HL["/_next/static/immutable/media/0c89a48fa5027cee-s.p.1ii7211jqpzi_.woff2","font",{"crossOrigin":"","type":"font/woff2"}] a:X 0:{"P":null,"c":["","zh-CN","download"],"q":"","i":false,"f":[[["",{"children":[["locale","zh-CN","d",["api","auth","providers","unsubscribe"]],{"children":["download",{"children":["__PAGE__",{},"$undefined","$undefined",4608]},"$undefined","$undefined",4608]},"$undefined","$undefined",4624]},"$undefined","$undefined",4624],[["$","$1","c",{"children":[null,["$","$L2",null,{"parallelRouterKey":"children","error":"$undefined","errorStyles":"$undefined","errorScripts":"$undefined","template":["$","$L3",null,{}],"templateStyles":"$undefined","templateScripts":"$undefined","notFound":[["$","html",null,{"lang":"en","suppressHydrationWarning":true,"children":["$","body",null,{"className":"flex min-h-screen flex-col items-center justify-center bg-background px-6 text-center","children":[["$","p",null,{"className":"text-7xl font-bold text-stone-200 select-none","children":"404"}],["$","h1",null,{"className":"mt-4 text-2xl font-bold text-foreground","children":"Page not found"}],["$","p",null,{"className":"mt-2 max-w-md text-muted-foreground","children":"The page you're looking for might have been moved, deleted, or never existed."}],["$","$L4",null,{"href":"/","className":"mt-6 inline-flex items-center rounded-lg bg-blue-600 px-5 py-2.5 text-sm font-medium text-white transition-colors hover:bg-blue-700","children":"Take me home"}]]}]}],[["$","link","0",{"rel":"stylesheet","href":"/_next/static/immutable/chunks/0bupflxufswob.css","precedence":"next","crossOrigin":"$undefined","nonce":"$undefined"}]]],"forbidden":"$undefined","unauthorized":"$undefined"}]]}],{"children":[["$","$1","c",{"children":[[["$","link","0",{"rel":"stylesheet","href":"/_next/static/immutable/chunks/0bupflxufswob.css","precedence":"next","crossOrigin":"$undefined","nonce":"$undefined"}],["$","link","1",{"rel":"stylesheet","href":"/_next/static/immutable/chunks/0o4yun1hxa7er.css","precedence":"next","crossOrigin":"$undefined","nonce":"$undefined"}],["$","script","script-0",{"src":"/_next/static/immutable/chunks/1ty5ydjjcibls.js","async":true,"nonce":"$undefined"}],["$","script","script-1",{"src":"/_next/static/immutable/chunks/0t-su5g1z5lkv.js","async":true,"nonce":"$undefined"}],["$","script","script-2",{"src":"/_next/static/immutable/chunks/21bcxxao7anzt.js","async":true,"nonce":"$undefined"}],["$","script","script-3",{"src":"/_next/static/immutable/chunks/2w6h9kjzbvct2.js","async":true,"nonce":"$undefined"}],["$","script","script-4",{"src":"/_next/static/immutable/chunks/1p2ndgirh2-it.js","async":true,"nonce":"$undefined"}],["$","script","script-5",{"src":"/_next/static/immutable/chunks/0_nv14y4sgdwj.js","async":true,"nonce":"$undefined"}],["$","script","script-6",{"src":"/_next/static/immutable/chunks/3c5c8c7dedp60.js","async":true,"nonce":"$undefined"}]],"$L5"]}],{"children":[["$","$1","c",{"children":[null,["$","$L2",null,{"parallelRouterKey":"children","error":"$undefined","errorStyles":"$undefined","errorScripts":"$undefined","template":["$","$L3",null,{}],"templateStyles":"$undefined","templateScripts":"$undefined","notFound":"$undefined","forbidden":"$undefined","unauthorized":"$undefined"}]]}],{"children":[["$","$1","c",{"children":["$L6",[["$","script","script-0",{"src":"/_next/static/immutable/chunks/1jjsyi7c0q8kj.js","async":true,"nonce":"$undefined"}]],["$","$L7",null,{"children":["$","$8",null,{"name":"Next.MetadataOutlet","children":"$@9"}]}]]}],{},null,false,null]},null,false,"$a"]},null,false,null]},null,false,"$a"],["$","$1","h",{"children":[null,["$","$Lb",null,{"children":"$Lc"}],["$","div",null,{"hidden":true,"children":["$","$Ld",null,{"children":["$","$8",null,{"name":"Next.Metadata","children":"$Le"}]}]}],["$","meta",null,{"name":"next-size-adjust","content":""}]]}],false]],"m":"$undefined","G":["$f",[]],"S":true,"h":null,"r":"$undefined","s":"$undefined","a":"$undefined","l":"$undefined","p":"$undefined","d":"$undefined","b":"CEsgTTzxQcjXQr2OALsbL"} 11:I[19598,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"GoogleTagManager"] a:C 10:T579,window.dataLayer=window.dataLayer||[];function gtag(){dataLayer.push(arguments);} gtag('consent','default',{ad_storage:'denied',ad_user_data:'denied',ad_personalization:'denied',analytics_storage:'denied',functionality_storage:'granted',personalization_storage:'denied',security_storage:'granted',region:["AT","BE","BG","HR","CY","CZ","DK","EE","FI","FR","DE","GR","HU","IE","IT","LV","LT","LU","MT","NL","PL","PT","RO","SK","SI","ES","SE","IS","LI","NO","GB","CH"],wait_for_update:500}); gtag('consent','default',{ad_storage:'granted',ad_user_data:'granted',ad_personalization:'granted',analytics_storage:'granted',functionality_storage:'granted',personalization_storage:'granted',security_storage:'granted'}); try{var _m=document.cookie.match(/(?:^|; )ow_consent=([^;]*)/);if(_m){var _v=JSON.parse(decodeURIComponent(_m[1]));if(_v&&_v.choices&&typeof _v.choices.marketing==='boolean'&&typeof _v.choices.analytics==='boolean'&&typeof _v.v==='string'&&/^\d{4}-\d{2}-\d{2}$/.test(_v.v)&&_v.v>='2026-06-01'){gtag('consent','update',{analytics_storage:_v.choices.analytics?'granted':'denied',ad_storage:_v.choices.marketing?'granted':'denied',ad_user_data:_v.choices.marketing?'granted':'denied',ad_personalization:_v.choices.marketing?'granted':'denied'});}}}catch(e){} if(navigator.globalPrivacyControl){gtag('consent','update',{ad_storage:'denied',ad_user_data:'denied',ad_personalization:'denied'});}5:["$","html",null,{"lang":"zh-Hans","dir":"ltr","className":"space_grotesk_e6988195-module__RNs2Mq__variable","suppressHydrationWarning":true,"children":[["$","head",null,{"children":[["$","script",null,{"dangerouslySetInnerHTML":{"__html":"$10"}}],["$","link",null,{"rel":"preconnect","href":"https://ddwl4m2hdecbv.cloudfront.net"}]]}],["$","body",null,{"className":"flex min-h-screen flex-col","children":[["$","$L11",null,{"gtmId":"GTM-NPK2WBC7"}],"$L12"]}]]}] 14:I[59302,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js","/_next/static/immutable/chunks/1jjsyi7c0q8kj.js"],"default"] 6:["$L13",["$","$L14",null,{}]] 13:["$","script",null,{"type":"application/ld+json","dangerouslySetInnerHTML":{"__html":"{\"@context\":\"https://schema.org\",\"@type\":\"BreadcrumbList\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"首页\",\"item\":\"https://openwhispr.com/zh-CN\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"下载\",\"item\":\"https://openwhispr.com/zh-CN/download\"}]}"}}] c:[["$","meta","0",{"charSet":"utf-8"}],["$","meta","1",{"name":"viewport","content":"width=device-width, initial-scale=1"}]] 15:I[75696,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"default"] 12:["$","$L15",null,{"formats":"$undefined","locale":"zh-CN","messages":{"common":{"nav":{"features":"功能","pricing":"定价","openMenu":"打开菜单","closeMenu":"关闭菜单","useCases":"使用案例","dictation":"语音输入","aiNoteTaker":"AI 会议记录","upload":"上传音频","downloadForFree":"免费下载"},"features":{"notepad":{"label":"AI 会议记录","description":"无需机器人的私密会议记录"},"aiChat":{"label":"AI 对话","description":"懂你会议内容的智能对话"},"audioUpload":{"label":"音频转文字","description":"在本地转录任何音频文件"},"dictation":{"label":"语音输入软件","description":"在任何应用中语音转文字"}},"footer":{"tagline":"开源的语音转文字工具，尊重你的隐私，由你掌控一切。","columns":{"product":"产品","resources":"资源","legal":"法律","followUs":"关注我们","company":"公司"},"links":{"aiNotepad":"AI 会议记录","aiChat":"AI 对话","audioUpload":"音频转文字","features":"功能","pricing":"定价","faq":"FAQ","models":"模型","documentation":"文档","trustCenter":"信任中心","comparisons":"产品对比","useCases":"应用场景","blog":"博客","changelog":"更新日志","privacy":"隐私政策","cookies":"Cookie 政策","terms":"服务条款","security":"安全","contact":"联系我们","download":"下载","dpa":"DPA","careers":"招聘"},"copyright":"© 2026 Gizmo Labs Inc.","madeWith":"用 {heart} 为开源社区打造","ariaLabel":"页脚","legalAriaLabel":"法律信息","socialLabel":"社交媒体","logoAlt":"OpenWhispr 徽标","postalAddress":"221 W 9th Street, Wilmington, Delaware 19801, USA"},"actions":{"github":"GitHub","download":"下载","starUsOnGitHub":"在 GitHub 上给我们加星","noAccountOfflineOpenSource":"无需帐户 · 可离线使用 · 永远开源","close":"关闭"},"localeSwitcher":{"label":"语言","select":"选择语言"},"useCases":{"medical":{"label":"医疗听写","description":"临床医生的私人听写"},"windows":{"label":"Windows","description":"Windows上的AI语音转文字"},"legal":{"label":"法律听写","description":"律师私人听写"},"macbook":{"label":"MacBook 听写","description":"macOS的AI语音转文字"}},"downloadButton":{"loading":"正在加载...","downloadOpenWhispr":"下载 OpenWhispr","downloadForPlatform":"下载 {platform} 版","allPlatforms":"适用于 macOS、Windows、Linux 和 iOS","platforms":{"macos":"macOS","windows":"Windows","linux":"Linux","unknown":"你的操作系统","macosAppleSilicon":"macOS (Apple Silicon)","macosIntel":"macOS (Intel)","ios":"iOS"},"getStartedOnIPhone":"在 iPhone 上开始"},"theme":{"toggle":"切换主题","light":"灯光模式","dark":"深色模式"},"siteMetadata":{"title":"OpenWhispr | 开源 AI 语音听写","description":"即时将语音转换为文字。完全私密、开源，由 OpenAI Whisper 和 NVIDIA Parakeet 提供支持。","imageAlt":"OpenWhispr — 私密、开源的 AI 语音输入与会议记录"},"adminMetadata":{"title":"管理后台 — OpenWhispr","description":"OpenWhispr 内部管理面板。"},"workspaceSsoMetadata":{"title":"工作区 SSO — OpenWhispr","description":"为你的 OpenWhispr 工作区配置单点登录。"}},"featureNotepad":{"metadata":{"title":"AI 会议记录：转写和总结会议 - OpenWhispr","description":"OpenWhispr 是在本地转写会议的 AI 会议记录工具。无需机器人加入通话，即可私密地获取转写文本、摘要和待办事项。","keywords":["AI 会议笔记","会议 AI 记事本","会议笔记应用","会议转录","AI 会议助手","开源会议笔记","私密会议笔记","Granola 替代品","Granola 开源替代品","无机器人会议笔记","AI 会议摘要","自动生成会议笔记","会议记录工具","AI 笔记应用","AI 会议纪要","会议笔记隐私","会议录制应用","AI 会议总结","会议待办事项","会议笔记整理"]},"breadcrumb":{"home":"首页","notepad":"记事本"},"schema":{"description":"会议 AI 记事本。转录、记笔记，并让 AI 优化它们。开源、从设计上即注重隐私，支持 macOS、Windows、Linux 和 iOS。","offerDescription":"免费且开源。另有可选的 Pro 套餐。","featureList":["AI 驱动的会议笔记优化","实时会议转录","无需会议机器人","自定义 AI 操作和模板","可搜索的会议历史","文件夹整理","跨平台支持","零数据留存","开源"]},"hero":{"title":"无需机器人的 AI 会议笔记","subtitle":"会后自动帮你整理的笔记工具。你只需随手记下要点，OpenWhispr 会在后台聆听，并将内容整理成清晰有序的摘要，包括决策、行动项和后续事项。","availability":"支持 macOS、Windows、Linux 和 iOS"},"howItWorks":{"eyebrow":"工作原理","title":"三步搞定。零配置。","subtitle":"无需插件、无需集成、无需会议机器人。只需打开应用，开始你的会议。"},"steps":[{"number":"01","title":"开始你的会议","description":"在 Zoom、Google Meet、Teams 或任何其他应用上开启会议。OpenWhispr 直接捕获你电脑的音频——没有机器人加入你的通话，也没人知道它正在运行。"},{"number":"02","title":"用你喜欢的方式记笔记","description":"随手写要点、零散想法，怎么顺手怎么来。OpenWhispr 在后台转录完整对话，让你专注于会议本身。"},{"number":"03","title":"获得清晰、结构化的笔记","description":"会议结束后，AI 会把你的粗略笔记和完整记录整理成清晰的会议纪要——决策、待办事项和关键要点，随时可分享。"}],"enhancedNotes":{"eyebrow":"AI 优化的笔记","title":"你写三个要点，剩下的交给 AI。","paragraph1":"会议中想到什么就记什么。会议一结束，OpenWhispr 就会把你的笔记和完整记录结合起来，整理成结构化的会议纪要——决策、待办事项和后续步骤。","paragraph2":"随时在原始笔记、完整记录和 AI 优化版本之间切换。任何内容都不会被覆盖——你的原始笔记始终在那里。"},"platforms":{"eyebrow":"无机器人、无插件","title":"适配每一款会议应用。","subtitle":"OpenWhispr 直接捕获你电脑的音频——没有机器人加入你的通话，无需安装插件，也无需配置集成。","footnote":"以及任何其他在你电脑上播放音频的应用。"},"aiActions":{"eyebrow":"AI 操作","title":"把你的笔记变成邮件、摘要和待办事项。","paragraph1":"在任何会议之后，对你的笔记运行 AI 操作——总结讨论、起草跟进邮件、提取所有待办事项，或用你团队的模板格式化所有内容。","paragraph2":"为每次会议后都要做的事创建你自己的自定义操作。省下重复性工作的时间，专注于真正重要的事。"},"organization":{"eyebrow":"井井有条","title":"为每个项目、团队和客户建文件夹。","paragraph1":"创建文件夹，按你的工作方式整理会议笔记。按项目、按团队、按客户分组——怎么合理怎么来。","paragraph2":"每次会议都会自动保存，轻松查找。再也不用在一长串笔记里翻来翻去。"},"search":{"eyebrow":"搜索","title":"在任何会议中找到任何内容。","subtitle":"在你所有的转录、笔记和摘要中搜索。即时找到讨论了什么、谁说了什么、决定了什么——哪怕是几个月前的内容。"},"privacy":{"eyebrow":"从设计上即注重隐私","title":"你的会议是你自己的事。","subtitle":"你的会议录音和笔记属于你自己。我们不存储你的音频，不留副本，也绝不会用你的数据训练模型。","points":["零数据留存——音频经处理后立即丢弃","我们绝不用你的数据训练模型，也不用它来改进我们的模型","你的笔记本地存储在你的设备上","开源——任何人都可以在 GitHub 上验证代码"],"cardTitle":"零数据留存","cardBody":"你的音频会被实时处理并立即丢弃——绝不存储录音，绝不留副本。你的笔记保存在你的设备上，我们的代码开源，每一项承诺你都可以自己验证。","cardCtaPrefix":"不用只听我们说。","cardCtaLink":"阅读源代码","cardCtaSuffix":"。"},"comparison":{"eyebrow":"为什么选 OpenWhispr","title":"处理会议笔记的更好方式。","subtitle":"拥有与其他工具相同的能力，但你的数据始终属于你。","headerFeature":"功能","headerOthers":"其他工具","rows":{"zeroRetention":"零数据留存","neverTrains":"绝不用你的数据训练","noBots":"无会议机器人","openSource":"开源","freeTier":"免费套餐","crossPlatform":"跨平台"},"values":{"freeTierOpenwhispr":"每月免费 5 小时","freeTierOthers":"有限","crossPlatformOpenwhispr":"macOS、Windows、Linux、iOS","crossPlatformOthers":"macOS、Windows"}},"faq":{"title":"常见问题","subtitle":"人们通常想了解的关于 AI 记事本的问题。"},"faqs":[{"question":"OpenWhispr 记事本是如何工作的？","answer":"OpenWhispr 在会议期间聆听你电脑的音频——没有机器人加入你的通话。你照常记笔记，会议结束后，AI 会把你的笔记和完整记录结合起来，生成结构化的会议纪要，包含决策、待办事项和关键要点。"},{"question":"会有机器人加入我的会议吗？","answer":"不会。OpenWhispr 直接从你电脑的扬声器和麦克风捕获音频。不会有机器人加入你的 Zoom、Google Meet、Teams 或任何其他通话。其他参会者永远不会知道你在录制。"},{"question":"它免费吗？","answer":"是的。OpenWhispr 免费且开源。你可以在自己的电脑上运行一切，无需账户。如果你想用云端 AI 实现更快的处理，还有可选的 Pro 套餐。"},{"question":"我的数据私密吗？","answer":"是的。OpenWhispr 从设计上即注重隐私。你的音频会被实时处理并立即丢弃——不存储任何内容，不留任何副本。我们绝不用你的数据训练模型。你的笔记本地存储在你的设备上，我们的代码完全开源。"},{"question":"支持哪些会议应用？","answer":"全部都支持。OpenWhispr 适配 Zoom、Google Meet、Microsoft Teams、Slack、Discord、Webex、FaceTime，以及任何其他在你电脑上播放音频的应用。无需插件或集成。"},{"question":"OpenWhispr 会存储我的录音吗？","answer":"不会。你的音频会被实时处理并立即丢弃。我们绝不存储录音，也绝不用你的数据训练 AI 模型。你的笔记保留在你的设备上。"},{"question":"支持哪些操作系统？","answer":"OpenWhispr 支持 macOS、Windows、Linux 和 iOS。"},{"question":"什么是 AI 操作？","answer":"AI 操作让你一键对笔记执行任务——比如总结会议、起草跟进邮件、提取待办事项，或用你团队的模板格式化笔记。你还可以创建自己的自定义操作。"}],"graphics":{"notesTransform":{"title":"冲刺规划","yourNotesLabel":"你的笔记","note1":"- 三月前交付语音流程","note2":"- 找设计团队聊新 UI","note3":"- 预算问题：问财务","enhancedLabel":"已优化","decisionsTitle":"决策","decisionsText":"语音流程是最高优先级——三月底前交付","actionItemsTitle":"待办事项","actionItemsText":"Gabriel：周五前就 Q2 预算与财务跟进","nextStepsTitle":"后续步骤","nextStepsText":"新编辑器设计评审——周二下午 2 点"},"aiActions":{"title":"AI 操作","actions":[{"name":"总结会议","desc":"要点和决策"},{"name":"起草跟进邮件","desc":"基于你的笔记撰写邮件"},{"name":"同步到模板","desc":"用团队模板格式化"},{"name":"列出待办事项","desc":"提取所有待办"},{"name":"创建你自己的...","desc":"自定义 AI 操作"}]},"folders":{"foldersLabel":"文件夹","folders":[{"name":"产品","count":8},{"name":"设计","count":3},{"name":"客户","count":5},{"name":"一对一","count":12}],"recentLabel":"最近","recent":[{"title":"冲刺规划","time":"10:30"},{"title":"产品同步","time":"9:15"},{"title":"设计评审","time":"昨天"}],"detailTitle":"冲刺规划","detailTime":"今天，上午 10:30","decisionsTitle":"决策","decisionsText":"语音流程三月底前交付。UI 改版推迟到流程上线后。","actionItemsTitle":"待办事项","actionItemsText":"Sarah：本次冲刺结束前完成前端迁移。Alex：API 重构 + 基准测试。"},"search":{"query":"预算讨论","results":[{"title":"Q2 预算规划","date":"3月8日","snippet":"已批准 $50k 用于新招聘。与财务跟进……"},{"title":"冲刺规划","date":"3月11日","snippet":"……需在周五前获得 Q2 预算的人员编制批准"},{"title":"团队站会","date":"3月5日","snippet":"预算评审推迟到下周，等待最终……"}]}}},"featureAiChat":{"meta":{"title":"AI 会议笔记与私密聊天 | OpenWhispr","description":"面向会议笔记的本地 AI 聊天：提问、获取摘要并提取行动项。完全在设备上运行，开源且免费。","keywords":["AI 会议助手","最佳 AI 会议助手","会议 AI 助手","与会议笔记聊天","向会议记录提问","会议问答","语音 AI 助手","私密 AI 助手","开源 AI 助手"]},"breadcrumbs":{"home":"首页","aiChat":"AI 聊天"},"structuredData":{"description":"了解你会议内容的 AI 聊天。通过语音或文字提问、获取摘要并提取行动事项。开源，从设计之初就注重隐私。","offerDescription":"免费且开源。可选 Pro 套餐。"},"hero":{"title":"已经了解你会议内容的 AI 聊天。","subtitle":"通过语音或文字询问你的工作。即时获得答案、摘要和行动事项，全部直接来自你的会议记录。","platforms":"支持 macOS、Windows、Linux 和 iOS"},"graphics":{"label":"AI 聊天"},"chatExchange":{"question":"本周的产品会议有哪些关键决策？","intro":"根据本周的 3 场会议：","points":["语音管线将于三月底交付——最高优先级","UI 改版推迟至管线完成之后","Q2 预算已获批——需在周五前提交招聘需求"],"followUpPlaceholder":"追问一个问题……"},"askAnything":{"eyebrow":"尽管问","title":"别再翻找笔记了，直接问就好。","paragraph1":"AI 聊天拥有你会议的完整上下文——转录、笔记和摘要。无论是询问决策、行动事项，还是某人说过的话，都能即刻得到答案。","paragraph2":"按住快捷键说出问题，或直接打字。无论哪种方式，答案都来自你真实的会议，而非泛泛而谈的 AI。"},"exampleQuestions":{"title":"你的会议，皆可问答。","subtitle":"任何你平时要翻遍笔记才能找到的内容，尽管问。","items":["上周有哪些关键决策？","列出分配给 Sarah 的所有行动事项","关于 API 时间线我们做了什么决定？","为今天的会议起草一封跟进邮件","总结 Q2 的预算讨论","谁对此次迁移提出了顾虑？"]},"lifecycle":{"eyebrow":"贯穿你的整个工作日","title":"每场会议前、中、后都能帮上忙。","subtitle":"AI 聊天不只用于会后。它能帮你提前准备、在通话中提出更明智的问题，并在会议结束后做好跟进。","stages":[{"label":"会前","title":"做好准备","desc":"询问过往会议内容，在下次通话前快速跟上进度"},{"label":"会中","title":"提出更好的问题","desc":"无需离开会议，即可实时查阅上下文"},{"label":"会后","title":"把事情办好","desc":"起草邮件、总结决策、规划后续步骤"}]},"contentGeneration":{"eyebrow":"把事情办好","title":"把对话转化为行动。","subtitle":"AI 聊天帮你撰写文档、跟进邮件和简报——全部基于会议中真正讨论过的内容。","items":["起草一封跟进邮件","总结会议内容","撰写 Slack 更新","创建行动事项清单","起草项目简报","撰写会议纪要","创建一份提案","起草进展更新"]},"folderContext":{"eyebrow":"文件夹上下文","title":"针对整个项目提问。","paragraph1":"将提问范围限定在特定文件夹——例如某个项目、团队或客户。AI 聊天会搜索该文件夹中的所有会议，找出规律、反复出现的话题和趋势。","paragraph2":"非常适合准备季度回顾、汇总反馈主题，或在离开一段时间后重新跟进某个项目。","scope":"在 Product 文件夹中","question":"哪些功能需求被提及得最多？","intro":"在 Product 文件夹的 12 场会议中：","items":["实时协作（被提及 8 次）","移动应用支持（被提及 6 次）","自定义模板（被提及 5 次）"]},"privacy":{"eyebrow":"从设计之初就注重隐私","title":"你的对话，始终属于你。","subtitle":"AI 聊天使用存储在你设备上的笔记。你的提问与答案始终保持私密——你可以自行选择使用本地还是云端 AI 模型。","bullets":["聊天记录本地存储在你的设备上","提供本地 AI 模型，实现完全离线的隐私保护","可选启用云端模型，获得更快的响应","开源——你能精确核实代码所做的一切"],"cardTitle":"你的数据，由你做主","cardBody":"使用本地模型运行 AI 聊天，实现完全隐私——任何数据都不会离开你的设备。或在需要更快、更强响应时选择云端模型。无论哪种方式，你的会议数据都绝不会存储在外部服务器上。","sourcePrefix":"无需只听我们一面之词。","sourceLink":"阅读源代码","sourceSuffix":"。"},"comparison":{"eyebrow":"为什么选择 OpenWhispr","title":"尊重你隐私的 AI 聊天。","subtitle":"懂上下文的 AI，与你的会议协同，而非与你的隐私为敌。","headerFeature":"功能","headerOthers":"其他产品","rows":[{"feature":"语音驱动的问答","openwhispr":true,"others":false},{"feature":"来自你会议的上下文","openwhispr":true,"others":true},{"feature":"从设计之初就注重隐私","openwhispr":true,"others":false},{"feature":"开源","openwhispr":true,"others":false},{"feature":"离线可用","openwhispr":true,"others":false},{"feature":"免费套餐","openwhispr":"每月免费 5 小时","others":"有限"},{"feature":"跨平台","openwhispr":"macOS、Windows、Linux、iOS","others":"macOS、Windows"}]},"faq":{"title":"常见问题","subtitle":"人们通常想了解的关于 AI 聊天的内容。"},"faqs":[{"question":"什么是 AI 聊天？","answer":"AI 聊天是一款内置助手，了解你会议的上下文。无论是询问讨论了什么、谁说了什么，还是接下来要做什么——都可通过语音或文字提问。它会直接从你的会议记录和转录中提取答案。"},{"question":"它是如何了解我会议内容的？","answer":"AI 聊天可以访问存储在你设备上的会议记录和转录。当你提问时，它会搜索你的笔记，给出准确、贴合上下文的答案。"},{"question":"我可以通过语音和它对话吗？","answer":"可以。按住快捷键，自然地说出你的问题，AI 聊天就会转录并作出回应。如果你愿意，也可以打字输入。"},{"question":"我的数据私密吗？","answer":"是的。你的笔记和聊天记录都保留在你的设备上。你可以选择使用本地 AI 模型（完全私密、离线可用）或云端模型（响应更快）。即便使用云端模型，你的会议数据也绝不会存储在外部。"},{"question":"我可以问哪些类型的问题？","answer":"任何关于你会议的问题——总结讨论、列出行动事项、查找谁说了什么、查阅决策、起草跟进邮件，或在你的全部会议历史中搜索。你也可以将问题范围限定在特定的文件夹或项目。"},{"question":"免费套餐包含此功能吗？","answer":"是的。AI 聊天可免费使用本地模型。云端驱动的模型在 Pro 套餐中提供，可获得更快、更强的响应。"}]},"featureAudioUpload":{"metadata":{"title":"音频转文字应用（免费下载）- OpenWhispr","description":"OpenWhispr 是免费的音频转文字应用，可在本地转写导入的音频文件，无需将录音上传到云端。","ogTitle":"音频上传与转录 — 隐私至上设计 | OpenWhispr","ogDescription":"上传任意音频或视频文件，几秒内获得精准转录文本。隐私至上设计，开源软件。","keywords":["音频转录","音频转文字","转录音频文件","本地音频转录","隐私音频转录","离线转录","MP3 转文字","WAV 转录","会议录音转录","语音备忘录转录","AI 音频转录","开源转录","whisper 转录","音频文件转笔记","本地转录录音"]},"breadcrumbs":{"home":"首页","audioUpload":"音频上传"},"schema":{"name":"OpenWhispr 音频上传","description":"上传任意音频文件，获得精准转录文本。隐私至上设计，可离线使用，开源软件。支持 99 多种语言。","offerDescription":"免费开源。可选 Pro 套餐。"},"hero":{"title":"在设备上将音频转成文字","subtitle":"将任何录音变成笔记。拖入音频文件，几秒内即可获得准确的转写文本，再用 AI 生成摘要、提取行动项，并将所有内容整理成笔记。","availability":"支持 macOS、Windows、Linux 和 iOS"},"howItWorks":{"eyebrow":"工作原理","title":"拖入、转录、搞定。","subtitle":"无需账号、无需注册、无需等待。只需拖入文件即可获得转录文本。"},"steps":[{"number":"01","title":"拖入文件","description":"拖放任意音频文件——会议录音、语音备忘录、讲座、访谈。支持 MP3、WAV、M4A 等众多格式。"},{"number":"02","title":"获得转录文本","description":"OpenWhispr 使用 AI 转录您的音频。默认在您的设备上完成处理——您的录音绝不会被上传到任何地方。"},{"number":"03","title":"整理成笔记","description":"您的转录文本会变成一条可搜索的笔记。运行 AI 操作即可总结要点、提取行动项，或用模板进行排版。"}],"transcriptGraphic":{"filename":"meeting-recording.mp3","transcriptLabel":"转录文本","enhancedLabel":"已增强","quote1":"“我觉得我们应该推进采用新的设计系统。Sarah，你能牵头前端迁移吗？”","quote2":"“没问题，我这个迭代就开始做组件梳理。Alex，API 那边你来负责可以吗？”","quote3":"“好的，我会把重构的范围理清楚，周三之前出一份 RFC。”","decisionsHeading":"决策","decisionsBody":"推进采用新的设计系统","actionItemsHeading":"行动项","actionItem1":"Sarah：本迭代完成组件梳理","actionItem2":"Alex：周三前提交 API 重构的 RFC"},"beyondTranscription":{"eyebrow":"不止于转录","title":"从原始录音到结构化笔记。","paragraph1":"不只是给你一堆文字。OpenWhispr 会把转录文本整理成有条理的笔记——决策、行动项、要点和后续步骤，全部源自实际所说的内容。","paragraph2":"对任意转录文本运行 AI 操作——总结要点、起草跟进邮件，或用团队模板进行排版。"},"formats":{"title":"你需要的每一种音频格式","subtitle":"上传来自任意设备或应用的录音。OpenWhispr 开箱即用，支持所有主流音频格式。","footnote":"另外还支持 FFmpeg 支持的所有格式。","list":[{"name":"MP3","popular":true},{"name":"WAV","popular":true},{"name":"M4A","popular":true},{"name":"WebM","popular":false},{"name":"OGG","popular":false},{"name":"FLAC","popular":false},{"name":"AAC","popular":false}]},"useCasesSection":{"eyebrow":"使用场景","title":"转录一切。","subtitle":"会议录音、语音备忘录、访谈、讲座——只要是音频，OpenWhispr 就能把它变成可搜索、有条理的笔记。"},"useCases":[{"title":"会议录音","description":"上传来自 Zoom、Teams 或任意平台的录音，获得包含决策和行动项的结构化会议纪要。"},{"title":"语音备忘录","description":"把随手录的语音笔记变成可搜索的文字。随时随地记录灵感，稍后再整理。"},{"title":"访谈与调研","description":"转录访谈、用户调研会话或焦点小组。提取关键引述和主题。"},{"title":"讲座与演讲","description":"把课堂录音或会议演讲转换成包含重点要点的学习笔记。"}],"languageGraphic":{"label":"支持的语言","total":"99 多种","more":"+ 另外 87 种","languages":["英语","西班牙语","法语","德语","葡萄牙语","日语","中文","韩语","意大利语","荷兰语","俄语","阿拉伯语"]},"multiLanguage":{"eyebrow":"多语言","title":"99 多种语言。自动识别。","paragraph1":"无需任何设置，即可转录数十种语言的音频。OpenWhispr 会自动识别语言并生成精准的转录文本。","paragraph2":"无论你的会议用的是英语、西班牙语、日语还是阿拉伯语——它都能直接搞定。"},"privacy":{"eyebrow":"隐私至上设计","title":"你的录音始终归你所有。","intro":"大多数转录服务会把你的音频上传到云端。OpenWhispr 默认在你的设备上处理录音——敏感音频无需离开你的电脑。","checklist":["音频文件默认在你的设备上处理","可选择开启云端处理以提升速度","转录文本本地存储——你拥有自己的数据","开源——任何人都可以在 GitHub 上验证代码"],"cardTitle":"敏感录音的理想之选","cardBody":"法律证词、医疗记录、财务通话、人力资源访谈——当你的录音包含敏感信息时，本地处理意味着它们绝不会接触任何外部服务器。","cardFootnotePrefix":"无需只听我们一面之词。","cardFootnoteLink":"阅读源代码","cardFootnoteSuffix":"。"},"comparison":{"eyebrow":"为何选择 OpenWhispr","title":"更好的音频转录方式。","subtitle":"精准转录，同时让你的录音保持私密。","featureHeader":"功能","openwhisprHeader":"OpenWhispr","othersHeader":"云端服务","rows":[{"feature":"隐私至上设计","openwhispr":true,"others":false},{"feature":"开源","openwhispr":true,"others":false},{"feature":"可离线使用","openwhispr":true,"others":false},{"feature":"无文件大小限制","openwhispr":true,"others":false},{"feature":"免费套餐","openwhispr":"每月免费 5 小时","others":"限量分钟数"},{"feature":"AI 笔记增强","openwhispr":true,"others":"需额外付费"},{"feature":"多语言","openwhispr":"99 多种语言","others":"因服务而异"}]},"faqSection":{"title":"常见问题","subtitle":"关于音频上传，大家通常想了解的内容。"},"faqs":[{"question":"我可以上传哪些音频格式？","answer":"所有常见格式——MP3、WAV、M4A、WebM、OGG、FLAC 和 AAC。无论你是用手机、笔记本还是其他任何设备录制的，都能正常使用。"},{"question":"转录的准确度如何？","answer":"非常准确。OpenWhispr 使用最先进的 AI 模型进行转录。它能很好地处理不同口音、背景噪音和多位说话人。你还可以在本地模型（私密、可离线使用）和云端模型（更快、更精准）之间选择。"},{"question":"有文件大小或时长限制吗？","answer":"没有人为限制。你可以上传任意时长的录音。文件越长，处理时间越久，但对你能转录的内容没有上限。"},{"question":"转录之后我能做什么？","answer":"你的转录文本会变成 OpenWhispr 中的一条笔记。从那里你可以对它运行 AI 操作——总结要点、提取关键点、用模板排版，或起草跟进邮件。它不只是一份转录文本，更是一个起点。"},{"question":"我的音频会被保密吗？","answer":"会的。OpenWhispr 采用隐私至上设计。使用本地处理时，你的音频绝不会离开你的设备。如果你为了更快的结果选择云端处理，你的音频会被实时处理并立即丢弃——不会存储任何内容。"},{"question":"支持哪些语言？","answer":"OpenWhispr 支持数十种语言的转录，包括英语、西班牙语、法语、德语、葡萄牙语、日语、中文等众多语言。它还能自动识别语言。"},{"question":"它免费吗？","answer":"免费。OpenWhispr 是免费开源软件。本地转录完全免费，且可离线使用。如果你想用云端驱动的 AI 获得更快、更精准的结果，还可选择 Pro 套餐。"}]},"compareHandy":{"metadata":{"title":"OpenWhispr vs Handy：开源听写工具对比","description":"比较 OpenWhispr 与 Handy 这两款免费开源听写应用，了解它们在 AI 模型、本地隐私和功能方面的差异，选出适合你的工具。","keywords":["OpenWhispr 与 Handy 对比","Handy 语音转文字","开源听写","免费语音听写应用","Whisper 听写应用","nvidia parakeet","本地语音转文字","隐私语音听写","跨平台听写","Handy 替代方案","Tauri 语音转文字","AI 听写对比"]},"breadcrumb":{"home":"首页","comparisons":"对比","handy":"对比 Handy"},"hero":{"title":"OpenWhispr 与 Handy 对比","subtitle":"两款免费、开源的听写应用，都将你的语音数据保留在本地。以下是它们在功能、灵活性和工作流方面的对比。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"快速了解","handyIntro":"是一款轻量级开源语音转文字应用，采用 Tauri 和 Rust 构建。它专注于把一件事做到极致：按下按键，开口说话，文字就会粘贴到任意文本框中。它同时支持 Whisper 和 Parakeet V3 模型、Silero 语音活动检测以及 GPU 加速，以极小的占用实现快速的本地转录。该项目已在 GitHub 上斩获超过 14,000 颗星，拥有活跃的社区。","openWhisprIntro":"拥有同样开源、隐私优先的基因，但将这一理念进一步延伸。除了本地转录之外，它还提供智能体模式（可用自定义 AI 系统提示词对听写内容进行后处理）、可选的云端转录（以获得更快结果），以及打磨精良的全系统听写体验。它支持 macOS、Windows、Linux 和 iOS，并提供免费版与可选的 Pro 版本。","chooseHandy":"如果你想要一款由 Rust 驱动、极简、仅本地的快速转录工具，且系统占用极小，请选择 Handy。","chooseOpenWhispr":"如果你想要 AI 后处理、云端转录选项以及功能更丰富的听写工作流，请选择 OpenWhispr。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"开源","openWhispr":{"status":"full","details":"在 GitHub 上完全开源"},"handy":{"status":"full","details":"采用 MIT 许可证，完全开源"}},{"feature":"隐私与本地处理","openWhispr":{"status":"full","details":"可 100% 本地运行，支持自带 API 密钥"},"handy":{"status":"full","details":"完全离线，不向云端传输任何数据"}},{"feature":"平台支持","openWhispr":{"status":"full","details":"macOS、Windows、Linux、iOS"},"handy":{"status":"full","details":"macOS、Windows、Linux"}},{"feature":"AI 模型多样性","openWhispr":{"status":"full","details":"OpenAI Whisper 和 NVIDIA Parakeet 模型"},"handy":{"status":"full","details":"Whisper 模型 + Parakeet V3"}},{"feature":"智能体模式 / AI 后处理","openWhispr":{"status":"full","details":"可用自定义系统提示词进行格式化、翻译等"},"handy":{"status":"none","details":"不支持"}},{"feature":"云端转录选项","openWhispr":{"status":"full","details":"OpenWhispr Cloud 或自带密钥（BYOK），快速出结果"},"handy":{"status":"none","details":"仅本地，无云端选项"}},{"feature":"语音活动检测","openWhispr":{"status":"partial","details":"基础静音检测"},"handy":{"status":"full","details":"Silero VAD 实现智能过滤"}},{"feature":"GPU 加速","openWhispr":{"status":"partial","details":"取决于模型和平台"},"handy":{"status":"full","details":"Whisper 支持 GPU（Intel、AMD、NVIDIA）"}},{"feature":"实时听写体验","openWhispr":{"status":"full","details":"全系统听写，带悬浮窗"},"handy":{"status":"full","details":"按住说话，自动粘贴"}},{"feature":"技术栈","openWhispr":{"status":"partial","details":"Electron + React + TypeScript"},"handy":{"status":"full","details":"Tauri（Rust）+ React，占用更小"}}],"whyOpenWhispr":{"title":"为什么选 OpenWhispr？","agentMode":{"title":"智能体模式与 AI 后处理：","body":"超越单纯的转录。用自定义系统提示词自动重新排版、翻译、总结或清理你的听写内容。"},"cloud":{"title":"云端转录选项：","body":"当本地处理不够快时，使用 OpenWhispr Cloud 或自带 API 密钥，即可获得闪电般的结果。"},"overlay":{"title":"精致的听写悬浮窗：","body":"全系统听写体验，配有可视化悬浮窗，可在你所有的应用中无缝运作。"},"pricing":{"title":"免费 + 可选 Pro：","body":"永久免费，包含全部核心功能。可选 Pro 版本（每用户每月 $8）提供无限云端转录、更多会议时长以及移动端 App。"}},"whyHandy":{"title":"为什么选 Handy？","architecture":{"title":"轻量级 Rust/Tauri 架构：","body":"采用 Tauri 和 Rust 而非 Electron 构建，因此二进制文件更小、内存占用更低、启动更快。"},"parakeet":{"title":"支持 Parakeet V3 模型：","body":"提供 NVIDIA 的 Parakeet V3 作为 Whisper 的替代方案，支持自动语言检测，仅靠 CPU 即可以约 5 倍实时速度运行。"},"vad":{"title":"先进的语音活动检测：","body":"采用 Silero VAD 实现智能静音过滤，只转录你真正说出的话。"},"community":{"title":"庞大社区与 MIT 许可证：","body":"超过 14,000 颗 GitHub 星，社区活跃。采用 MIT 许可证，灵活性最大化。"}},"pricing":{"title":"真实成本对比","ow":{"freeLabel":"永久免费","free1":"包含全部核心功能","free2":"本地处理","free3":"智能体模式与系统提示词","free4":"社区支持","proLabel":"Pro（可选）","pro1":"无限云端转录","pro2":"每月 20 小时会议录音","pro3":"移动端 App","pro4":"邮件支持"},"handy":{"freeLabel":"免费且开源","free1":"包含全部功能","free2":"仅本地处理","free3":"多种模型选择","free4":"社区支持","noPaidTier":"无付费版本","sponsorsLabel":"可选的 GitHub Sponsors 赞助","paid1":"没有任何高级功能被设限","paid2":"所有功能皆免费","paid3":"赞助以支持开发"},"callout":{"lead":"两款应用都免费：","body":"这是一次罕见的对比——成本并非区分因素。两款工具都是真正免费且开源的。问题在于：哪一套功能和架构更契合你的工作流。"}},"featureShowdown":{"title":"功能对决","bothHaveTitle":"两者都有的功能：","bothHave":["开源且永久免费","本地 Whisper 模型处理","隐私优先，无需云端","macOS、Windows 和 Linux","全局快捷键触发","在任意应用中可用","按住说话录音模式","多种 Whisper 模型尺寸"],"openWhisprExclusivesTitle":"OpenWhispr 独有：","openWhisprExclusives":["带自定义 AI 系统提示词的智能体模式","云端转录（OpenWhispr Cloud 或自带密钥）","跨应用的可视化听写悬浮窗","可选 Pro 版本，含无限云端转录与移动端 App"],"handyExclusivesTitle":"Handy 独有：","handyExclusives":["Parakeet V3 模型（仅 CPU，自动语言检测）","Tauri/Rust 架构（占用更小）","Silero 语音活动检测","广泛的 GPU 加速（Intel、AMD、NVIDIA）"]},"evidence":{"title":"基于事实的说明","body":"本对比基于厂商文档、定价页面和公开产品资料，而非道听途说的个人评价。"},"migration":{"title":"从 Handy 切换到 OpenWhispr","intro":"已经在用 Handy，又对 AI 驱动的后处理感到好奇？以下是试用 OpenWhispr 的方法：","step1":{"title":"下载 OpenWhispr","body":"免费下载。可选 Pro 版本（每用户每月 $8）提供无限云端转录、更多会议时长以及移动端 App。"},"step2":{"title":"设置你的快捷键","body":"配置与 Handy 中相同的按住说话快捷键，获得熟悉的使用体验。"},"step3":{"title":"试试智能体模式","body":"设置一个系统提示词，自动为你的听写内容排版、清理或翻译——这正是 Handy 所没有的功能。"},"callout":{"lead":"两个都留着！","body":"既然两款应用都免费且开源，你无需二选一。许多用户两个都装，针对不同工作流各取所长。"}},"bottomLine":{"title":"结论","intro":"这是一次势均力敌的对比。OpenWhispr 和 Handy 都是免费、开源、注重隐私的听写工具，在 macOS、Windows 和 Linux 上均可使用。两者都不会把你锁进订阅，也不会未经同意就将你的语音发送到云端。","teamHandy":{"title":"Handy 阵营","body":"你想要一款尽可能精简的工具，配备 Rust 驱动的后端、Parakeet V3 支持和出色的语音活动检测。你偏好仅本地处理，希望对系统的资源占用尽可能小。"},"teamOpenWhispr":{"title":"OpenWhispr 阵营","body":"你想要的听写不止于转录。智能体模式、云端处理选项、可视化听写悬浮窗以及不断壮大的功能集，让 OpenWhispr 成为追求 AI 增强语音输入的作家、开发者和专业人士的理想之选。"},"closing":"两个出色的开源项目，价值观相同却各有所长。最棒的是？两者都免费，你大可全都试一试。"},"finalCta":{"title":"准备好体验 AI 增强的听写了吗？","body":"获得你喜爱的、与 Handy 同样出色的本地转录，外加智能体模式、云端选项和精致的全系统悬浮窗。永久免费。","starButton":"在 GitHub 上给我们点星","footnote":"无需信用卡 • 离线可用 • 永久开源"},"viewAllComparisons":"查看所有听写应用对比"},"compareMacwhisper":{"metadata":{"title":"MacWhisper 替代方案：免费开源听写","description":"适合实时听写的 MacWhisper 替代方案。OpenWhispr 免费开源，支持 Mac、Windows、Linux 和 iOS。立即免费体验。","keywords":["MacWhisper 替代方案","OpenWhispr 与 MacWhisper 对比","跨平台 whisper 应用","开源听写","实时听写应用","whisper 转录","nvidia parakeet","免费听写软件","MacWhisper Windows 替代方案","MacWhisper Linux 替代方案"]},"breadcrumb":{"home":"首页","comparisons":"对比","macwhisper":"对比 MacWhisper"},"hero":{"title":"OpenWhispr 与 MacWhisper 对比","subtitle":"两款 AI 驱动的语音工具，两种截然不同的理念。实时听写遇上文件转录。","githubButton":"在 GitHub 上查看"},"quickTake":{"title":"快速了解","paragraph1":"由 Jordi Bruin 打造，是一款备受好评的 macOS 应用，专为转录音频和视频文件而生。拖入一段播客、会议录音或视频文件，即可获得排版精良的文字稿，附带说话人标注、时间戳和多种导出格式。它在所擅长的领域——录后转录——表现卓越，已获得超过 2,000 名用户的 4.5 星评分。","paragraph2":"解决的是另一个问题：可在任意应用中全系统运作的实时语音听写。按下快捷键，开口说话，文字便会出现在光标所在之处。它开源、跨平台（Mac、Windows、Linux、iOS）、永久免费——由 OpenAI Whisper 和 NVIDIA Parakeet 驱动。","chooseMacWhisper":"如果你需要转录已有的音频/视频文件，并要求说话人标注、时间戳和字幕导出，请选择 MacWhisper。","chooseOpenWhispr":"如果你想要在任意应用中实时听写、跨平台支持或开源的自由，请选择 OpenWhispr。"},"featureTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"主要用途","openWhispr":{"status":"full","details":"在任意应用中实时听写"},"macWhisper":{"status":"full","details":"文件转录与音频处理"}},{"feature":"开源","openWhispr":{"status":"full","details":"在 GitHub 上完全开源"},"macWhisper":{"status":"none","details":"专有软件，闭源"}},{"feature":"平台支持","openWhispr":{"status":"full","details":"macOS、Windows、Linux、iOS"},"macWhisper":{"status":"partial","details":"macOS（近期新增 iOS/iPad）"}},{"feature":"定价","openWhispr":{"status":"full","details":"永久免费，可选 $8/月 Pro 版本"},"macWhisper":{"status":"partial","details":"免费基础版，另有可选付费版本（价格因渠道而异）"}},{"feature":"隐私与本地处理","openWhispr":{"status":"full","details":"100% 本地，云端 API 支持自带密钥"},"macWhisper":{"status":"full","details":"在设备上本地运行 Whisper 处理"}},{"feature":"AI 模型选项","openWhispr":{"status":"full","details":"OpenAI Whisper 和 NVIDIA Parakeet 模型"},"macWhisper":{"status":"full","details":"多种模型，包括 WhisperKit"}},{"feature":"全系统听写","openWhispr":{"status":"full","details":"按住说话快捷键，随处可用"},"macWhisper":{"status":"partial","details":"具备听写模式，但主要以文件为核心"}},{"feature":"批量文件处理","openWhispr":{"status":"none","details":"专注于实时听写"},"macWhisper":{"status":"full","details":"拖放式批量转录"}},{"feature":"导出格式","openWhispr":{"status":"partial","details":"剪贴板粘贴、文本历史记录"},"macWhisper":{"status":"full","details":"SRT、VTT、HTML、纯文本、字幕"}},{"feature":"说话人识别","openWhispr":{"status":"none","details":"不支持"},"macWhisper":{"status":"full","details":"自动说话人识别（v12+）"}}],"whyOpenWhispr":{"title":"为什么选 OpenWhispr？","realTimeTitle":"实时听写：","realTimeText":"边说边看到文字即时出现在任意应用中。不只是事后转录文件。","crossPlatformTitle":"跨平台：","crossPlatformText":"在 Mac、Windows、Linux 和 iOS 上均可使用。MacWhisper 仅限 Apple 平台。","openSourceTitle":"开源的自由：","openSourceText":"可查看代码、修改功能、贡献改进。没有黑箱。","freeTitle":"永久免费：","freeText":"全部核心功能均免费。Pro 为可选项，每月 $8，图个方便。"},"whyMacWhisper":{"title":"为什么选 MacWhisper？","fileTranscriptionTitle":"文件转录：","fileTranscriptionText":"专为转录音频和视频文件而打造。拖入、放下，即得排版精良的文字稿。","speakerTitle":"说话人识别：","speakerText":"自动识别并标注说话人，文字稿按颜色区分。非常适合会议和访谈。","exportTitle":"丰富的导出选项：","exportText":"可导出为 SRT、VTT、HTML、纯文本和字幕格式。内置视频播放器，支持同步回放。","appleTitle":"Apple 生态：","appleText":"原生 macOS 设计，近期支持 iPhone 和 iPad。WhisperKit 模型针对 Apple Silicon 优化。"},"pricing":{"title":"真实成本对比","openWhispr":{"free":{"label":"永久免费","features":["在任意应用中实时听写","包含 Whisper 和 Parakeet 模型","通过系统提示词进行 AI 清理","跨平台支持"]},"pro":{"label":"Pro","features":["无限云端转录","每月 20 小时会议录音","移动端 App","邮件支持"]}},"macWhisper":{"free":{"price":"免费","label":"基础版","features":["基础 Whisper 转录","仅限较小的模型","本地处理"]},"pro":{"price":"Pro","label":"一次性买断","features":["全部模型，包括 WhisperKit","说话人识别","高级导出格式","语音活动检测"]}},"calloutLead":"工具不同，分工各异：","calloutText":"MacWhisper 对于转录工作流是一笔超值的投资。OpenWhispr 则免费提供实时听写——你或许两个都想要。"},"featureShowdown":{"title":"功能对决","sharedTitle":"两者共有的功能：","shared":["由 OpenAI Whisper 和 NVIDIA Parakeet 驱动","100% 设备本地处理","隐私优先架构","多语言支持（99+ 种）","提供免费版本","高精度转录"],"openWhisprExclusivesTitle":"OpenWhispr 独有：","openWhisprExclusives":["按住说话快捷键的实时听写","在任意文本框中全系统可用","跨平台（Mac、Windows、Linux、iOS）","开源且可自由 fork","用自定义系统提示词进行 AI 清理","OpenAI/Groq 云端 API 支持自带密钥"],"macWhisperExclusivesTitle":"MacWhisper 独有：","macWhisperExclusives":["批量音频/视频文件转录","自动说话人识别","字幕导出（SRT、VTT）","内置视频播放器，支持同步回放","文字稿自动翻译","支持 WhisperKit + Parakeet 模型"]},"evidence":{"title":"基于事实的说明","text":"本对比基于厂商文档、定价页面和公开产品资料，而非道听途说的个人评价。"},"migration":{"title":"为你的工作流加入实时听写","intro":"已经在用 MacWhisper 做文件转录？OpenWhispr 是实时听写的完美补充：","step1Title":"下载 OpenWhispr","step1Text":"免费下载。可选 Pro 版本（每用户每月 $8）提供无限云端转录、更多会议时长以及移动端 App。","step2Title":"设置你的快捷键","step2Text":"配置按住说话（长按或切换），并挑选你偏好的 Whisper 模型，在速度与精度之间取舍。","step3Title":"随处听写","step3Text":"用 OpenWhispr 在邮件、文档和聊天中实时听写——用 MacWhisper 做文件转录。两全其美。","calloutLead":"任意平台均可使用：","calloutText":"如果你也用 Windows 或 Linux，OpenWhispr 能把同样的听写体验带到你拥有的每一台设备上。"},"bottomLine":{"title":"结论","paragraph":"OpenWhispr 和 MacWhisper 是两款真正不同的工具，只是恰好采用了相似的 AI 语音引擎。MacWhisper 擅长录后转录——把音频文件转成带说话人标注、时间戳和字幕导出的文字。OpenWhispr 擅长实时听写——在你工作时，把你的想法直接说进任意应用。","teamMacWhisperTitle":"MacWhisper 阵营","teamMacWhisperText":"你经常转录播客、会议或视频文件。你需要说话人标注、字幕格式和批量处理。你身处 Apple 生态。","teamOpenWhisprTitle":"OpenWhispr 阵营","teamOpenWhisprText":"你想在任意应用中实时听写。你看重开源、需要跨平台支持，或想要一款可自定义、可扩展的免费工具。","closing":"许多用户两个都用——MacWhisper 用于转录，OpenWhispr 用于听写。它们彼此完美互补。"},"cta":{"title":"准备好体验实时听写了吗？","subtitle":"获得这款免费、开源的语音听写工具，可在每个平台的每个应用中使用。由 OpenAI Whisper 和 NVIDIA Parakeet 驱动。","githubButton":"在 GitHub 上给我们点星","footnote":"无需信用卡 • 离线可用 • 永久开源"},"compareHubLink":"← 将 OpenWhispr 与其他工具对比"},"compareHub":{"metadata":{"title":"对比 OpenWhispr — 看看我们的实力 | OpenWhispr","description":"诚实、详尽的听写应用对比。看看 OpenWhispr 与 SuperWhisper、WisprFlow、Dragon NaturallySpeaking、Otter.ai 等产品相比如何。","keywords":["听写应用对比","语音转文字对比","openwhispr 对比","最佳听写应用","听写软件对比","语音转文字替代方案"]},"breadcrumb":{"home":"首页","comparisons":"对比"},"hero":{"title":"对比 OpenWhispr","subtitle":"与每一款主流听写应用并排对比。功能、价格、隐私与取舍，一目了然。","description":"我们坚信透明。每一篇对比都涵盖功能、价格、隐私、平台支持以及真实的取舍——包括竞品确实更出色的地方。"},"card":{"versus":"对比 {name}","readFullComparison":"阅读完整对比 →"},"competitors":{"granola":{"name":"Granola","description":"基于云端的 AI 会议记事本，无需会议机器人即可记录，并可同步日历。","differentiator":"本地优先的开源替代方案，支持设备端说话人分离"},"meetily":{"name":"Meetily","description":"开源、本地优先的 AI 会议助手，基于 Whisper。","differentiator":"同样秉持本地优先理念，且已支持日历、说话人分离与笔记对话"},"otter":{"name":"Otter.ai","description":"基于云端的会议转录与笔记服务。","differentiator":"私密的本地听写，对比云端会议转录"},"macwhisper":{"name":"MacWhisper","description":"使用 Whisper 模型转录音频文件的 macOS 应用。","differentiator":"跨平台实时听写，对比仅限 Mac 的文件转录"},"superwhisper":{"name":"SuperWhisper","description":"精致的闭源听写应用，覆盖 macOS、Windows 与 iOS。","differentiator":"开源的本地掌控，对比闭源的托管体验"},"wisprflow":{"name":"WisprFlow","description":"具备情境感知自动编辑的闭源 AI 听写工具。","differentiator":"开源带来的自由，对比闭源带来的便利"},"willowvoice":{"name":"WillowVoice","description":"面向桌面与移动端的闭源云端语音写作应用。","differentiator":"本地开源听写，对比云端托管的语音写作"},"dragon":{"name":"Dragon Professional","description":"Nuance Dragon Professional 语音识别（前身为 NaturallySpeaking），现归微软所有。","differentiator":"免费开源 AI，对比售价 $699、仅限 Windows 的老牌听写软件"},"windowsSpeech":{"name":"Windows Speech Recognition","description":"Windows 内置的语音输入，具备基础听写功能。","differentiator":"强大的 AI 听写，对比基础的内置工具"},"appleDictation":{"name":"Apple Dictation","description":"macOS 与 iOS 内置的听写功能，自定义选项有限。","differentiator":"可定制的 AI 听写，对比功能受限的内置特性"},"talon":{"name":"Talon Voice","description":"为免手操作而设计的复杂语音控制系统。","differentiator":"简单的听写，对比复杂的语音编程"},"handy":{"name":"Handy","description":"基于 Tauri 的轻量级语音转文字应用，支持 Parakeet 与 Whisper。","differentiator":"功能完备的听写，对比轻量级的本地转录"},"voiceink":{"name":"VoiceInk","description":"原生 macOS 听写应用，具备屏幕感知的智能模式与强力模式。","differentiator":"跨平台的自由，对比精致的原生 macOS 体验"},"spokenly":{"name":"Spokenly","description":"macOS 与 iOS 听写应用，具备 Agent Mode 与免费的本地 Whisper 模型。","differentiator":"跨平台开源，对比精致的仅限 Apple 体验"},"typeless":{"name":"Typeless","description":"AI 语音键盘，具备智能编辑、去除口头禅与语气适配功能。","differentiator":"无限量离线听写，对比 AI 润色的云端语音输入"},"fathom":{"name":"Fathom","description":"云端 AI 会议笔记工具，提供慷慨的免费方案和自动摘要。","differentiator":"在设备上私密采集，而非让云端机器人录制会议"},"fireflies":{"name":"Fireflies.ai","description":"云端 AI 会议助手，机器人会加入通话，并提供 CRM 同步和分析。","differentiator":"在本地无机器人记录会议，而非由云端机器人上传通话"}},"whyCompare":{"title":"为什么要对比听写应用？","paragraph1":"选择语音转文字工具是一个因人而异的决定，取决于你的平台、隐私需求、预算与工作流。有些听写应用在 macOS 上表现出色，却把 Windows 和 Linux 用户抛在一旁。还有些必须依赖云端处理，会把你的语音数据发送到远程服务器。少数应用甚至把本该免费的功能锁进昂贵的订阅里。","paragraph2":"OpenWhispr 开源、跨平台、隐私优先——但这并不意味着它就适合每一个人。这些对比页面为你提供做出决定所需的信息。我们诚实地介绍每个竞品的长处，因为最好的工具，是最契合你需求的那一款。","paragraph3":"无论你是在寻找 Dragon NaturallySpeaking 的免费替代品、SuperWhisper 的跨平台替代方案，还是只想了解 AI 驱动的听写与内置语音识别相比如何，都能在上方找到详尽的拆解。"}},"compareAppleDictation":{"metadata":{"title":"Apple 听写替代方案 — Mac 版 OpenWhispr","description":"Apple 听写替代方案，支持本地 AI、离线隐私保护和开源代码。OpenWhispr 还提供智能体模式与跨平台支持，完全免费。","keywords":["Apple Dictation 替代方案","更好的 Mac 听写","升级 Mac 听写","Mac 听写应用","比 Apple Dictation 更好","Whisper 听写 Mac","nvidia parakeet","AI 听写 macOS","macOS 听写替代方案","语音转文字 Mac"]},"breadcrumb":{"home":"首页","comparisons":"对比","appleDictation":"对比 Apple Dictation"},"hero":{"title":"OpenWhispr 对比 Apple Dictation","subtitle":"Apple Dictation 内置且便捷。OpenWhispr 则为你带来 AI 驱动的精准度、智能体模式与彻底的掌控权。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"速览","appleIntro":"免费、内置于每一台 Mac，且在近期的 macOS 版本中有了显著提升。它如今支持约 29 种语言的设备端处理、19 种语言的自动标点，并能在任意文本框中使用、没有超时限制。对于快速、随意的听写，它确实相当扎实。","openWhisprIntro":"基于 OpenAI Whisper 与 NVIDIA Parakeet 打造——这是全球研究者与开发者所采用的顶尖语音识别技术。它支持 99 种以上的语言，提供多种模型规格以满足你对精度/速度的取舍需求，并新增了带自定义系统提示词的智能体模式，可用 AI 整理、重排或转换你听写的文字。此外，它还开源、跨平台，且永久免费。","stickWithApple":"继续使用 Apple Dictation，如果：你只需要为简短笔记和消息做基础听写。","upgradeToOpenWhispr":"升级到 OpenWhispr，如果：你想要更高的精准度、AI 文本整理、自定义提示词或跨平台支持。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"AI 模型质量","openWhispr":{"status":"full","details":"Whisper 与 Parakeet 模型，顶尖精准度"},"appleDictation":{"status":"partial","details":"Apple 自家语音模型，表现不错，但在技术内容上精度稍逊"}},{"feature":"智能体模式 / AI 整理","openWhispr":{"status":"full","details":"自定义系统提示词可重排、总结或整理文本"},"appleDictation":{"status":"none","details":"无 AI 后处理或文本转换"}},{"feature":"隐私与本地处理","openWhispr":{"status":"full","details":"借助 Whisper 与 Parakeet 实现 100% 本地处理，可选 BYOK 云端 API"},"appleDictation":{"status":"full","details":"视语言与 Apple 功能设置而定，提供设备端与服务器辅助两种模式"}},{"feature":"开源","openWhispr":{"status":"full","details":"在 GitHub 上完全开源"},"appleDictation":{"status":"none","details":"闭源专有"}},{"feature":"平台支持","openWhispr":{"status":"full","details":"macOS、Windows、Linux、iOS"},"appleDictation":{"status":"partial","details":"仅限 macOS 与 iOS"}},{"feature":"语言支持","openWhispr":{"status":"full","details":"借助 Whisper 与 Parakeet 模型支持 99 种以上语言"},"appleDictation":{"status":"partial","details":"语言与自动标点的覆盖范围因语言与地区而异"}},{"feature":"可定制性","openWhispr":{"status":"full","details":"快捷键、系统提示词、模型选择、BYOK，且可自由 fork"},"appleDictation":{"status":"partial","details":"仅可设置键盘快捷键与选择语言"}},{"feature":"自动标点","openWhispr":{"status":"full","details":"Whisper 与 Parakeet 在所有语言中原生添加标点"},"appleDictation":{"status":"partial","details":"支持的语言覆盖范围因地区与 Apple 功能可用性而异"}}],"whyOpenWhispr":{"title":"为什么要升级到 OpenWhispr？","agentMode":{"title":"智能体模式与系统提示词：","body":"AI 后处理可清除口头禅、重排文本、修正语法，或将听写内容转换为要点、邮件或代码注释。"},"models":{"title":"顶尖 AI 模型：","body":"可在 Whisper 与 Parakeet 模型之间选择。在技术词汇、口音与嘈杂环境下，它们都稳定地胜过内置听写。"},"crossPlatform":{"title":"跨平台：","body":"可在 macOS、Windows、Linux 与 iOS 上运行。你的工作流不再被锁定在 Apple 硬件上。"},"openSource":{"title":"开源且可自由改造：","body":"查看代码、添加功能、融入你的工作流。没有黑箱，也没有供应商锁定。"}},"whereAppleShines":{"title":"Apple Dictation 的过人之处","zeroSetup":{"title":"无需任何设置：","body":"你的 Mac 上早已自带。按下麦克风键或连按两次 Fn，即可开始说话。无需下载或配置。"},"osIntegration":{"title":"深度系统集成：","body":"在每一个文本框中都能原生使用。与 iOS、iPadOS 和 iCloud 无缝衔接，无需通过剪贴板粘贴。"},"free":{"title":"完全免费：","body":"随 macOS 附带，无需额外付费。没有账户、没有订阅、没有附加推销。"},"voiceCommands":{"title":"语音命令：","body":"内置命令可处理标点、大写、换行，并能按名称插入表情符号。"}},"pricing":{"title":"真实成本对比","ow":{"freeLabel":"永久免费","free1":"包含 Whisper 与 Parakeet 模型","free2":"智能体模式与系统提示词","free3":"99 种以上语言","free4":"macOS、Windows、Linux","proLabel":"Pro","pro1":"包含 Free 的全部功能","pro2":"无限量云端转录","pro3":"每月 20 小时会议录音","pro4":"移动应用","pro5":"邮件支持"},"apple":{"builtInLabel":"内置于 macOS","pro1":"基础听写","pro2":"自动标点（19 种语言）","pro3":"用于格式化的语音命令","con1":"无智能体模式或 AI 整理","con2":"无法选择模型","con3":"仅限 Apple 硬件"},"callout":{"lead":"两者都免费。","body":"区别在于你能得到什么：Apple Dictation 给你基础的语音转文字；OpenWhispr 给你一整套 AI 听写工具，具备 Whisper 与 Parakeet 的精准度、智能体模式，以及随心定制一切的自由。"}},"featureShowdown":{"title":"功能对决","bothHaveTitle":"两者都具备的功能：","bothHave":["设备端语音识别","可在任意文本框中使用","自动标点","听写无时长限制","免费使用","多语言支持"],"openWhisprExclusivesTitle":"OpenWhispr 独有：","openWhisprExclusives":["带自定义系统提示词的智能体模式","多种 AI 模型规格（Whisper 与 Parakeet）","99 种以上语言，标点完整","跨平台（macOS、Windows、Linux、iOS）","支持 BYOK 云端 API","开源且可自由 fork"],"appleDictationExclusivesTitle":"Apple Dictation 独有：","appleDictationExclusives":["内置于 macOS（无需设置）","原生文本框集成（无需剪贴板）","语音命令（大写、表情符号）","支持 iOS 与 iPadOS","歧义词高亮提示"]},"evidence":{"title":"有据可循的说明","body":"本对比基于厂商文档、价格页面与公开产品资料，而非道听途说的用户评价。"},"migration":{"title":"从 Apple Dictation 升级到 OpenWhispr","intro":"准备好体验更精准、AI 驱动的听写了吗？上手步骤如下：","step1":{"title":"下载 OpenWhispr","body":"免费下载。可选 Pro（每位用户每月 $8），享受无限量云端转录、更多会议时长以及移动应用。"},"step2":{"title":"设置你的快捷键","body":"配置一个按住说话的快捷键。许多用户会沿用之前用于 Apple Dictation 的那个 Fn 键。"},"step3":{"title":"选择 AI 模型","body":"在 OpenAI Whisper 与 NVIDIA Parakeet 之间选择。可从 “base” 起步，兼顾速度与精度，也可直接选择 “large” 以获得最佳质量。"},"step4":{"title":"试用智能体模式","body":"设置一个系统提示词，自动整理你的听写内容。这正是 Apple Dictation 所没有的功能。"},"callout":{"lead":"你可以两者并用。","body":"用 Apple Dictation 处理快速消息，需要精准度、格式化或 AI 整理时则使用 OpenWhispr。"}},"bottomLine":{"title":"结论","paragraph1":"Apple Dictation 是一款能干的内置工具，在近期的 macOS 版本中已大幅改进。对于基础听写——快速消息、简短笔记——它都能胜任。","paragraph2":"但如果你对语音输入有更高要求——技术内容上更高的精准度、AI 驱动的文本整理、自定义格式，或在 Windows 与 Linux 上工作的能力——OpenWhispr 正是你一直在寻找的升级。","keepApple":{"title":"继续使用 Apple Dictation","body":"你只在 Mac 上听写简短的文本和邮件，不需要 AI 整理或自定义格式。内置方案足以满足你的需求。"},"upgradeOpenWhispr":{"title":"升级到 OpenWhispr","body":"你想要 Whisper 与 Parakeet 级别的精准度、带自定义提示词的智能体模式、跨平台支持，或开源带来的透明。Apple Dictation 能做的它都能做，它做不到的也一并奉上。"},"closing":"Apple Dictation 是一个不错的起点。当你超越它时，OpenWhispr 就是去处。"},"finalCta":{"title":"准备好升级你的 Mac 听写了吗？","body":"获得 AI 驱动的听写，拥有 Whisper 与 Parakeet 的精准度、智能体模式与完整的自定义能力。永久免费、开源，处处可用。","starButton":"在 GitHub 上给我们点 Star","footnote":"无需信用卡 • 可离线使用 • 永久开源","viewAllComparisons":"查看所有对比"}},"compareDragon":{"metadata":{"title":"Dragon Dictation 替代方案 — 免费开源","description":"OpenWhispr 是 Dragon 的免费开源替代方案，可在 Mac、Windows、Linux 和 iOS 上进行本地私密听写，无需购买 699 美元的许可证。","keywords":["Dragon NaturallySpeaking 替代方案","Dragon Professional 替代方案","Dragon Professional Speech Recognition 替代方案","Nuance Dragon 替代方案","免费 Dragon 替代方案","Dragon 开源替代方案","Dragon 听写替代品","Dragon NaturallySpeaking 替代品","Dragon Home 停产替代方案","Dragon Professional v16 替代方案","Dragon Professional Anywhere 替代方案","免费语音听写软件","Whisper 听写","nvidia parakeet","开源听写","免费语音识别软件","Mac 版 Dragon 替代方案","Microsoft Dragon 替代方案"]},"breadcrumb":{"home":"首页","comparisons":"对比","dragon":"对比 Dragon Professional"},"hero":{"title":"OpenWhispr 对比 Dragon Professional Speech Recognition","subtitle":"Nuance Dragon 的免费开源替代方案。现代化的 AI 听写，在每个平台上都能运行——不只是 Windows。","note":"前身为 Dragon NaturallySpeaking——现为 Dragon Professional v16，通过 Nuance 归 Microsoft 所有。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"速览","dragonIntro":"（最初名为 Dragon NaturallySpeaking，现为 Dragon Professional Speech Recognition）在 25 年间定义了听写。它最初由 Dragon Systems 开发，后被 Nuance 收购，而 Microsoft 又在 2022 年以 197 亿美元买下了 Nuance。收购之后，Microsoft 将重心放在了医疗 AI 上——推出了面向临床医生的 Microsoft Dragon Copilot——而通用桌面产品则陷入停滞。Dragon Home 于 2023 年 2 月停产，且无任何替代品。Dragon Professional v16 仍以约 $699 出售，但自 2023 年发布以来再未有过重大更新。","openWhisprIntro":"是 Dragon Professional 的现代化开源替代方案。它基于 OpenAI Whisper 和 NVIDIA Parakeet 构建，完全免费，可在 macOS、Windows、Linux 和 iOS 上运行，并支持 99+ 种语言。对于数以百万计被弃于升级路径之外的 Dragon Home 用户——以及任何不愿为仅限 Windows 的听写支付 $699 的人来说——OpenWhispr 以零成本提供顶尖的识别准确度。","legacy":"Dragon 的传承：它开创了消费级听写，并在 25 余年间赢得了业界领先准确度的声誉。这份传承值得尊重。","reality":"如今的现实：Dragon Home 已死，Mac 版 Dragon 在 2018 年被砍，Professional 售价 $699，而 Microsoft 的重心是医疗 AI——而非桌面听写。OpenWhispr 免费、跨平台，且持续积极开发。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能","headerDragon":"Dragon Professional"},"comparisonData":[{"feature":"定价","openWhispr":{"status":"full","details":"永久免费，可选 $8/月 Pro"},"dragon":{"status":"none","details":"Dragon Professional v16：约 $699 一次性买断；Anywhere：约 $660/年"}},{"feature":"平台支持","openWhispr":{"status":"full","details":"macOS、Windows、Linux、iOS"},"dragon":{"status":"partial","details":"仅限 Windows（Mac 版已于 2018 年 10 月停产）"}},{"feature":"开源","openWhispr":{"status":"full","details":"在 GitHub 上完全开源"},"dragon":{"status":"none","details":"专有，闭源"}},{"feature":"消费者版本","openWhispr":{"status":"full","details":"人人免费，无任何限制"},"dragon":{"status":"none","details":"Dragon Home 已于 2023 年 2 月停产，无替代品"}},{"feature":"AI 技术","openWhispr":{"status":"full","details":"OpenAI Whisper 与 NVIDIA Parakeet 模型"},"dragon":{"status":"full","details":"Nuance 深度学习引擎"}},{"feature":"隐私与本地处理","openWhispr":{"status":"full","details":"100% 本地，你的数据绝不离开你的设备"},"dragon":{"status":"partial","details":"桌面版 v16：本地；Anywhere：基于云端"}},{"feature":"准确度","openWhispr":{"status":"full","details":"顶尖的 Whisper 与 Parakeet 模型，99+ 种语言"},"dragon":{"status":"full","details":"高达 99% 准确度，深度学习引擎"}},{"feature":"语音命令","openWhispr":{"status":"partial","details":"通过系统提示词进行 AI 整理，支持基础命令"},"dragon":{"status":"full","details":"高级自定义语音命令与宏"}},{"feature":"活跃开发","openWhispr":{"status":"full","details":"活跃的开源社区，频繁发布更新"},"dragon":{"status":"partial","details":"桌面版 v16 自 2023 年起停滞；云端产品有更新"}},{"feature":"多语言支持","openWhispr":{"status":"full","details":"99+ 种语言，支持自动检测"},"dragon":{"status":"partial","details":"语言包有限，需单独购买"}}],"whyOpenWhispr":{"title":"为什么选 OpenWhispr？","free":{"title":"永久免费：","body":"无 $699 授权费。无需订阅。免费方案下可自带 API 密钥使用云端模型。可选 Pro（$8/月），享无限云端转录与会议录音。"},"crossPlatform":{"title":"跨平台：","body":"可在 macOS、Windows、Linux 和 iOS 上运行。Dragon 桌面版仅限 Windows——Mac 版 Dragon 已于 2018 年 10 月停产，无法在现代 macOS 上运行。"},"modernAi":{"title":"现代 AI：","body":"由 OpenAI Whisper 与 NVIDIA Parakeet 驱动，支持 99+ 种语言。Dragon 的桌面引擎自 2023 年 v16 发布以来再未有过重大更新。"},"openSource":{"title":"开源：","body":"查看代码、修改功能、贡献改进。没有厂商锁定，也不会有 Dragon Home 用户所经历的那种终止支持的意外。"}},"dragonStrengths":{"title":"Dragon 的优势","macros":{"title":"语音命令与宏：","body":"经 25 余年打磨的深层系统级语音命令、自定义宏和自动文本规则。对于依赖复杂语音工作流的高级用户来说无可匹敌。"},"enterprise":{"title":"企业与行业解决方案：","body":"Dragon Legal、Dragon Medical One 和 Dragon Professional Anywhere 为组织提供专业词汇库与集中管理。"},"accuracy":{"title":"久经验证的准确度：","body":"数十年的打磨，开箱即用准确度高达 99%。语音配置文件会随时间适应你的说话方式，带来更好的效果。"},"windows":{"title":"深度 Windows 集成：","body":"应用专属语音命令、直接文本插入，以及用于批量处理音频文件的 Auto Transcribe Folder Agent。"}},"pricing":{"title":"真实成本对比","ow":{"freeLabel":"永久免费","free1":"无限本地听写","free2":"Whisper 与 Parakeet 模型","free3":"macOS、Windows、Linux","free4":"99+ 种语言","free5":"自带 API 密钥","proLabel":"Pro（可选）","pro1":"无限云端转录","pro2":"每月 20 小时会议录音","pro3":"邮件支持"},"dragonPro":{"heading":"Dragon Professional v16","oneTimeLabel":"一次性授权","pro1":"高级语音命令","pro2":"自定义词汇库","pro3":"Nuance 深度学习引擎","con1":"仅限 Windows","con2":"自 2023 年起无重大更新"},"dragonHome":{"label":"Dragon Home（已停产）","con1":"已于 2023 年 2 月停产","con2":"无升级至 v16 的路径","con3":"仅限 Windows"},"dragonAnywhere":{"subLabel":"Professional Anywhere（订阅制）","pro1":"基于云端，自动更新","pro2":"无需训练语音配置文件","pro3":"含移动应用","con1":"需要联网","con2":"音频发送至云端"},"dragonMobile":{"label":"Dragon Anywhere Mobile","pro1":"iOS 与 Android","con1":"仅限移动端","con2":"需要云端处理"},"callout":{"lead":"算笔账：","body":"Dragon Professional v16 售价 $699，仅限 Windows 授权，且不保证未来更新。Dragon Professional Anywhere 每年 $660——也就是 $82/月。OpenWhispr 免费，在每个平台上都能运行，并由前沿 AI 驱动。即便是 $8/月的 Pro，也比 Dragon 的订阅便宜 10 倍。"}},"landscape":{"title":"2026 年的 Dragon 产品全景","intro":"自 Microsoft 收购 Nuance 以来，Dragon 产品线已经分化。以下是如今仍存的产品和已停产的产品：","available":{"title":"仍在售","item1":"—— $699 桌面授权，仅限 Windows，最后更新于 2023 年","item2":"—— 约 $660/年云端订阅，定期收到更新","item3":"—— 专业法律听写","item4":"—— 面向医疗的云端临床文档","item5":"—— AI 临床助手（2025 年 3 月推出）"},"discontinued":{"title":"已停产","item1":"—— 已于 2023 年 2 月停产，无 v16 替代品","item2":"—— 已于 2018 年 10 月停产，最后版本 v6.0","item3":"—— 品牌于 2015 年退役，现为“Dragon Professional”","item4":"—— 自 2023 年 2 月起不再销售","item5":"—— 由 Dragon Medical One（云端）取代"}},"featureShowdown":{"title":"功能对决","bothHaveTitle":"双方都有的功能：","openWhisprExclusivesTitle":"OpenWhispr 独有：","dragonExclusivesTitle":"Dragon 独有：","bothHave":["高准确度语音识别","本地/离线处理选项","在任意应用中可用","连续听写","按键说话热键激活","自定义词汇库支持"],"openWhisprExclusives":["跨平台（macOS、Windows、Linux、iOS）","完全免费且开源","99+ 种语言，支持自动检测","用于文本整理与重排版的 AI 智能体模式","自带 API 密钥（OpenAI、Anthropic 等）","会议录音，支持自动检测","活跃的社区开发"],"dragonExclusives":["高级语音命令宏与自动文本","专业法律与医疗词汇库","企业管理控制台（Nuance Management Center）","用于批量处理的 Auto Transcribe Folder Agent","随时间学习的自适应语音配置文件","25 余年英语听写打磨"]},"evidence":{"title":"基于证据的说明","item1":"Dragon Professional v16 定价（约 $699）和 Dragon Professional Anywhere（约 $660/年）依据 Nuance 官方定价及授权经销商截至 2026 年 3 月的报价。","item2":"Dragon Home 停产经 Nuance 2023 年 2 月产品通告确认。Mac 版 Dragon 于 2018 年 10 月停产，最后版本 6.0。","item3":"Microsoft 于 2022 年 3 月以约 197 亿美元收购 Nuance Communications。面向医疗的 Microsoft Dragon Copilot 于 2025 年 3 月推出。","item4":"Dragon Professional v16 系统要求：Windows 10（1909+）或 Windows 11、Intel Core i3+ 或 AMD Ryzen 3+、至少 4 GB 内存。不支持 macOS 或 Linux。","item5":"本对比基于厂商文档、官方发布说明和公开产品信息——而非道听途说的用户评价。"},"migration":{"title":"从 Dragon 切换到 OpenWhispr","intro":"无论是 Dragon Home 在你身上停产、Mac 版 Dragon 将你抛下，还是你已厌倦为一个停滞不前、仅限 Windows 的产品支付 $699——以下是切换方法：","step1":{"title":"下载 OpenWhispr","body":"在 macOS、Windows 和 Linux 上免费。自带 API 密钥即可使用云端模型，无需额外费用。可选 Pro（$8/月），享无限云端转录与会议录音。"},"step2":{"title":"选择你的 AI 模型","body":"从 OpenAI Whisper 或 NVIDIA Parakeet 模型中挑选。先从“base”或“small”开始，在速度与准确度间取得良好平衡——模型会自动下载。"},"step3":{"title":"设置你的热键","body":"配置一个按键说话热键，就像 Dragon 的麦克风开关一样。按住录音，松开转录——在任意应用中全系统通用。"},"step4":{"title":"启用 AI 智能体模式（可选）","body":"使用搭配系统提示词的智能体模式，自动整理、重排版并改进你的转录内容——类似 Dragon 的格式化规则，但由现代 LLM 驱动。"},"callout":{"lead":"曾是 Dragon Home 用户？","body":"你的 v15 授权没有升级路径——Nuance 已于 2023 年 2 月停产 Dragon Home，Windows 10 也在 2025 年 10 月终止支持。OpenWhispr 面向未来、开源，并可在 Windows 11、macOS 和 Linux 上运行。"}},"bottomLine":{"title":"结论","intro":"Dragon NaturallySpeaking 当之无愧地赢得了它的传奇地位。数十年间，它就是市面上最好的听写软件。但 Microsoft 对 Nuance 的收购将重心转向了医疗 AI，让通用桌面听写陷入了维护模式。","dragonIf":{"title":"在以下情况，Dragon 仍然合适……","body":"你是依赖高级语音宏的 Windows 高级用户，需要专业的法律或医疗词汇库，或者你的企业已经部署了 Dragon 基础设施。Dragon Professional Anywhere 的云端模式也适合需要集中管理的组织。"},"openWhisprIf":{"title":"在以下情况，OpenWhispr 是更好的选择……","body":"你想要在任意平台上都能运行的免费现代化听写。如果你曾是 Dragon Home 或 Mac 版 Dragon 用户、需要 Linux 支持、看重开源，或者干脆拒绝为一个自 2023 年起就没有实质更新、仅限 Windows 的语音转文字支付 $699。"},"closing":"Dragon 推开了门。Whisper、Parakeet 和开源把它彻底敞开。听写的未来是免费、跨平台、由社区驱动的。"},"finalCta":{"title":"准备好迎接现代化听写了吗？","body":"加入成千上万已切换到 OpenWhispr 的前 Dragon 用户。同样出色的准确度，零成本，覆盖每个平台。","starButton":"在 GitHub 上为我们点 Star","footnote":"无需信用卡 • 离线可用 • 永久开源","viewAllComparisons":"← 查看全部对比"}},"compareGranola":{"metadata":{"title":"Granola 替代方案 — 本地开源 AI 会议笔记","description":"比较 OpenWhispr 与 Granola。免费获得本地 AI 会议笔记、设备端转录和说话人分离，完全开源。立即体验。","keywords":["Granola 替代方案","Granola ai 替代方案","私密会议记录","本地 ai 会议记录","Granola 开源替代方案","Granola 对比 OpenWhispr","隐私 ai 会议记事本","设备端会议转录","本地说话人分离","与会议记录对话","无机器人会议转录","linux 版 granola"]},"breadcrumb":{"home":"首页","comparisons":"对比","granola":"对比 Granola"},"hero":{"title":"OpenWhispr：私密、开源 Granola 替代方案","subtitle":"正在寻找 Granola 替代方案来保证您的会议私密性？ OpenWhispr 为您提供相同的人工智能会议笔记工作流程，但转录在您的设备上本地运行，并且代码完全开源。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"速览","granolaIntro":"是一款为连轴会议打造的精致 AI 记事本。它无需机器人即可捕获你的系统音频、在云端转录、用 AI 增强你的笔记，并让你跨多场会议对话。它与 Google Calendar 和 Outlook 同步，并集成 Slack、Notion、HubSpot、Attio 和 Zapier。","openWhisprIntro":"是私密的开源 Granola 替代方案。它为您提供相同的会议工作流程，但管道在您的设备上运行。音频使用 Whisper 或 Parakeet 本地转录，说话人分离在本地运行，并且默认情况下您的笔记存储在本地。它还兼作系统范围的听写工具。","chooseGranola":"如果你符合以下情况，选 Granola：你深耕于云优先的工作流，想要最精致的商业体验，且不介意会议音频离开你的机器。","chooseOpenWhispr":"如果满足以下条件，请选择 OpenWhispr：您想要 Granola 的所有功能，但具有本地转录、设备端说话人分离和开源自由。"},"privacy":{"title":"隐私，并排对比","openWhisprPoints":["音频在你的设备上捕获并转录","说话人分离在本地运行","笔记默认存储在本地","可选的云端同步需主动开启且经过加密","免费层云端转录零数据留存","开源——你可以验证每一项说法"],"granolaPoints":["音频上传至云端转录","AI 处理在 Granola 的服务器上进行","转录文本存储于服务器端，附自动删除策略","闭源——你无法审计整条流程","数据控制取决于套餐层级与管理员设置"]},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"隐私与数据控制","openWhispr":{"status":"full","details":"本地优先：音频与转录默认在你的设备上运行"},"granola":{"status":"none","details":"在 Granola 的服务器上进行云端转录与 AI 处理"}},{"feature":"开源","openWhispr":{"status":"full","details":"在 GitHub 上完全开源——可审计每一行代码"},"granola":{"status":"none","details":"专有，闭源"}},{"feature":"音频绝不离开你的设备","openWhispr":{"status":"full","details":"使用 Whisper 与 Parakeet 模型 100% 在设备端转录"},"granola":{"status":"none","details":"音频被上传并在云端处理"}},{"feature":"本地说话人分离","openWhispr":{"status":"full","details":"谁说了什么的识别完全在你的机器上运行"},"granola":{"status":"none","details":"无设备端说话人分离——处理在云端进行"}},{"feature":"无机器人会议捕获","openWhispr":{"status":"full","details":"直接捕获系统音频——不会有机器人加入你的通话"},"granola":{"status":"full","details":"直接捕获系统音频——不会有机器人加入你的通话"}},{"feature":"AI 会议记录与摘要","openWhispr":{"status":"full","details":"自动生成笔记、待办事项和自定义模板"},"granola":{"status":"full","details":"自动生成笔记，附可定制模板"}},{"feature":"与你的笔记和数据对话","openWhispr":{"status":"full","details":"跨你的所有笔记和转录文本提问"},"granola":{"status":"full","details":"在单场及跨场会议中进行 AI 对话"}},{"feature":"日历集成","openWhispr":{"status":"full","details":"Google Calendar 同步即将到来的会议"},"granola":{"status":"full","details":"Google Calendar 与 Microsoft Outlook 同步"}},{"feature":"通用语音听写","openWhispr":{"status":"full","details":"在任意应用或文本框中按键说话听写"},"granola":{"status":"none","details":"并非听写工具——专注于会议捕获"}},{"feature":"平台支持","openWhispr":{"status":"full","details":"macOS、Windows、Linux、iOS"},"granola":{"status":"partial","details":"macOS、Windows、iPhone——无 Linux 或 Android"}},{"feature":"离线可用","openWhispr":{"status":"full","details":"使用本地模型，无需联网即可完整转录与记录"},"granola":{"status":"none","details":"需要联网——云端处理"}},{"feature":"免费层","openWhispr":{"status":"full","details":"无限本地转录，每月 5 小时会议录音"},"granola":{"status":"partial","details":"Basic 层会议历史有限"}}],"whyOpenWhispr":{"title":"为什么选 OpenWhispr？","localAudio":{"title":"你的音频绝不离开你的设备：","body":"转录用 Whisper 或 Parakeet 在设备端运行。无云端上传、无第三方服务器、无数据经纪关系。你的会议属于你。"},"diarization":{"title":"本地说话人分离：","body":"知道谁说了什么——完全在你的机器上计算。Granola 不提供设备端说话人分离。"},"chat":{"title":"与你的数据对话：","body":"跨每一场会议、笔记和转录提问。智能体模式可从你的笔记、日历和网络中获取信息，起草后续跟进。"},"openSource":{"title":"开源：","body":"在 GitHub 上查看、复刻并审计每一行代码。精确验证你的数据去向——无需盲目信任。"},"platforms":{"title":"Linux + Windows + macOS：","body":"三大桌面平台上均为一流的原生应用。Granola 仅提供 macOS 和 Windows 版本。"},"free":{"title":"对个人免费：","body":"永久无限本地转录。Granola 的 Business 层为每位用户 $14/月。"}},"whyGranola":{"title":"为什么选 Granola？","ux":{"title":"非常精致的用户体验：","body":"Granola 在视觉设计和会议模板的易用性上投入巨大。这款产品用起来感觉很棒。"},"calendar":{"title":"Outlook + Google Calendar：","body":"开箱即支持 Google 和 Microsoft 日历。OpenWhispr 目前支持 Google Calendar。"},"integrations":{"title":"深度云端集成：","body":"付费层提供 Attio、HubSpot、Affinity、Notion、Slack 和 Zapier。如果你的团队重度依赖 CRM，这一点很重要。"},"admin":{"title":"团队管理控制：","body":"Enterprise 提供 SSO、集中计费、组织级留存策略和管理员共享控制。"}},"pricing":{"title":"真实成本对比","ow":{"freeLabel":"免费——永久","free":["无限本地转录","本地说话人分离","每月 5 小时会议录音","100+ 种语言","离线可用","macOS、Windows、Linux"],"proLabel":"Pro","pro":["包含 Free 的全部功能","每月 20 小时会议录音","无限云端转录","移动配套应用"],"businessLabel":"Business","business":["无限会议录音","智能体模式","与你的数据对话"]},"granola":{"basicLabel":"Basic","basic":["✓ 会议历史有限","✓ 会议内 AI 对话","✓ 共享文件夹","✗ 无无限笔记"],"businessLabel":"Business","business":["无限会议记录","高级集成","集中计费","MCP 访问"],"enterpriseLabel":"Enterprise","enterprise":["SSO","优先支持","组织级留存控制","管理员控制"]},"callout":{"lead":"云端与本地的成本对比：","body":"Granola 的 Business 层为每位用户每年 $168。OpenWhispr Pro 为 $96/年，提供同样的会议工作流——而且你的数据绝不离开你的机器。"}},"featureShowdown":{"title":"功能对决","bothDoWellTitle":"双方都做得好的：","bothDoWell":["无机器人会议捕获","AI 增强的会议记录","可定制的笔记模板","与你的会议数据对话","同步即将到来的会议日历","待办事项提取"],"openWhisprExclusivesTitle":"OpenWhispr 独有：","openWhisprExclusives":["100% 本地转录（音频从不上传）","本地说话人分离（在设备端识别谁说了什么）","在 GitHub 上完全开源","端到端离线可用","原生 Linux 支持","在任意应用中按键说话语音听写","为 OpenAI、Anthropic、Groq、Ollama 自带密钥（BYOK）"],"granolaExclusivesTitle":"Granola 独有：","granolaExclusives":["原生 Microsoft Outlook 日历支持","深度 CRM 集成（Attio、HubSpot、Affinity）","付费层的 Zapier 工作流","成熟的团队管理与 SSO 控制","iPhone 配套应用（现已上线）"]},"evidence":{"title":"基于证据的说明","bodyPrefix":"本对比基于 Granola 截至 2026 年的公开定价页面、帮助中心和产品营销材料。厂商文档、套餐内容和平台支持可能变动——最新信息请参阅","bodySuffix":"。"},"migration":{"title":"从 Granola 切换到 OpenWhispr","intro":"如果你喜欢 Granola 的工作流，但希望会议音频留在你的机器上，切换大约只需五分钟：","step1":{"title":"下载 OpenWhispr","body":"在 macOS、Windows 和 Linux 上免费下载。无需账号即可开始。"},"step2":{"title":"挑选你的转录模型","body":"选择一个适合你硬件的本地 Whisper 或 NVIDIA Parakeet 模型——或自带 OpenAI/Groq 密钥，使用由你掌控的云端转录。"},"step3":{"title":"连接 Google Calendar","body":"可选——同步即将到来的会议，让笔记自动关联到正确的日程上。"},"step4":{"title":"启用本地说话人分离","body":"开启说话人识别，让你的会议记录标明谁在何时发言——全部在你的设备上计算。"},"step5":{"title":"与你的笔记对话","body":"跨你捕获的每一场会议提问。用智能体模式起草后续跟进，以你的笔记、日历和网络作为上下文。"}},"bottomLine":{"title":"结论","intro":"Granola 和 OpenWhispr 在同样的会议工作流上竞争：从你的电脑捕获音频，让 AI 增强你的笔记，并让整份转录文本可对话。真正的区别在于这些工作发生在何处。","teamGranola":{"title":"Granola 阵营","body":"你的团队已经在用 HubSpot/Attio/Notion，并想要最精致的商业体验。云端处理不是障碍。"},"teamOpenWhispr":{"title":"OpenWhispr 阵营","body":"你想要 Granola 的工作流，但不要 Granola 的云端。你看重本地处理、本地说话人分离、开源、Linux 支持，以及自带密钥的自由。"},"closing":"最好的会议记事本，是那个不会让你用隐私换智能的。OpenWhispr 两者兼得。"},"finalCta":{"title":"准备好迎接私密、本地的 AI 会议记录了吗？","body":"别再把你的会议上传到云端。OpenWhispr 为你提供 AI 笔记、说话人分离、日历同步，以及与你的数据对话——全部在你的机器上运行。","starButton":"在 GitHub 上为我们点 Star","footnote":"无需信用卡 • 离线可用 • 永久开源","viewAllComparisons":"查看全部对比"}},"compareMeetily":{"metadata":{"title":"Meetily 替代方案：本地开源 AI 笔记","description":"Meetily 替代方案：OpenWhispr 本地优先且开源，支持日历同步、说话人分离和笔记对话。立即免费体验。","keywords":["Meetily 替代方案","Meetily 对比 OpenWhispr","开源会议助手","本地会议转录","本地说话人分离","开源会议记录","日历集成会议记录","与会议记录对话","隐私会议助手","设备端会议转录","开源 AI 会议记事本","无机器人会议记录"]},"breadcrumb":{"home":"首页","comparisons":"对比","meetily":"对比 Meetily"},"hero":{"title":"OpenWhispr 对比 Meetily","subtitle":"两款开源、本地优先的 AI 会议助手。看看 OpenWhispr 如何加入日历同步、本地说话人分离和与笔记对话——这些功能今天就能用,而非停留在路线图上。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"快速了解","meetilyIntro":"是一款开源、本地优先的 AI 会议助手。它使用本地 Whisper 模型录制、转录并总结会议,可选搭配本地 LLM 或 BYOK 生成摘要。你的数据永不离开设备。它采用 MIT 许可证,持续活跃开发,且对个人使用免费。","openWhisprIntro":"与 Meetily 一样坚守本地优先、开源的理念——并已交付 Meetily 还在路线图上规划的功能。日历集成、本地说话人分离和与笔记对话今天就能使用。OpenWhispr 同时也是一款完整的语音听写应用:在桌面任何文本框中一键说话即可输入。","chooseMeetily":"选择 Meetily,如果:你想要一款极简、采用 MIT 许可证的会议录制工具,并愿意等待日历同步、说话人分离和对话等功能。","chooseOpenWhispr":"选择 OpenWhispr,如果:你想要同样的本地优先保障,外加完整功能集——日历同步、本地说话人分离、与数据对话以及通用语音听写——而且这些都已交付。"},"sharedValues":{"title":"我们的共识","body":"OpenWhispr 和 Meetily 怀有相同的根本承诺——我们认为这是好事。本地优先、开源的会议记录这一品类,因我们双方共同推动而更加出色。","items":["音频永不离开你的设备","开源——流程可审计","基于 Whisper 的本地转录","如有需要,可 BYOK 接入云端 LLM","免费社区版","无机器人的系统音频捕获"]},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"隐私与数据控制","openWhispr":{"status":"full","details":"本地优先:音频和转录默认在你的设备上运行"},"meetily":{"status":"full","details":"本地优先:所有处理均在设备上完成"}},{"feature":"开源","openWhispr":{"status":"full","details":"在 GitHub 上完全开源"},"meetily":{"status":"full","details":"采用 MIT 许可证开源"}},{"feature":"音频永不离开你的设备","openWhispr":{"status":"full","details":"使用 Whisper 和 Parakeet 进行设备端转录"},"meetily":{"status":"full","details":"使用 Whisper 模型进行设备端转录"}},{"feature":"本地说话人分离","openWhispr":{"status":"full","details":"谁说了什么的识别如今完全在你的机器上运行"},"meetily":{"status":"none","details":"列在路线图上——目前尚不可用"}},{"feature":"与你的笔记和数据对话","openWhispr":{"status":"full","details":"今天就能跨所有笔记和转录内容提问"},"meetily":{"status":"none","details":"列在路线图上——目前尚不可用"}},{"feature":"日历集成","openWhispr":{"status":"full","details":"Google Calendar 同步——会议自动关联到笔记"},"meetily":{"status":"none","details":"列在路线图上——目前尚不可用"}},{"feature":"AI 会议记录与摘要","openWhispr":{"status":"full","details":"自动生成笔记、待办事项和自定义模板"},"meetily":{"status":"full","details":"AI 摘要,内置六款模板"}},{"feature":"无机器人会议捕获","openWhispr":{"status":"full","details":"直接捕获系统音频——无机器人加入你的通话"},"meetily":{"status":"full","details":"直接捕获系统音频——无机器人加入你的通话"}},{"feature":"通用语音听写","openWhispr":{"status":"full","details":"在任意应用或文本框中一键说话听写"},"meetily":{"status":"none","details":"仅限会议——无通用听写"}},{"feature":"移动应用","openWhispr":{"status":"partial","details":"iOS 配套应用"},"meetily":{"status":"none","details":"无移动应用——标注为即将推出"}},{"feature":"平台支持","openWhispr":{"status":"full","details":"原生支持 macOS、Windows、Linux、iOS"},"meetily":{"status":"partial","details":"原生支持 macOS 和 Windows;Linux 需通过 Docker"}},{"feature":"离线可用","openWhispr":{"status":"full","details":"无需联网即可完整转录和记录"},"meetily":{"status":"full","details":"使用本地 Whisper,无需联网即可完整转录"}},{"feature":"免费版","openWhispr":{"status":"full","details":"无限本地转录,每月 5 小时会议录制"},"meetily":{"status":"full","details":"免费社区版,包含所有核心功能"}}],"whyOpenWhispr":{"title":"为什么选择 OpenWhispr?","diarization":{"title":"本地说话人分离今天就能用:","body":"谁说了什么的识别完全在你的设备上运行。Meetily 将高级说话人分离列在路线图上——OpenWhispr 现在就已交付。"},"calendar":{"title":"日历集成今天就能用:","body":"Google Calendar 同步会自动将会议记录关联到正确的日程。Meetily 的日历集成仍在路线图上。"},"chat":{"title":"与你的笔记对话今天就能用:","body":"跨每一场会议、笔记和转录内容提问。用智能体模式起草跟进内容。Meetily 的对话功能仍在路线图上。"},"dictation":{"title":"通用语音听写:","body":"一键说话即可输入任意应用——邮件、文档、代码、终端。Meetily 仅限会议。"},"linux":{"title":"原生 Linux 应用:","body":"为 Linux 而构建,而非通过 Docker 运行。在 macOS、Windows 和 Linux 上都是地道的一流桌面应用。"},"mobile":{"title":"移动配套应用:","body":"iOS 配套应用,随时随地进行语音捕获。"}},"whyMeetily":{"title":"为什么选择 Meetily?","mit":{"title":"MIT 许可证:","body":"宽松许可——可轻松 fork、换标或将 Meetily 嵌入你自己的工具中。"},"scope":{"title":"极简范围:","body":"如果你只需要录制和摘要,Meetily 聚焦的范围正好匹配——没有听写、没有日历、没有对话。"},"docker":{"title":"Docker 部署:","body":"如果符合你的工作流,可通过 Docker 在任意 Linux 发行版或服务器上自托管。"},"community":{"title":"社区势头:","body":"活跃的 GitHub 项目,开源社区对本地优先的会议工具兴趣浓厚。"}},"pricing":{"title":"真实成本对比","ow":{"freeLabel":"免费——永久","free1":"无限本地转录","free2":"本地说话人分离","free3":"Google Calendar 同步","free4":"每月 5 小时会议录制","free5":"语音听写至任意应用","free6":"100 多种语言","proLabel":"Pro","pro1":"每月 20 小时会议录制","pro2":"无限云端转录","pro3":"移动配套应用","businessLabel":"Business","business1":"无限会议录制","business2":"智能体模式","business3":"与你的数据对话"},"meetily":{"communityLabel":"社区版","community1":"实时转录","community2":"AI 摘要","community3":"6 款模板","community4":"本地 Whisper 模型","community5":"无说话人分离","community6":"无日历同步","community7":"无法与笔记对话","proLabel":"Pro(仅限年付——无月付选项)","pro1":"更高准确度","pro2":"高级导出","pro3":"自动检测功能","pro4":"无月付选项","enterprisePrice":"定制","enterpriseLabel":"Enterprise","enterprise1":"专属支持","enterprise2":"合规框架"},"callout":{"lead":"定价相近,功能深度不同:","body":"两款工具都提供带本地转录的免费版。OpenWhispr 的免费版包含说话人分离、日历同步和与笔记对话——而这些在 Meetily 上还是路线图项目。"}},"featureShowdown":{"title":"功能对决","openWhisprExclusivesTitle":"OpenWhispr 独有:","openWhisprExclusives":["本地说话人分离(设备端识别谁说了什么)","Google Calendar 同步","跨所有会议与你的笔记对话","智能体模式(草稿、跟进、网络 + 笔记 + 日历)","在任意应用中一键说话语音听写","除 Whisper 外还支持 NVIDIA Parakeet 模型","原生 Linux 桌面应用(非 Docker)","iOS 配套应用"],"meetilyExclusivesTitle":"Meetily 独有:","meetilyExclusives":["MIT 许可证(相比 OpenWhispr 许可证更宽松)","基于 Docker 的 Linux 部署","若你只需要录制 + 摘要,范围更窄"]},"evidence":{"title":"基于证据的说明","bodyBefore":"Meetily 功能可用性依据其公开路线图、GitHub 项目以及","bodyAfter":"截至 2026 年的情况。撰写本文时,日历集成、高级说话人分离、与会议记录对话以及移动应用均被 Meetily 标注为计划中/即将推出。请查看其官网了解最新动态。"},"migration":{"title":"从 Meetily 切换到 OpenWhispr","intro":"已经在用 Meetily,却想要那些仍在路线图上的功能?OpenWhispr 保留你当初选择 Meetily 时看重的本地优先理念:","step1":{"title":"下载 OpenWhispr","body":"提供 macOS、Windows 和 Linux 原生安装包。免费上手,本地使用无需注册账户。"},"step2":{"title":"选择本地模型","body":"与你在 Meetily 中熟悉的同一 Whisper 模型家族,另可选用 NVIDIA Parakeet。全部在设备端运行。"},"step3":{"title":"启用说话人分离","body":"开启本地说话人分离,让转录内容把每句话归属到正确的说话人——无需云端。"},"step4":{"title":"连接 Google Calendar","body":"可选——自动将笔记关联到正确的会议,并让智能体模式读取你即将到来的日程。"},"step5":{"title":"与你的会议对话","body":"跨你录制过的每一场会议提问。一条命令即可起草跟进邮件。"}},"bottomLine":{"title":"结论","intro":"Meetily 和 OpenWhispr 属于同一品类:开源、本地优先、将音频留在你设备上的 AI 会议助手。区别在于功能深度与范围。","teamMeetily":{"title":"Meetily 阵营","body":"你想要一款小巧、专注、采用 MIT 许可证的会议录制工具。你愿意等待说话人分离、对话和日历功能。"},"teamOpenWhispr":{"title":"OpenWhispr 阵营","body":"你想要同样的本地优先、开源保障,外加本地说话人分离、Google Calendar 同步、与你的笔记对话以及通用语音听写——今天就已交付。"},"closing":"两个项目,同样的理念。OpenWhispr 只是在今天交付了更完整的会议助手能力栈。"},"finalCta":{"title":"准备好体验完整的本地优先会议能力栈了吗?","body":"OpenWhispr 开源、本地优先、功能完备。会议记录、说话人分离、日历同步以及与你的数据对话——全部在你的机器上运行。","starButton":"在 GitHub 上给我们 Star","footnote":"无需信用卡 • 离线可用 • 永久开源","viewAllComparisons":"查看所有对比"}},"compareOtter":{"metadata":{"title":"Otter.ai 替代方案 — 私密、本地、开源","description":"免费开源的 Otter.ai 替代方案。OpenWhispr 完全在设备端运行听写，离线且私密，无需云端，也没有使用上限。免费体验。","keywords":["Otter.ai 替代方案","Otter 免费替代方案","隐私听写应用","Otter.ai 对比 OpenWhispr","本地语音转文字","隐私转录替代方案","开源听写","离线听写应用","nvidia parakeet"]},"breadcrumb":{"home":"首页","comparisons":"对比","otter":"对比 Otter.ai"},"hero":{"title":"私密、开源 Otter.ai 替代方案","subtitle":"OpenWhispr 是 Otter.ai 的替代品，适合那些想要 100% 在自己的设备上运行语音转文本的人。您的音频永远不会离开您的机器，它可以离线工作，并且整个应用程序是开源的。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"快速了解","otterIntro":"是一个强大的基于云端的会议转录平台。它能加入你的 Zoom、Teams 和 Google Meet 通话,实时转录对话、生成摘要、提取待办事项,并与 Salesforce、HubSpot 等工具集成。它在自己擅长的领域表现出色——但本质上是一款会议工具,而非听写工具。","openWhisprIntro":"采取完全不同的方法。它是一款隐私第一的听写工具，完全在使用 OpenAI Whisper 和 NVIDIA Parakeet 的计算机上运行。它还包括一个设备上的 AI Notepad，用于捕获会议记录，而无需机器人加入通话。","chooseOtter":"选择 Otter.ai,如果:你需要自动化会议转录,搭配团队协作和 CRM 集成。","chooseOpenWhispr":"选择 OpenWhispr,如果:你想要隐私、本地的语音听写,在任意应用中可用且无需将音频发送到云端。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"隐私与数据控制","openWhispr":{"status":"full","details":"100% 本地处理,你的数据永不离开机器"},"otter":{"status":"none","details":"依据 Otter 政策进行云端处理和存储"}},{"feature":"开源","openWhispr":{"status":"full","details":"在 GitHub 上完全开源"},"otter":{"status":"none","details":"专有、闭源"}},{"feature":"平台支持","openWhispr":{"status":"full","details":"macOS、Windows、Linux、iOS"},"otter":{"status":"full","details":"网页版、macOS、Windows、iOS、Android"}},{"feature":"离线可用","openWhispr":{"status":"full","details":"无需联网即可完整使用"},"otter":{"status":"none","details":"所有功能均需联网"}},{"feature":"通用听写","openWhispr":{"status":"full","details":"专为在任意文本框中听写而打造"},"otter":{"status":"partial","details":"主要聚焦会议,听写用途有限"}},{"feature":"会议转录","openWhispr":{"status":"partial","details":"含会议录制(免费 5 小时,Pro 20 小时)"},"otter":{"status":"full","details":"核心功能,集成 Zoom/Teams/Meet"}},{"feature":"AI 整理与重新排版","openWhispr":{"status":"full","details":"可为任意排版任务自定义系统提示词"},"otter":{"status":"partial","details":"自动生成摘要和待办事项"}},{"feature":"免费版限制","openWhispr":{"status":"full","details":"无限制,包含所有核心功能"},"otter":{"status":"partial","details":"每月 300 分钟,终身仅 3 次文件导入"}}],"whyOpenWhispr":{"title":"为什么选择 OpenWhispr?","privacy":{"title":"真正的隐私——本地处理:","body":"您的音频永远不会离开您的设备。没有云上传，没有第三方服务器，没有任何内容存储在设备外。 Otter.ai 在云端处理所有内容，并将录音存储在 Otter 的云端。"},"dictation":{"title":"专为听写打造:","body":"OpenWhispr 专为用语音输入文字而设计——起草邮件、撰写代码注释、填写表单。Otter 则是为转录会议而设计。"},"noLimits":{"title":"无使用限制:","body":"想听写多少就听写多少,永久如此。Otter 的免费版将你限制在每月 300 分钟,即便是 Pro 版也只有 1,200 分钟。"},"openSource":{"title":"开源且可自由改造:","body":"审阅每一行代码。fork 它、扩展它、审计它。确切了解你的音频数据发生了什么。"}},"whyOtter":{"title":"为什么选择 Otter.ai?","meetings":{"title":"会议转录:","body":"Otter 会自动加入 Zoom、Teams 和 Meet 通话,实时转录并识别说话人。这是它的核心优势。"},"collaboration":{"title":"团队协作:","body":"共享转录内容、标记关键时刻、将待办事项分配给队友。专为需要共享会议记录的团队而打造。"},"summaries":{"title":"AI 摘要与待办事项:","body":"自动生成会议摘要、提取决策,并创建可供分配和跟踪的待办事项。"},"integrations":{"title":"深度集成:","body":"可连接 50 多款工具,包括 Salesforce、HubSpot、Slack、Notion、Jira 和 Asana。对于销售和项目管理工作流极为强大。"}},"pricing":{"title":"真实成本对比","ow":{"freeLabel":"永久免费","free1":"无限听写","free2":"包含 Whisper 和 Parakeet 模型","free3":"100% 本地处理","free4":"通过系统提示词进行 AI 整理","free5":"macOS、Windows、Linux","proLabel":"Pro(可选)","pro1":"包含免费版全部功能","pro2":"无限云端转录","pro3":"每月 20 小时会议录制","pro4":"移动应用","pro5":"邮件支持"},"otter":{"basicLabel":"Basic","basic1":"每月 300 分钟","basic2":"3 次文件导入(终身)","basic3":"会议集成","basic4":"有限存储","proLabel":"Pro(年付每月 $8.33)","pro1":"每月 1,200 分钟","pro2":"每月 10 次文件导入","pro3":"高级搜索与导出","pro4":"单场会议最长 90 分钟","businessLabel":"Business(年付每月 $13.33)","business1":"无限录制","business2":"单场会议 4 小时上限","business3":"使用分析","business4":"优先支持"},"callout":{"lead":"不同的工具,不同的成本:","body":"Otter.ai 的定价面向需要会议转录的团队。OpenWhispr 免费为你提供无限的隐私听写——即便是 Pro 版,价格也不到 Otter 最便宜付费方案的一半。"}},"featureShowdown":{"title":"功能对决","bothHaveTitle":"两者都有的功能:","bothHave":["语音转文字转录","多语言支持","标点与排版","AI 驱动的文本处理","提供免费版"],"openWhisprExclusivesTitle":"OpenWhispr 独有:","openWhisprExclusives":["100% 本地/离线处理","开源(可审阅并 fork 代码)","通过剪贴板粘贴在任意文本框中使用","自定义 AI 系统提示词(智能体模式)","无限免费使用,无分钟上限","Linux 支持","为 OpenAI 和 Groq API BYOK"],"otterExclusivesTitle":"Otter.ai 独有:","otterExclusives":["自动化会议机器人(替你加入通话)","转录内容的实时团队协作","CRM 集成(Salesforce、HubSpot)","自动生成会议摘要","待办事项提取与分配","iOS 和 Android 移动应用","企业级功能(SSO、HIPAA、管理控制)"]},"evidence":{"title":"基于证据的说明","body":"本对比基于厂商文档、定价页面和公开产品文档,而非道听途说的用户评价。"},"migration":{"title":"将听写从 Otter.ai 切换到 OpenWhispr","intro":"正用 Otter.ai 做听写,却想要一款专门打造的工具?以下是切换方法:","step1":{"title":"下载 OpenWhispr","body":"免费下载。可选 Pro($8/用户/月),享无限云端转录、更多会议时长和移动应用。支持 macOS、Windows、Linux 和 iOS。"},"step2":{"title":"选择你的 AI 模型","body":"根据你的硬件和需求,从 OpenAI Whisper 或 NVIDIA Parakeet 模型中选择。可从 \"base\" 入手,在各方面取得不错的平衡。"},"step3":{"title":"设置你的热键","body":"配置一键说话(按住录音或切换模式)。然后只需说话——你的文字会出现在光标所在之处。"},"step4":{"title":"自定义 AI 整理","body":"设置自定义系统提示词,自动为你的听写排版——修正语法、排版为邮件、翻译,或任何你需要的处理。"},"callout":{"lead":"会议仍用 Otter:","body":"OpenWhispr 和 Otter.ai 解决的是不同的问题。许多用户保留 Otter 用于会议转录,同时使用 OpenWhispr 进行日常听写——两全其美。"}},"bottomLine":{"title":"结论","intro":"Otter.ai 和 OpenWhispr 为不同的任务而生。在每一项指标上正面对比它们会偏离重点——正确的选择完全取决于你的需求。","teamOtter":{"title":"Otter.ai 阵营","body":"你整天都在开会,需要在团队间共享转录内容,并希望将 AI 生成的摘要导入 Salesforce 或 Slack。Otter 正是为此而打造,且做得极为出色。"},"teamOpenWhispr":{"title":"OpenWhispr 阵营","body":"你想要私密地听写文字,无需云端处理、使用限制或订阅费用。你看重开源、跨多个平台工作,并想要一款把一件事做到完美的工具:将你的语音转为文字。"},"closing":"最好的工具,是为你真实工作流而打造的那一款。对于隐私听写,OpenWhispr 是不二之选。对于会议转录,Otter.ai 当之无愧。"},"finalCta":{"title":"准备好体验隐私、开源的听写了吗?","body":"别再把你的语音发送到云端。OpenWhispr 为你提供快速、准确的听写,完全在你的机器上运行——免费且开源。","starButton":"在 GitHub 上给我们 Star","footnote":"无需信用卡 • 离线可用 • 永久开源","viewAllComparisons":"查看所有对比"}},"compareSpokenly":{"metadata":{"title":"OpenWhispr 与 Spokenly 对比 — 详细比较 | OpenWhispr","description":"将 OpenWhispr 与 Spokenly 逐项对比。开源跨平台听写应用，对比一款精致的 macOS 与 iOS 语音转文字应用（配备 Agent Mode）。","keywords":["Spokenly 替代品","Spokenly 与 OpenWhispr","Mac 听写应用","Whisper 听写","Mac 语音转文字","离线听写"]},"breadcrumb":{"home":"首页","comparisons":"对比","spokenly":"对比 Spokenly"},"hero":{"title":"OpenWhispr 与 Spokenly 对比","subtitle":"Spokenly 是一款广受欢迎、配备 Agent Mode 的 macOS 与 iOS 听写应用。下面看看它与开源跨平台替代方案的对比。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"快速概览","spokenlyIntro":"是一款面向 macOS 与 iOS 的精致语音转文字应用，拥有超过 10 万名活跃用户。其免费的本地模型（Whisper 和 Parakeet）没有使用限制，Agent Mode 还能让你用语音指令操控 Mac。Pro 版每月 9.99 美元，提供托管的云端模型。","openWhisprIntro":"提供的是另一种价值主张。它完全开源，可在 macOS、Windows、Linux 和 iOS 上运行，让你完全掌控自己的数据与工作流程。本地处理永久免费，云端功能与移动端访问则可选每月 8 美元的 Pro 版。","chooseSpokenly":"选择 Spokenly，如果：你使用 Mac/iOS，想要 Agent Mode、智能提示词以及精致的原生体验。","chooseOpenWhispr":"选择 OpenWhispr，如果：你需要 Windows 或 Linux 支持、看重开源，或想要自定义你的工具。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"隐私与数据掌控","openWhispr":{"status":"full","details":"100% 本地选项，可自带 API 密钥"},"spokenly":{"status":"full","details":"提供本地模型，云端音频处理后即删除"}},{"feature":"开源","openWhispr":{"status":"full","details":"在 GitHub 上完全开源"},"spokenly":{"status":"none","details":"专有软件，闭源"}},{"feature":"平台支持","openWhispr":{"status":"full","details":"macOS、Windows、Linux、iOS"},"spokenly":{"status":"partial","details":"仅限 macOS 和 iOS"}},{"feature":"定价模式","openWhispr":{"status":"full","details":"永久免费，可选每月 8 美元的 Pro 版"},"spokenly":{"status":"full","details":"本地模型免费且不限量，云端 Pro 版每月 9.99 美元"}},{"feature":"AI 模型选择","openWhispr":{"status":"full","details":"OpenAI Whisper 与 NVIDIA Parakeet 模型"},"spokenly":{"status":"full","details":"Whisper、Parakeet，以及通过 BYOK 或 Pro 提供的云端模型"}},{"feature":"语音指令","openWhispr":{"status":"partial","details":"基本的标点指令"},"spokenly":{"status":"full","details":"Agent Mode 支持 Mac 自动化、启动应用、网页搜索"}},{"feature":"离线支持","openWhispr":{"status":"full","details":"完整的离线功能"},"spokenly":{"status":"full","details":"纯本地模式屏蔽所有网络请求"}},{"feature":"自定义","openWhispr":{"status":"full","details":"完全可自定义，便于 fork"},"spokenly":{"status":"partial","details":"智能提示词与设置，但不可扩展"}}],"whyOpenWhispr":{"title":"为什么选 OpenWhispr？","crossPlatform":{"title":"跨平台：","body":"可在 macOS、Windows、Linux 和 iOS 上运行。Spokenly 仅限 macOS 和 iOS。"},"openSource":{"title":"开源自由：","body":"查看代码、修改功能、贡献改进。没有黑盒。"},"pricing":{"title":"更低的 Pro 费用：","body":"Pro 版每月 8 美元，而 Spokenly Pro 版每月 9.99 美元，并且包含会议录音和移动端。"},"hackable":{"title":"可深度定制：","body":"添加功能、创建插件，随心所欲地融入你的工作流程。"}},"whySpokenly":{"title":"为什么选 Spokenly？","agentMode":{"title":"Agent Mode：","body":"用语音指令操控 Mac——搜索网页、启动应用、免动手运行快捷指令。"},"nativeApple":{"title":"原生 Apple 体验：","body":"专为 macOS 和 iOS 打造，针对 Apple Silicon 优化，App Store 评分高达 4.9 星。"},"cloudModels":{"title":"广泛的云端模型支持：","body":"自带 OpenAI、Claude、Gemini、Deepgram 和 Groq 的密钥——或交由 Pro 版处理。"},"mcp":{"title":"MCP 集成：","body":"内置面向 AI 编程助手的 MCP 服务器，让你直接将语音听写到开发工作流程中。"}},"pricing":{"title":"真实成本对比","ow":{"freeLabel":"永久免费","free1":"包含全部本地功能","free2":"本地 Whisper 与 Parakeet 模型","free3":"无使用限制","free4":"macOS、Windows 和 Linux","proLabel":"Pro","pro1":"无限量云端转录","pro2":"每月 20 小时会议录音","pro3":"移动应用","pro4":"邮件支持"},"spokenly":{"freeLabel":"本地模型","free1":"Whisper 与 Parakeet，不限量","free2":"无需注册","free3":"仅限 macOS 和 iOS","proLabel":"Pro","pro1":"包含托管云端模型","pro2":"优先支持","pro3":"覆盖 macOS 和 iPhone"},"callout":{"lead":"两者都提供免费的本地模型：","body":"Spokenly 和 OpenWhispr 都能让你免费在本地运行 Whisper 和 Parakeet。关键区别在于平台支持和是否开源。"}},"featureShowdown":{"title":"功能大比拼","bothHaveTitle":"两者都具备的功能：","bothHave":["免费的本地 Whisper 与 Parakeet 模型","完整的离线功能","快捷键激活","在任意应用中可用","支持 100 多种语言","实时转录"],"openWhisprExclusivesTitle":"OpenWhispr 独有：","openWhisprExclusives":["支持 Windows 和 Linux","完全开源，可 fork","会议录音与转录","社区驱动的开发"],"spokenlyExclusivesTitle":"Spokenly 独有：","spokenlyExclusives":["面向 macOS 自动化的 Agent Mode","支持多个云端服务商的 BYOK","面向 AI 编程助手的内置 MCP 服务器","用于语法和格式的智能提示词"]},"evidence":{"title":"基于事实的说明","body":"本对比基于 Spokenly 的官方网站、App Store 页面及公开文档。Spokenly 报告称拥有超过 10 万名活跃用户，App Store 评分为 4.9/5。定价与功能截至 2026 年 3 月准确。"},"migration":{"title":"从 Spokenly 切换到 OpenWhispr","intro":"已经在 Mac 上使用 Spokenly？下面教你如何把 OpenWhispr 加入工作流程：","step1":{"title":"下载 OpenWhispr","body":"从 GitHub 获取免费版，或升级到 Pro 版使用云端功能。"},"step2":{"title":"匹配你的设置","body":"选择与你在 Spokenly 中相同的 Whisper 或 Parakeet 模型，并设置你偏好的快捷键。"},"step3":{"title":"扩展到其他平台","body":"把你的语音听写工作流程也带到 Windows 或 Linux 设备上。"},"callout":{"lead":"少了 Agent Mode？","body":"OpenWhispr 专注于听写准确度与跨平台覆盖。如果你高度依赖语音驱动的 Mac 自动化，可以让两款工具并行使用。"}},"bottomLine":{"title":"结论","intro":"Spokenly 和 OpenWhispr 有很多共同点——两者都提供免费的本地 Whisper 和 Parakeet 模型、出色的隐私保护以及全系统听写。选择取决于平台需求和理念：","teamSpokenly":{"title":"Spokenly 阵营","body":"你全身心投入 Apple 生态，想要用于 Mac 自动化的 Agent Mode，并欣赏通过 BYOK 支持广泛云端模型的精致原生体验。"},"teamOpenWhispr":{"title":"OpenWhispr 阵营","body":"你需要 Windows 或 Linux，信奉开源，想要会议录音，或偏好完全自定义工具的自由。"},"closing":"两者都是出色的听写工具。挑选最契合你的平台和工作流程的那一款。"},"finalCta":{"title":"准备好试试开源听写了吗？","body":"获得与 Spokenly 同样出色的本地模型，外加跨平台支持和开源带来的自由。","starButton":"在 GitHub 上给我们点 Star","footnote":"无需信用卡 • 离线可用 • 永久开源","viewAllComparisons":"查看全部对比"}},"compareSuperwhisper":{"metadata":{"title":"SuperWhisper 替代方案：免费开源","description":"SuperWhisper 的免费开源替代方案，可在 Mac、Windows、Linux 和 iOS 上本地运行。无需订阅，直接使用强大的设备端听写。","keywords":["SuperWhisper 替代品","SuperWhisper 与 OpenWhispr","Mac 听写应用","Whisper 听写","NVIDIA Parakeet"]},"breadcrumb":{"home":"首页","comparisons":"对比","superwhisper":"对比 SuperWhisper"},"hero":{"title":"OpenWhispr 与 SuperWhisper 对比","subtitle":"SuperWhisper 现已支持 macOS、Windows 和 iOS。下面是开源方案与精致专有体验的客观对比。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"快速概览","superWhisperIntro":"是一款精致的听写产品，最初以 Mac 为先，如今已登陆 macOS、Windows 和 iOS。它提供免费的每月额度，以及面向重度使用和更多云端功能的付费套餐。","openWhisprIntro":"提供的是另一种价值主张。它完全免费、开源，并且适用于所有平台——而不只是 Mac。你将获得强大的 OpenAI Whisper 和 NVIDIA Parakeet 模型，并拥有修改、扩展和真正掌控自己工具的自由。此外，它背后还有一个看重隐私与透明的社区。","chooseSuperWhisper":"选择 SuperWhisper，如果：你想要一款精致的专有应用，并内置跨设备同步。","chooseOpenWhispr":"选择 OpenWhispr，如果：你看重开源、需要跨平台支持，或想要自定义。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"隐私与数据掌控","openWhispr":{"status":"full","details":"100% 本地选项，可自带 API 密钥"},"superWhisper":{"status":"full","details":"本地处理，可选云端功能"}},{"feature":"开源","openWhispr":{"status":"full","details":"在 GitHub 上完全开源"},"superWhisper":{"status":"none","details":"专有软件，闭源"}},{"feature":"平台支持","openWhispr":{"status":"full","details":"macOS、Windows、Linux、iOS"},"superWhisper":{"status":"partial","details":"macOS、Windows、iOS（无 Linux）"}},{"feature":"定价模式","openWhispr":{"status":"full","details":"永久免费，可选每月 8 美元的 Pro 版"},"superWhisper":{"status":"partial","details":"免费套餐外加付费额度/订阅（详见官方定价）"}},{"feature":"AI 模型选择","openWhispr":{"status":"full","details":"OpenAI Whisper 与 NVIDIA Parakeet 模型"},"superWhisper":{"status":"full","details":"提供多个 Whisper 模型"}},{"feature":"系统集成","openWhispr":{"status":"full","details":"在任意应用中可用"},"superWhisper":{"status":"full","details":"深度的 macOS 集成"}},{"feature":"自定义","openWhispr":{"status":"full","details":"完全可自定义，便于 fork"},"superWhisper":{"status":"partial","details":"仅限设置与偏好"}},{"feature":"语音指令","openWhispr":{"status":"partial","details":"基本的标点指令"},"superWhisper":{"status":"full","details":"高级指令与快捷方式"}}],"whyOpenWhispr":{"title":"为什么选 OpenWhispr？","openSource":{"title":"开源自由：","body":"查看代码、修改功能、贡献改进。没有黑盒。"},"crossPlatform":{"title":"跨平台：","body":"可在 Mac、Windows、Linux 和 iOS 上运行。把你的工作流程带到任何地方。"},"foreverFree":{"title":"永久免费：","body":"无需购买。Pro 版为可选项，仅需每月 8 美元。"},"hackable":{"title":"可深度定制：","body":"添加功能、创建插件，随心所欲地融入你的工作流程。"}},"whySuperWhisper":{"title":"为什么选 SuperWhisper？","nativeFeel":{"title":"原生 macOS 质感：","body":"专为 Mac 设计，界面精美、动画流畅。"},"polish":{"title":"精致与设计：","body":"每个细节都为最佳用户体验精心打磨，用起来确实令人愉悦。"},"advancedFeatures":{"title":"高级功能：","body":"开箱即用的语音指令、自定义模式以及针对特定应用的设置。"},"support":{"title":"专业支持：","body":"专属的支持团队，以及来自专注开发团队的定期更新。"}},"pricing":{"title":"真实成本对比","ow":{"freeLabel":"永久免费","free1":"包含全部功能","free2":"本地处理","free3":"无任何限制","free4":"社区支持","proLabel":"Pro","pro1":"无限量云端转录","pro2":"每月 20 小时会议录音","pro3":"移动应用","pro4":"邮件支持"},"super":{"freeAmount":"免费","freeLabel":"免费每月额度","free1":"试用全部 Pro 功能","free2":"小型 AI 模型","free3":"基本功能","paidAmount":"付费套餐","paidLabel":"付费套餐","paid1":"无限量使用","paid2":"大型 AI 模型","paid3":"全部功能","paid4":"macOS + Windows + iOS"},"callout":{"lead":"价格并非一切：","body":"虽然 SuperWhisper 凭借其分级定价提供了很高的性价比，但 OpenWhispr 给你的是金钱买不到的东西——完全的掌控与透明。"}},"featureShowdown":{"title":"功能大比拼","bothHaveTitle":"两者都具备的功能：","bothHave":["多种 AI 模型选择（Whisper 和 Parakeet）","本地处理以保护隐私","快捷键激活","在任意应用中可用","高精度转录","多语言支持"],"openWhisprExclusivesTitle":"OpenWhispr 独有：","openWhisprExclusives":["跨平台（Mac、Windows、Linux、iOS）","开源且可 fork","永久免费选项","社区驱动的开发"],"superWhisperExclusivesTitle":"SuperWhisper 独有：","superWhisperExclusives":["原生 macOS 界面/体验","高级语音指令","针对特定应用的模式","菜单栏集成"]},"evidence":{"title":"基于事实的说明","body":"本对比基于供应商文档、定价页面和公开产品文档，而非道听途说的用户评价。"},"migration":{"title":"从 SuperWhisper 切换到 OpenWhispr","intro":"喜欢 SuperWhisper，但对开源感到好奇？下面教你如何完成切换：","step1":{"title":"下载 OpenWhispr","body":"从 GitHub 获取免费版，或升级到 Pro 版以获得便利。"},"step2":{"title":"匹配你的设置","body":"设置你偏好的 Whisper 模型和快捷键，以匹配你的 SuperWhisper 配置。"},"step3":{"title":"享受自由","body":"继续在 Mac 上使用，或把你的工作流程也带到 Windows 或 Linux 设备上。"},"callout":{"lead":"少了语音指令？","body":"OpenWhispr 社区正在积极开发高级功能。来 GitHub 加入我们，一起塑造未来！"}},"bottomLine":{"title":"结论","intro":"OpenWhispr 和 SuperWhisper 都是出色的工具，都尊重你的隐私并提供强大的 AI 转录。选择取决于你的价值观和需求：","teamSuperWhisper":{"title":"SuperWhisper 阵营","body":"你是一名 Mac 用户，看重精致质感、原生集成以及开箱即用的完美体验。每天都会用到的工具花 29 美元很划算。"},"teamOpenWhispr":{"title":"OpenWhispr 阵营","body":"你信奉开源，想要跨平台支持，或喜欢自定义工具的理念。永久免费，并可选择支持开发。"},"closing":"最好的工具就是契合你工作流程和理念的那一款。两者各有千秋，都是赢家。"},"finalCta":{"title":"准备好加入开源运动了吗？","body":"体验社区驱动开发的力量。获得与 SuperWhisper 同样出色的功能，外加将其打造成自己专属工具的自由。","starButton":"在 GitHub 上给我们点 Star","footnote":"无需信用卡 • 离线可用 • 永久开源","viewAllComparisons":"查看全部对比"}},"compareTalon":{"metadata":{"title":"OpenWhispr vs Talon Voice：简单的 AI 听写","description":"OpenWhispr 是用于听写的简单 Talon Voice 替代方案，提供 AI 文本整理和即时设置，完全免费开源，无需编写脚本。","keywords":["Talon Voice 替代方案","Talon 对比 OpenWhispr 听写","简单语音听写","无需 Talon 的语音输入","用于听写的 Talon Voice 替代方案","轻松语音转文字","无需脚本的语音听写","OpenWhispr 对比 Talon","nvidia parakeet"]},"breadcrumb":{"home":"首页","comparisons":"对比","talon":"对比 Talon Voice"},"hero":{"title":"OpenWhispr 对比 Talon Voice","subtitle":"两款工具，两种需求。简单的 AI 听写，对比全程免手操控电脑。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"快速了解","talonIntro":"是一套强大的免手输入系统，专为需要或希望完全用语音操控整台电脑的人打造。它支持眼动追踪、噪声识别，以及一套 Python 脚本层，让你能自定义语音交互的每一个细节。它在免手编程方面表现出色，对无障碍场景更是不可或缺。但它需要大量前期设置——安装社区脚本文件、学习一套语音字母表，以及编写自定义命令。","openWhisprIntro":"只专注于一件事：尽可能准确、轻松地把语音变成文字。下载安装，按下热键，自然开口说话，文字便会出现——并由 AI 整理和排版。无需脚本、无需语音字母表、没有学习曲线。它为任何想要快速、私密、AI 驱动听写的人而生。","chooseTalon":"选择 Talon，如果：你需要全程免手操控电脑、语音编程，或有超出听写范畴的无障碍需求。","chooseOpenWhispr":"选择 OpenWhispr，如果：你想要简单、准确、带 AI 整理且零设置的语音听写。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"核心用途","openWhispr":{"status":"full","details":"语音听写与文字输入"},"talon":{"status":"full","details":"全程免手操控电脑"}},{"feature":"设置复杂度","openWhispr":{"status":"full","details":"下载、安装、开始听写"},"talon":{"status":"partial","details":"需要脚本和社区文件"}},{"feature":"学习曲线","openWhispr":{"status":"full","details":"极低——自然开口即可"},"talon":{"status":"partial","details":"较陡——语音字母表、命令、脚本"}},{"feature":"AI 文本整理","openWhispr":{"status":"full","details":"通过系统提示词进行 AI 排版"},"talon":{"status":"none","details":"仅原始转录"}},{"feature":"开源","openWhispr":{"status":"full","details":"在 GitHub 上完全开源"},"talon":{"status":"partial","details":"社区脚本开源，核心为专有"}},{"feature":"平台支持","openWhispr":{"status":"full","details":"macOS、Windows、Linux、iOS"},"talon":{"status":"full","details":"macOS、Windows、Linux（X11）"}},{"feature":"通用听写","openWhispr":{"status":"full","details":"针对自然语音听写优化"},"talon":{"status":"partial","details":"可以实现，但并非主要侧重点"}},{"feature":"免手编程","openWhispr":{"status":"partial","details":"可在任意文本框中听写"},"talon":{"status":"full","details":"为语音编程量身打造"}},{"feature":"眼动追踪","openWhispr":{"status":"none","details":"不支持"},"talon":{"status":"full","details":"集成 Tobii 眼动仪"}},{"feature":"价格","openWhispr":{"status":"full","details":"永久免费，可选 $8/月 Pro"},"talon":{"status":"full","details":"核心应用免费，可选择付费 Patreon 档位以获取尝鲜功能和支持"}}],"whyOpenWhispr":{"title":"为什么选 OpenWhispr？","instantSetup":{"title":"即装即用：","body":"下载、安装、按下热键，即可开始听写。无需脚本、无需配置文件，也不必背诵语音字母表。"},"aiCleanup":{"title":"AI 文本整理：","body":"系统提示词会自动为你的语音排版、加标点并加以润色。Talon 给你的只是原始转录。"},"naturalSpeech":{"title":"自然语音：","body":"自然开口说话，即可得到精炼的文字。无需学习诸如 “word cap hello” 或 “phrase hello world” 之类的特殊命令。"},"freeForever":{"title":"永久免费：","body":"全功能免费使用。Pro 为可选，$8/月。Talon 的 Beta 功能需在 Patreon 上支付 $25/月。"}},"whyTalon":{"title":"为什么选 Talon Voice？","fullControl":{"title":"全面操控电脑：","body":"用语音导航整个操作系统——窗口管理、应用切换、文件浏览等等。OpenWhispr 只专注于文字输入。"},"eyeTracking":{"title":"眼动追踪：","body":"与 Tobii 眼动仪集成以控制鼠标。对于完全无法使用鼠标的用户至关重要。"},"voiceCoding":{"title":"语音编程：","body":"为编写代码量身打造的命令——语言专属代码片段、IDE 集成，以及通过语音控制终端。"},"accessibility":{"title":"无障碍：","body":"对于患有 RSI、运动障碍或其他限制手部使用状况的人来说，它真正改变了人生。Talon 满足了一项至关重要的需求。"}},"pricing":{"title":"真实成本对比","ow":{"freeLabel":"永久免费","free1":"包含全部功能","free2":"本地 Whisper 与 Parakeet 处理","free3":"AI 文本整理","free4":"99+ 种语言","proLabel":"Pro","pro1":"无限云端转录","pro2":"每月 20 小时会议录音","pro3":"移动应用","pro4":"邮件支持"},"talon":{"freeAmount":"免费","freeLabel":"公开版","free1":"核心语音控制","free2":"Conformer 语音引擎","free3":"社区脚本","betaLabel":"Beta（Patreon）","beta1":"抢先体验新功能","beta2":"优先支持","beta3":"Beta 通道访问","beta4":"VIP Slack 频道"},"callout":{"lead":"不同的工具，不同的成本：","body":"Talon 的 Patreon 支持着一位全职开发无障碍软件的开发者。OpenWhispr 的免费档位让你零成本获得全功能听写。两种模式都很公道。"}},"featureShowdown":{"title":"功能大比拼","bothHaveTitle":"两者都有的功能：","bothHave":["跨平台（macOS、Windows、Linux）","本地语音处理","隐私优先设计","可通过开放社区自定义","离线可用","活跃的社区支持"],"openWhisprExclusivesTitle":"OpenWhispr 独有：","openWhisprExclusives":["AI 文本整理与排版","即装即用——无需脚本","自然语音听写（正常说话即可）","支持 BYOK 自带云端 API 密钥","完全开源（含核心）"],"talonExclusivesTitle":"Talon Voice 独有：","talonExclusives":["眼动追踪鼠标控制","用语音全面操控操作系统","带语言专属命令的语音编程","噪声识别（爆破音/嘶音点击）","用于自定义命令的 Python 脚本"]},"evidence":{"title":"基于实证的说明","body":"本对比基于厂商文档、定价页面和公开的产品文档，而非道听途说的用户评价。"},"migration":{"title":"从 Talon 而来？这是你能期待的","intro":"如果你用 Talon 来听写、想要更简单的方案，或想在 Talon 之外加入一款专门的听写工具：","step1":{"title":"下载 OpenWhispr","body":"免费下载。可选 Pro（$8/用户/月），畅享无限云端转录、更多会议时长以及移动应用。"},"step2":{"title":"设置你的热键","body":"选一个不与 Talon 命令冲突的按住说话键。两者可以并行运行。"},"step3":{"title":"自然听写","body":"无需语音字母表，无需 “phrase” 前缀。开口说话即可，让 AI 处理排版和标点。"},"callout":{"lead":"它们配合得很好：","body":"用 Talon 来免手编程和操控操作系统。当你想用 AI 整理来听写长文、邮件或笔记时，用 OpenWhispr。"}},"bottomLine":{"title":"结论","intro":"Talon Voice 和 OpenWhispr 解决的是根本不同的问题。Talon 是一套全面的免手输入系统——它能彻底取代你的键盘和鼠标。OpenWhispr 是一款专注的听写工具——它把你的语音变成精炼的文字。直接比较两者，就像拿瑞士军刀和厨刀作比较：都能切，但各为不同的活计而设计。","teamTalon":{"title":"Talon 阵营","body":"你需要免手操控电脑、语音编程或眼动追踪。你愿意投入时间学习这套系统，因为回报——彻底的语音驱动计算——物有所值。Talon 在无障碍方面真正改变人生。"},"teamOpenWhispr":{"title":"OpenWhispr 阵营","body":"你想要准确、AI 增强且即装即用的听写。你说话，它书写——干净、排好版、即取即用。无需脚本，没有学习曲线。免费、开源、私密。"},"closing":"许多用户两者并用——Talon 用于免手编程和操控电脑，OpenWhispr 用于快速自然听写。它们彼此互补，相得益彰。"},"finalCta":{"title":"准备好轻松听写了吗？","body":"跳过繁琐设置，几分钟内开始听写。AI 驱动、私密、开源、免费。","starButton":"在 GitHub 上给我们点 Star","footnote":"无需信用卡 • 离线可用 • 永久开源"},"viewAllComparisons":"查看所有对比"},"compareTypeless":{"metadata":{"title":"Typeless 替代方案 — OpenWhispr vs Typeless","description":"真正支持离线处理的 Typeless 替代方案。OpenWhispr 开源并在本地运行，没有字数限制，还支持 Linux。免费体验。","keywords":["typeless 替代方案","typeless 对比 openwhispr","ai 语音听写","语音输入应用","听写软件对比","智能语音听写"]},"breadcrumb":{"home":"首页","comparisons":"对比","typeless":"对比 Typeless"},"hero":{"title":"OpenWhispr 对比 Typeless","subtitle":"Typeless 是一款跨桌面与移动端、支持智能编辑的 AI 语音键盘。看看它与开源、隐私优先的替代方案相比如何。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"快速了解","typelessIntro":"是一款 AI 驱动的语音键盘，能把自然语音转化为精炼文字。它会自动去除口头语、处理自我纠错，并根据你正在使用的应用调整语气。它支持 macOS、Windows、iOS 和 Android，免费档位为每周 4,000 词。","openWhisprIntro":"采取了不同的路线。它完全开源，借助本地 Whisper 和 Parakeet 模型完全离线运行，让你对自己的数据拥有完全掌控。除 macOS 和 Windows 外，它还支持 Linux，且本地处理没有每周词数限制。","chooseTypeless":"选择 Typeless，如果：你想要 AI 驱动的智能编辑、移动端语音输入和自动去除口头语。","chooseOpenWhispr":"选择 OpenWhispr，如果：你需要离线支持、Linux 兼容、开源或无限本地听写。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"隐私与数据掌控","openWhispr":{"status":"full","details":"可 100% 本地运行，支持自带 API 密钥"},"typeless":{"status":"partial","details":"云端零数据留存，但处理并非完全本地"}},{"feature":"开源","openWhispr":{"status":"full","details":"在 GitHub 上完全开源"},"typeless":{"status":"none","details":"专有、闭源"}},{"feature":"平台支持","openWhispr":{"status":"full","details":"macOS、Windows、Linux、iOS"},"typeless":{"status":"full","details":"macOS、Windows、iOS、Android（无 Linux）"}},{"feature":"定价模式","openWhispr":{"status":"full","details":"永久免费，可选 $8/月 Pro"},"typeless":{"status":"partial","details":"免费每周 4,000 词，Pro $12/月（年付）或 $30/月"}},{"feature":"智能编辑","openWhispr":{"status":"partial","details":"通过云端模型进行 AI 后处理"},"typeless":{"status":"full","details":"自动去除口头语、自我纠错、语气适配"}},{"feature":"离线支持","openWhispr":{"status":"full","details":"完整的离线功能"},"typeless":{"status":"none","details":"转录需要联网"}},{"feature":"自定义","openWhispr":{"status":"full","details":"完全可自定义，方便 fork"},"typeless":{"status":"partial","details":"个人词典、按应用设置语气偏好"}},{"feature":"移动端支持","openWhispr":{"status":"partial","details":"通过 Pro 计划支持 iOS 和 Android"},"typeless":{"status":"full","details":"原生 iOS 和 Android 应用，免费包含"}}],"whyOpenWhispr":{"title":"为什么选 OpenWhispr？","privacy":{"title":"真正的离线隐私：","body":"一切本地处理。你的语音永不离开设备。Typeless 需要联网。"},"openSource":{"title":"开源自由：","body":"审查、修改并扩展代码。永远不会被厂商锁定。"},"noLimits":{"title":"没有词数限制：","body":"免费计划即享无限本地听写。Typeless 将免费用量限制在每周 4,000 词。"},"linux":{"title":"支持 Linux：","body":"此处唯一能在 Linux 上运行的选项。Typeless 仅覆盖 Mac、Windows 和移动端。"}},"whyTypeless":{"title":"为什么选 Typeless？","smartEditing":{"title":"智能编辑：","body":"自动去除口头语、处理自我纠错，并为列表和步骤排版，无需手动整理。"},"tone":{"title":"情境感知语气：","body":"根据当前应用调整写作风格——邮件中专业，消息里随意。"},"mobile":{"title":"原生移动应用：","body":"免费包含完整的 iOS 和 Android 键盘应用，取代你默认的语音输入。"},"voiceEditing":{"title":"语音编辑命令：","body":"选中文字后开口说话，即可将其精简、改变语气、生成回复或即时翻译。"}},"pricing":{"title":"真实成本对比","ow":{"freeLabel":"永久免费","free1":"无限本地听写","free2":"本地 Whisper 与 Parakeet 模型","free3":"完整离线支持","free4":"macOS、Windows 与 Linux","proLabel":"Pro","pro1":"无限云端转录","pro2":"每月 20 小时会议录音","pro3":"移动应用","pro4":"邮件支持"},"typeless":{"freeLabel":"每周 4,000 词","free1":"包含 AI 自动编辑","free2":"支持 100+ 种语言","free3":"个人词典","proLabel":"Pro（年付）","pro1":"每周无限词数","pro2":"优先功能需求","pro3":"抢先体验新功能","pro4":"按月付费为 $30/月"},"callout":{"lead":"关键的取舍：","body":"Typeless 提供精炼的 AI 编辑，但需要联网且对免费用量设限。OpenWhispr 让你无限离线听写，并可完整访问源代码。"}},"bestFor":{"title":"适合不同的用户","openWhisprTitle":"OpenWhispr 最适合：","openWhisprItems":["在离线状态下处理敏感数据、注重隐私的用户","没有其他智能听写选择的 Linux 用户","需要无限听写、不受每周上限约束的重度用户","想要审查和扩展自己工具的开发者","需要会议录音和转录的团队"],"typelessTitle":"Typeless 最适合：","typelessItems":["想要自动整理、精炼输出的写作者","以 iOS 和 Android 移动端为先的用户","在正式与随意情境间切换的职场人士","句中混用多种语言的多语用户","偏好免设置托管服务的用户"]},"evidence":{"title":"基于实证的说明","body":"本对比基于 Typeless 的官方网站、定价页面、App Store 与 Play Store 上架信息，以及公开的产品文档。定价与功能截至 2026 年 3 月准确无误。"},"migration":{"title":"从 Typeless 切换到 OpenWhispr","intro":"正在使用 Typeless，想尝试开源方案？方法如下：","step1":{"title":"下载 OpenWhispr","body":"从 GitHub 获取免费版本，或升级到 Pro 以使用云端和移动功能。"},"step2":{"title":"选择你的模型","body":"挑选一个本地 Whisper 或 Parakeet 模型。无需联网，没有词数限制。"},"step3":{"title":"无限制听写","body":"没有每周词数上限，核心功能无需订阅。"},"callout":{"lead":"舍不得智能编辑？","body":"OpenWhispr 在 Pro 上支持借助云端模型进行 AI 后处理。输出干净且带标点，不过 Typeless 的口头语去除和语气适配更为先进。"}},"bottomLine":{"title":"结论","intro":"Typeless 和 OpenWhispr 解决的是同一个问题——把语音变成文字——但路线根本不同。Typeless 押注于 AI 驱动的智能编辑和广泛的移动覆盖。OpenWhispr 押注于隐私、开放和无限的离线访问。","teamTypeless":{"title":"Typeless 阵营","body":"你想要从语音直接得到精炼、可直接发送的文字。口头语去除、语气匹配和原生移动键盘支持，让听写轻松自如。"},"teamOpenWhispr":{"title":"OpenWhispr 阵营","body":"你想要完全的掌控。离线处理、没有词数限制、支持 Linux，并能审查处理你语音数据的每一行代码。"},"closing":"不同的理念，同样有效。选择与你工作方式相契合的那一个。"},"finalCta":{"title":"准备好掌控自己的语音数据了吗？","body":"无限离线听写、没有每周词数上限，以及开源带来的透明。免费试用 OpenWhispr。","starButton":"在 GitHub 上给我们点 Star","footnote":"无需信用卡 • 离线可用 • 永久开源","viewAllComparisons":"查看所有对比"}},"compareVoiceink":{"metadata":{"title":"VoiceInk 替代方案 — 免费跨平台听写","description":"比较两款开源听写应用。OpenWhispr 永久免费，支持 Mac、Windows、Linux 和 iOS；VoiceInk 则仅支持 macOS。","keywords":["OpenWhispr 对比 VoiceInk","VoiceInk 替代品","语音听写对比","开源听写","whisper 语音转文字","nvidia parakeet","macOS 听写应用","跨平台语音听写","AI 转录对比","本地语音转文字","隐私语音听写","2025 最佳听写应用","VoiceInk 评测"]},"breadcrumb":{"home":"首页","comparisons":"对比","voiceink":"对比 VoiceInk"},"hero":{"title":"OpenWhispr 对比 VoiceInk","subtitle":"两款基于 Whisper、隐私优先的开源听写应用。让我们把跨平台的竞争者与原生 Mac 的强力之选放在一起比一比。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"快速速览","voiceInkIntro":"是一款用 Swift 打造的原生 macOS 听写应用，具备出色的情境感知 AI 功能。它采用 whisper.cpp 进行本地转录，提供按应用配置的 Power Mode，并内置可自定义术语的个人词典。它在 GPLv3 协议下开源，拥有 3,700+ GitHub 星标的活跃社区。售价 39.99 美元的付费授权，对于想要精致、离线优先语音输入体验的 Mac 用户来说，可以是物超所值之选。","openWhisprIntro":"则采用了不同的思路。它完全免费，可在 macOS、Windows、Linux 和 iOS 上运行，并同时提供本地与云端转录选项。借助用于 AI 后处理的智能体模式和自定义系统提示词，它让你灵活掌控语音转文字的方式。可选的 Pro 套餐（每位用户每月 8 美元）增加了无限云端转录、更多会议时长以及移动 App，但核心应用本身完全免费且功能齐全。","chooseVoiceInk":"如果你符合以下情况，请选择 VoiceInk：你使用搭载 Apple Silicon 的 Mac，并希望开箱即用地获得屏幕感知 AI 模式、个人词典和按应用设置。","chooseOpenWhispr":"如果你符合以下情况，请选择 OpenWhispr：你需要跨平台支持、想要永久免费的选项，或更倾向于在本地处理之外保留 BYOK 云端的灵活性。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"隐私与数据控制","openWhispr":{"status":"full","details":"100% 本地处理选项，自带 API 密钥"},"voiceInk":{"status":"full","details":"100% 离线处理，无需云端"}},{"feature":"开源","openWhispr":{"status":"full","details":"在 GitHub 上完全开源"},"voiceInk":{"status":"full","details":"基于 GPLv3 协议开源"}},{"feature":"平台支持","openWhispr":{"status":"full","details":"macOS、Windows、Linux、iOS"},"voiceInk":{"status":"partial","details":"仅限 macOS（需 Apple Silicon）"}},{"feature":"定价模式","openWhispr":{"status":"full","details":"永久免费，可选每月 8 美元的 Pro"},"voiceInk":{"status":"partial","details":"付费授权（请查看官方定价）"}},{"feature":"AI 模型选项","openWhispr":{"status":"full","details":"OpenAI Whisper 和 NVIDIA Parakeet 模型"},"voiceInk":{"status":"full","details":"通过 whisper.cpp 支持 Whisper + Parakeet 模型"}},{"feature":"系统集成","openWhispr":{"status":"full","details":"可在任意应用中使用"},"voiceInk":{"status":"full","details":"系统级支持，并提供按应用配置的 Power Mode"}},{"feature":"AI 后处理","openWhispr":{"status":"full","details":"带自定义系统提示词的智能体模式"},"voiceInk":{"status":"full","details":"具备屏幕感知情境的 Smart Modes"}},{"feature":"自定义","openWhispr":{"status":"full","details":"完全可自定义，便于二次开发"},"voiceInk":{"status":"full","details":"个人词典、文本替换、Power Mode"}}],"whyOpenWhispr":{"title":"为什么选择 OpenWhispr？","crossPlatform":{"title":"真正的跨平台：","body":"可在 Mac、Windows、Linux 和 iOS 上运行。而 VoiceInk 仅支持搭载 Apple Silicon 的 Mac——不支持 Intel Mac、不支持 Windows、不支持 Linux。"},"freeCore":{"title":"核心功能永久免费：","body":"所有核心功能均为 0 元。无需 39.99 美元的前期费用——立即上手，只有当你想要 Pro 带来的便利时才付费。"},"flexibility":{"title":"云端 + 本地双重灵活：","body":"为隐私选择本地处理，或为速度选择云端转录。自带 API 密钥，掌控全局。"},"hackable":{"title":"可改造、可扩展：","body":"基于 Electron 和 React 打造——Web 开发者熟悉的技术栈。尽情 fork、扩展，让它真正属于你。"}},"whyVoiceInk":{"title":"为什么选择 VoiceInk？","native":{"title":"原生 macOS 性能：","body":"用 Swift 打造，针对 Apple Silicon 优化。充分利用神经网络引擎实现快速、高效的设备端转录。"},"smartModes":{"title":"屏幕感知 Smart Modes：","body":"情境感知 AI 会读取你的屏幕内容，带来更贴切的转录与排版。"},"powerMode":{"title":"Power Mode：","body":"自动按应用和网址应用不同设置。你的听写会自动适配 Slack、邮件、代码编辑器等场景。"},"dictionary":{"title":"个人词典：","body":"用你的自定义术语、人名、缩写和行话训练系统。还支持用于快捷输入的文本替换。"}},"pricing":{"title":"真实成本对比","ow":{"freeLabel":"永久免费","free1":"包含所有核心功能","free2":"本地 + 云端转录","free3":"带系统提示词的智能体模式","free4":"macOS、Windows、Linux","proLabel":"Pro","pro1":"无限云端转录","pro2":"每月 20 小时会议录音","pro3":"移动 App","pro4":"邮件支持"},"voiceInk":{"freeAmount":"免费","freeLabel":"试用","free1":"先试后买","free2":"从源码构建（GPLv3）","free3":"核心转录功能","paidLabel":"一次性购买","paid1":"终身使用全部功能","paid2":"包含未来更新","paid3":"优先支持","paid4":"提供学生折扣"},"callout":{"lead":"两种诚实的模式：","body":"如果你笃定使用 macOS，VoiceInk 39.99 美元的付费方案可以非常物超所值。OpenWhispr 永久免费，并提供可选的 Pro 订阅——非常适合想先试后买，或需要跨平台支持的用户。"}},"featureShowdown":{"title":"功能巅峰对决","bothShareTitle":"两者共有的功能：","bothShare":["基于 Whisper 的本地转录","100% 离线隐私选项","全局快捷键激活","系统级支持，可在任意应用中使用","可获取开源代码","AI 增强的文本排版","多语言支持","活跃的社区开发"],"openWhisprExclusivesTitle":"OpenWhispr 独有功能：","openWhisprExclusives":["跨平台（Mac、Windows、Linux、iOS）","支持 Intel 与 Apple Silicon Mac","支持 BYOK API 密钥的云端转录","带自定义系统提示词的智能体模式","永久免费，不设功能门槛"],"voiceInkExclusivesTitle":"VoiceInk 独有功能：","voiceInkExclusives":["针对 Apple Silicon 优化的原生 Swift 应用","用于情境感知的屏幕感知 Smart Modes","支持按应用/网址设置的 Power Mode","个人词典与文本替换","在 Whisper 之外支持 Parakeet 模型"]},"evidence":{"title":"基于实证的说明","body":"本对比依据厂商文档、定价页面及公开产品资料，而非道听途说的用户评价。"},"migration":{"title":"从 VoiceInk 切换到 OpenWhispr","intro":"喜欢 VoiceInk，但需要跨平台支持或想避免前期费用？以下是切换方法：","step1":{"title":"下载 OpenWhispr","body":"免费下载。可选 Pro（每位用户每月 8 美元），享受无限云端转录、更多会议时长以及移动 App。支持任意平台。"},"step2":{"title":"设置你的快捷键与模型","body":"沿用你在 VoiceInk 中的键盘快捷键，并选择 Whisper 模型大小。建议从 “small” 开始，在速度与准确度间取得良好平衡。"},"step3":{"title":"配置智能体模式","body":"使用 OpenWhispr 的智能体模式和自定义系统提示词，重现你在 VoiceInk 中 Smart Mode 的排版与情境行为。"},"step4":{"title":"迈向多平台","body":"也把它装到你的 Windows 或 Linux 设备上。同样的工作流，同样的质量——伴你工作的每一处。"},"callout":{"lead":"少了 Power Mode？","body":"OpenWhispr 的智能体模式搭配自定义系统提示词，能带来类似的 AI 后处理灵活性。社区也在打造更多情境感知功能——快来 GitHub 加入我们，一起塑造路线图！"}},"bottomLine":{"title":"结论","intro":"这是一场难得的对比——两款工具都开源且注重隐私。最终的选择，真正取决于你的平台和工作流偏好：","teamVoiceInk":{"title":"VoiceInk 阵营","body":"你全身心投入搭载 Apple Silicon 的 macOS，想要原生 Swift 性能，而 Power Mode、屏幕感知 Smart Modes 等功能也完美契合你的工作流。为你所得到的体验，付费授权可以物有所值。"},"teamOpenWhispr":{"title":"OpenWhispr 阵营","body":"你跨平台工作，想要一款没有前期费用的免费听写工具，或需要用自己的 API 密钥进行云端转录的灵活性。智能体模式让你零成本获得强大的 AI 后处理能力。"},"closing":"两款应用都拥护开源与隐私。最佳选择取决于你是身处 Apple 生态，还是需要在任何地方自由工作。"},"finalCta":{"title":"准备好试试这款跨平台替代方案了吗？","body":"获得与 VoiceInk 同样强大的 Whisper 和 Parakeet 转录能力——覆盖每个平台，完全免费。无需 Apple Silicon，无需前期购买，隐私毫不妥协。","starButton":"在 GitHub 上点亮 Star","footnote":"无需信用卡 • 可离线使用 • 永久开源"},"viewAllComparisons":"← 查看所有听写应用对比"},"compareWillowvoice":{"metadata":{"title":"Willow Voice 替代方案：本地、免费、开源","description":"比较 OpenWhispr 与 Willow Voice：本地开源听写，对比覆盖桌面端和移动端的专有云端语音写作。","keywords":["willow voice 替代品","willow voice 对比 openwhispr","云端听写应用","语音写作软件","nvidia parakeet"]},"breadcrumb":{"home":"首页","comparisons":"对比","willowVoice":"对比 Willow Voice"},"hero":{"title":"OpenWhispr 对比 Willow Voice","subtitle":"对比两款强大的语音转文字方案。一款开放免费，一款专有。让我们帮你做出选择。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"快速速览","willowVoiceIntro":"是一款专有听写工具，专注于快速的云端语音写作、命令工作流以及改写/语气预设。它支持桌面和移动端使用场景，提供基于账户的同步与托管式更新。","openWhisprIntro":"则采用了不同的思路。它完全开源，可在你的所有应用中使用（而不仅仅是浏览器），并让你完全掌控自己的数据。你可以完全离线运行，也可以使用自己的 API 密钥。而且，只要你乐于使用开源软件，它就永久免费。","chooseWillowVoice":"如果你符合以下情况，请选择 Willow Voice：你想要一套托管的云端工作流，内置改写/语气工具与账户同步。","chooseOpenWhispr":"如果你符合以下情况，请选择 OpenWhispr：你重视隐私、想要完全掌控，或更偏好开源软件。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"隐私与数据控制","openWhispr":{"status":"full","details":"100% 本地处理选项，自托管 API 密钥"},"willowVoice":{"status":"partial","details":"云端优先处理，配有账户级控制"}},{"feature":"开源","openWhispr":{"status":"full","details":"在 GitHub 上完全开源"},"willowVoice":{"status":"none","details":"专有软件"}},{"feature":"平台支持","openWhispr":{"status":"full","details":"系统级支持，可在任意应用中使用"},"willowVoice":{"status":"full","details":"支持 macOS、Windows 以及 iPhone App"}},{"feature":"定价模式","openWhispr":{"status":"full","details":"永久免费，并提供付费 Pro 选项"},"willowVoice":{"status":"partial","details":"免费套餐含 10,000 字，每周补充 2,000 字；付费方案每月 15 美元起"}},{"feature":"语音命令","openWhispr":{"status":"partial","details":"用于标点的基础命令"},"willowVoice":{"status":"full","details":"高级听写命令、预设与编辑流程"}},{"feature":"准确度","openWhispr":{"status":"full","details":"业界领先的 Whisper 和 Parakeet 模型"},"willowVoice":{"status":"full","details":"云端 AI 听写，带改写/语气选项"}},{"feature":"自定义","openWhispr":{"status":"full","details":"完全可自定义，便于二次开发"},"willowVoice":{"status":"partial","details":"可配置的预设与命令（不支持开源扩展）"}},{"feature":"离线支持","openWhispr":{"status":"full","details":"完整的离线功能"},"willowVoice":{"status":"partial","details":"核心听写流程需要联网"}}],"whyOpenWhispr":{"title":"为什么选择 OpenWhispr？","privacy":{"title":"真正的隐私：","body":"一切均在本地处理。除非你选择使用云端 API，否则你的语音绝不离开你的设备。"},"crossPlatform":{"title":"真正的跨平台：","body":"支持 macOS、Windows、Linux 和 iOS。而 Willow Voice 支持 macOS 和 Windows，外加一款 iPhone App。"},"openSource":{"title":"开源的自由：","body":"修改它、扩展它，或回馈社区。绝无厂商锁定，永远不会。"},"cost":{"title":"高性价比：","body":"本地处理永久免费，或使用你自己的 API 密钥获得云端功能。"}},"whyWillowVoice":{"title":"为什么选择 Willow Voice？","voiceCommands":{"title":"语音命令：","body":"内置命令工作流，快速编辑和排版听写出的文本。"},"autoEditing":{"title":"AI 自动编辑：","body":"专为云端 AI 后处理设计的改写、语气与写作预设。"},"managedService":{"title":"托管服务：","body":"托管式云端产品，支持账户同步与自动更新。"},"commandMode":{"title":"命令模式：","body":"为常见写作任务提供预设驱动的听写与编辑流程。"}},"pricing":{"title":"定价详解","ow":{"freeLabel":"永久免费","freeDetails":"本地处理，无限使用","proLabel":"Pro","proDetails":"无限云端、移动 App、技术支持"},"willow":{"freeLabel":"免费","freeWords":"10,000 字 + 每周补充 2,000 字","freeDetails":"入门级用量，需注册账户","proLabel":"Pro / Team 方案","proDetails":"更多字数、高级功能与团队选项"}},"useCases":{"title":"适合不同的用户","openWhisprTitle":"OpenWhispr 最适合：","willowVoiceTitle":"Willow Voice 最适合：","openWhisprList":["想要自定义工具的开发者","处理敏感数据、注重隐私的用户","需要系统级听写支持的团队","注重预算的个人与初创团队","在多个应用之间切换工作的用户"],"willowVoiceList":["偏好托管云端听写工作流的用户","使用改写/语气预设的写作者","希望获得账户同步与集中管理的团队","不想进行自托管或开源配置的人","能接受云端处理要求的用户"]},"migration":{"title":"从 Willow Voice 切换到 OpenWhispr","intro":"切换比你想象的更简单。方法如下：","step1":{"title":"下载 OpenWhispr","body":"从 GitHub 获取免费版本，或选择 Pro 实现即时配置。"},"step2":{"title":"配置你的偏好","body":"设置你的快捷键，选择本地或云端处理，并挑选你的模型。"},"callout":{"lead":"专业提示：","body":"在过渡期间，你可以同时使用这两款工具。OpenWhispr 不会干扰 Willow Voice 的桌面/移动端工作流。"}},"finalCta":{"title":"准备好掌控你的语音数据了吗？","body":"加入数以千计已转用 OpenWhispr 的用户，享受真正的隐私、系统级支持以及开源的自由。","starButton":"在 GitHub 上探索","footnote":"无需信用卡 • 可离线使用 • 永久开源","viewAllComparisons":"查看所有对比"}},"compareWindowsSpeech":{"metadata":{"title":"Windows 语音输入替代方案 — OpenWhispr","description":"从 Windows 语音输入切换到 OpenWhispr：基于 Whisper 模型的开源 AI 听写，100% 离线保护隐私，免费支持多平台。","keywords":["Windows voice typing 替代方案","更好的 Windows 听写","Windows 语音识别升级","Windows 语音识别替代方案","Win+H 替代方案","Whisper Windows 语音输入","nvidia parakeet","Windows 离线听写","Windows AI 听写"]},"breadcrumb":{"home":"首页","comparisons":"对比","windowsSpeech":"对比 Windows Speech"},"hero":{"title":"OpenWhispr 对比 Windows Voice Typing","subtitle":"Windows 自带听写功能，但它够用吗？将这款免费内置工具与顶尖的 AI 语音识别一较高下。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"快速概览","windowsIntro":"（Win+H）是一款便捷的内置听写工具，随每一份 Windows 11 免费附带。它能很好地处理基础听写，支持 45 种以上语言，并提供用于编辑的语音命令。不过，它要获得最佳效果需依赖网络连接，使用微软专有的语音引擎，且可定制性极为有限。Copilot+ PC 上较新的“Fluid Dictation”功能新增了语法纠正，但仅支持英语。","openWhisprIntro":"将 OpenAI Whisper 和 NVIDIA Parakeet 模型带到你的桌面 —— 正是这一技术驱动着全球部分最精准的语音识别。它 100% 本地运行，可在 Windows、Mac 和 Linux 上使用，并让你通过 agent 模式全面掌控模型大小、热键以及 AI 驱动的文本清理。它免费、开源，绝不会把你的语音数据发往任何地方。","stickWith":"继续使用 Windows Voice Typing，如果：你只需要快速、偶尔的听写，且不想安装任何东西。","chooseOpenWhispr":"选择 OpenWhispr，如果：你想要显著更高的准确度、离线隐私、AI 文本清理，或需要跨多个平台工作。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"语音识别引擎","openWhispr":{"status":"full","details":"OpenAI Whisper 和 NVIDIA Parakeet 模型"},"windowsSpeech":{"status":"partial","details":"微软专有引擎"}},{"feature":"离线处理","openWhispr":{"status":"full","details":"100% 本地，无需联网"},"windowsSpeech":{"status":"partial","details":"基础离线；最佳准确度需联网"}},{"feature":"开源","openWhispr":{"status":"full","details":"在 GitHub 上完全开源"},"windowsSpeech":{"status":"none","details":"闭源、专有"}},{"feature":"平台支持","openWhispr":{"status":"full","details":"macOS、Windows、Linux、iOS"},"windowsSpeech":{"status":"partial","details":"仅限 Windows"}},{"feature":"AI 模型选择","openWhispr":{"status":"full","details":"可从 tiny、base、small、medium、large 中选择"},"windowsSpeech":{"status":"none","details":"无模型可选，单一引擎"}},{"feature":"智能体模式 / AI 清理","openWhispr":{"status":"full","details":"用于格式化与清理的系统提示词"},"windowsSpeech":{"status":"none","details":"无 AI 后处理"}},{"feature":"语言支持","openWhispr":{"status":"full","details":"通过 Whisper 和 Parakeet 支持 99 种以上语言"},"windowsSpeech":{"status":"partial","details":"45 种以上语言（参见微软语言列表）"}},{"feature":"可定制性","openWhispr":{"status":"full","details":"热键、模型、系统提示词、BYOK"},"windowsSpeech":{"status":"partial","details":"自动标点、麦克风选择"}}],"whyOpenWhispr":{"title":"为什么升级到 OpenWhispr？","accuracy":{"title":"Whisper 与 Parakeet 准确度：","body":"OpenAI Whisper 和 NVIDIA Parakeet 在各种口音、背景噪音和专业词汇上都能带来顶尖的准确度。可从 tiny（快速）到 large（最高准确度）任选。"},"offline":{"title":"真正的离线隐私：","body":"每一个字都在你自己的机器上处理。没有云端、不收集数据、永远无需联网。"},"agentMode":{"title":"智能体模式清理：","body":"AI 驱动的系统提示词会自动修正语法、去除口头禅、格式化文本 —— 远超基础听写。"},"crossPlatform":{"title":"跨平台：","body":"可在 Windows、macOS 和 Linux 上使用。一款工具，通吃所有操作系统。带着你的工作流去任何地方。"}},"whyWindows":{"title":"Windows Voice Typing 的优势所在","zeroSetup":{"title":"零设置：","body":"已预装在每一台 Windows 11 PC 上。按下 Win+H 即可开口说话 —— 无需下载，无需配置。"},"voiceCommands":{"title":"语音命令：","body":"内置“删除那段”“选中那段”“撤销那段”等命令，并可按名称添加标点。便于快速编辑。"},"autoPunctuation":{"title":"自动标点：","body":"无需说出标点名称即可自动插入标点。在设置中一键开关。"},"fluidDictation":{"title":"Fluid Dictation（Copilot+ PC）：","body":"在较新的 Copilot+ 硬件上，语法纠正和口头禅去除在设备端运行。仅支持英语。"}},"pricing":{"title":"两者都免费 —— 但并不对等","ow":{"freeLabel":"免费且开源","free1":"内含 Whisper 和 Parakeet 模型","free2":"带系统提示词的 agent 模式","free3":"100% 离线，无需联网","free4":"macOS、Windows、Linux","free5":"99 种以上语言","free6":"支持 BYOK 云端 API","proLabel":"Pro（可选）","pro1":"无限云端转录","pro2":"每月 20 小时会议录制","pro3":"移动应用","pro4":"邮件支持"},"windows":{"builtInLabel":"内置于 Windows 11","pro1":"无需安装","pro2":"基础听写与语音命令","pro3":"自动标点","con1":"最佳准确度需联网","con2":"无模型可选","con3":"无 AI 清理或格式化","con4":"仅限 Windows"},"copilot":{"title":"需要 Copilot+ PC","subtitle":"用于 Fluid Dictation","pro1":"设备端语法纠正","pro2":"口头禅去除","con1":"仅支持英语","con2":"需要特定硬件"},"callout":{"lead":"两者都免费，但“免费”并非全部。","body":"Windows Voice Typing 虽便捷却有局限。OpenWhispr 带给你 Whisper 和 Parakeet AI 的准确度、完整的离线隐私、用于智能文本清理的 agent 模式，以及开源的自由 —— 全部 $0。"}},"featureShowdown":{"title":"功能对决","bothHaveTitle":"两者都具备的功能：","bothHave":["免费使用","可在任意文本框中工作","多语言支持","自动标点","按键说话激活","实时转录"],"openWhisprExclusivesTitle":"OpenWhispr 独有：","openWhisprExclusives":["顶尖的 Whisper 和 Parakeet AI 模型","智能体模式：AI 驱动的文本清理与格式化","自选模型大小（速度与准确度的权衡）","100% 离线 —— 不联网，没有例外","跨平台（macOS、Windows、Linux、iOS）","开源 —— 可审查、可修改、可贡献","99 种以上语言，准确度高","BYOK：使用你自己的云端 API 密钥"],"windowsExclusivesTitle":"Windows Voice Typing 独有：","windowsExclusives":["预装，零设置","内置语音编辑命令","Copilot+ PC 上的 Fluid Dictation","与 Windows 无障碍功能集成"]},"evidence":{"title":"基于事实的说明","body":"本对比基于厂商文档、定价页面和公开产品文档，而非道听途说的用户推荐。"},"migration":{"title":"从 Windows Voice Typing 升级","intro":"准备好体验更准确、更私密、AI 驱动的听写了吗？上手方法如下：","step1":{"title":"下载 OpenWhispr","body":"免费下载。可选 Pro（每用户每月 $8），享受无限云端转录、更多会议时长和移动应用。"},"step2":{"title":"挑选你的 AI 模型","body":"从“base”开始，兼顾速度与准确度。当你需要极致精度时，升级到“large”。"},"step3":{"title":"设置你的热键","body":"用你喜欢的按键配置按键说话。用起来就像 Win+H，但快捷键由你自定。"},"step4":{"title":"试试智能体模式","body":"设置系统提示词，自动修正语法、去除口头禅，或针对特定场景格式化文本。"},"callout":{"lead":"你可以两者兼用。","body":"OpenWhispr 和 Windows Voice Typing 可以并行运行。下次写作时试试 OpenWhispr —— 我们相信你会立刻察觉到准确度上的差别。"}},"bottomLine":{"title":"结论","intro":"Windows Voice Typing 是一款用于日常听写的可靠内置工具。它便捷、免费，应付快速记笔记绰绰有余。但如果你在意准确度、隐私或强大功能，它就力有未逮。","windowsFineFor":{"title":"Windows Voice Typing 适合","body":"快速消息、随手笔记，以及便捷比准确度更重要的场合。无需安装、无需配置 —— 按下 Win+H 即可。"},"openWhisprBuiltFor":{"title":"OpenWhispr 专为以下人群打造","body":"作家、开发者、专业人士，以及任何追求极致准确度、彻底隐私、AI 驱动文本清理和开源自由的人。处处可用 —— 不止 Windows。"},"closing":"Windows 给了你基础，OpenWhispr 给你升级。"},"finalCta":{"title":"准备好升级你的 Windows 听写了吗？","body":"获得 Whisper 和 Parakeet 的准确度、本地处理的隐私，以及 agent 模式的强大 —— 全部免费，全部开源。","starButton":"在 GitHub 上为我们点 Star","badge1":"无需信用卡","badge2":"100% 离线工作","badge3":"永久开源","viewAllComparisons":"查看所有对比"}},"compareWisprflow":{"metadata":{"title":"Wispr Flow 替代方案：免费、本地、开源","description":"OpenWhispr 是 Wispr Flow 的免费开源替代方案，完全在你的设备上运行。听写私密且离线，无需订阅。","keywords":["wisprflow 替代方案","wisprflow 对比 openwhispr","AI 听写","语音转文字","nvidia parakeet"]},"breadcrumb":{"home":"首页","comparisons":"对比","wisprflow":"对比 WisprFlow"},"hero":{"title":"OpenWhispr：开源 Wispr Flow 替代品","subtitle":"Wispr Flow 与 OpenWhispr 比较。如果您想要一个完全在您自己的设备上运行的私密、开源语音到文本工具，以下是两者的对比。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"快速概览","wisprFlowIntro":"是一款人工智能驱动的听写工具，适用于任何应用程序，从 Slack 到电子邮件再到人工智能聊天界面。它专注于具有高级语音命令和自动编辑功能的快速云听写。","openWhisprIntro":"则另辟蹊径。它完全开源，可在你所有的应用中工作（不止浏览器），并让你全面掌控自己的数据。你可以让它完全离线运行，也可以使用自己的 API 密钥。而且，只要你能接受开源，它就永久免费。","chooseWisprFlow":"如果满足以下条件，请选择 Wispr Flow：您需要人工智能驱动的自动编辑、上下文感知写作和高级语音命令。","chooseOpenWhispr":"选择 OpenWhispr，如果：你重视隐私、想要完全掌控，或更青睐开源软件。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"隐私与数据掌控","openWhispr":{"status":"full","details":"可选 100% 本地处理，自托管 API 密钥"},"wisprFlow":{"status":"partial","details":"基于云端，附带隐私控制"}},{"feature":"开源","openWhispr":{"status":"full","details":"在 GitHub 上完全开源"},"wisprFlow":{"status":"none","details":"专有软件"}},{"feature":"平台支持","openWhispr":{"status":"full","details":"在任意应用中全系统通用"},"wisprFlow":{"status":"full","details":"处处可用 —— 任意应用的任意文本框"}},{"feature":"定价模式","openWhispr":{"status":"full","details":"永久免费，另有付费 pro 选项"},"wisprFlow":{"status":"partial","details":"免费档（桌面端每周 2,000 词），另有付费 Pro 套餐"}},{"feature":"语音命令","openWhispr":{"status":"partial","details":"用于标点的基础命令"},"wisprFlow":{"status":"full","details":"高级语音命令与快捷方式"}},{"feature":"准确度","openWhispr":{"status":"full","details":"顶尖的 Whisper 和 Parakeet 模型"},"wisprFlow":{"status":"full","details":"高准确度，配 AI 驱动的自动编辑"}},{"feature":"可定制性","openWhispr":{"status":"full","details":"完全可定制，便于 fork"},"wisprFlow":{"status":"partial","details":"仅限于内置设置"}},{"feature":"离线支持","openWhispr":{"status":"full","details":"完整的离线功能"},"wisprFlow":{"status":"partial","details":"部分功能需联网"}}],"whyOpenWhispr":{"title":"为什么选择 OpenWhispr？","privacy":{"title":"真正的隐私：","body":"一切都在本地处理。除非你选择使用云端 API，否则你的语音绝不会离开你的设备。"},"crossPlatform":{"title":"真正的跨平台：","body":"支持 macOS、Windows、Linux 和 iOS。WisprFlow 仅支持 Mac 和 Windows。"},"openSource":{"title":"开源自由：","body":"可修改、可扩展、可回馈贡献。永远不被厂商绑定。"},"cost":{"title":"高性价比：","body":"本地处理永久免费，或使用你自己的 API 密钥解锁云端功能。"}},"whyWisprFlow":{"title":"为什么选择 WisprFlow？","voiceCommands":{"title":"语音命令：","body":"高级语音控制系统，支持用于复杂工作流的自定义命令和宏。"},"autoEditing":{"title":"AI 自动编辑：","body":"情境感知 AI，能判断你是在写邮件、短信还是文档，并相应调整。"},"managedService":{"title":"托管服务：","body":"无需设置。更新与改进自动完成。"},"commandMode":{"title":"命令模式：","body":"语音编辑命令，比如“改得正式些”或“变成要点列表”。"}},"pricing":{"title":"定价详解","ow":{"freeLabel":"永久免费","freeDetails":"本地处理，无限使用","proLabel":"Pro","proDetails":"无限云端、移动应用、支持服务"},"wisprFlow":{"freeLabel":"免费","freeSub":"有限的免费套餐","freeDetails":"基础功能，用量受限","proLabel":"Pro","proSub":"订阅制","proDetails":"无限词数，全部功能"}},"useCases":{"title":"适合不同用户","openWhisprTitle":"OpenWhispr 最适合：","openWhisprList":["想要定制工具的开发者","处理敏感数据、注重隐私的用户","需要全系统听写支持的团队","注重预算的个人和初创公司","需要跨多个应用工作的用户"],"wisprFlowTitle":"WisprFlow 最适合：","wisprFlowList":["需要高级语音命令的资深用户","需要 AI 驱动自动编辑的写作者","想要现成工作流自动化的用户","比起亲自折腾更偏爱托管服务的人","需要多语言云端听写的用户"]},"migration":{"title":"从 WisprFlow 切换到 OpenWhispr","intro":"切换比你想的更简单。方法如下：","step1":{"title":"下载 OpenWhispr","body":"从 GitHub 获取免费版，或升级 Pro 享受即装即用。"},"step2":{"title":"配置你的偏好","body":"设置热键，选择本地或云端处理，并挑选你的模型。"},"callout":{"lead":"专业提示：","body":"过渡期间你可以两款工具并用。OpenWhispr 不会干扰 WisprFlow 的浏览器扩展。"}},"finalCta":{"title":"准备好掌控你的语音数据了吗？","body":"加入成千上万已转向 OpenWhispr 的用户，享受真正的隐私、全系统支持和开源的自由。","starButton":"在 GitHub 上探索","footnote":"无需信用卡 • 离线工作 • 永久开源","viewAllComparisons":"查看所有对比"}},"useCasesHub":{"metadata":{"title":"适合各种工作流的听写软件 | OpenWhispr","description":"探索听写软件在医疗、法律、开发、写作、Mac、Windows 和 Linux 中的用法。私密、本地、开源，找到适合你的工作流。","keywords":["语音听写应用场景","面向开发者的语音转文字","面向写作者的听写","语音输入工作流","openwhispr 应用场景"]},"breadcrumb":{"home":"首页","useCases":"应用场景"},"hero":{"title":"应用场景","subtitle":"语音听写从来不是“一招通吃”。看看 OpenWhispr 如何融入你的具体工作流——从编写代码到起草法律文书。","description":"每个应用场景页面都涵盖配置技巧、真实示例，以及对你的职业或平台最关键的功能。"},"learnMore":"了解更多 →","useCases":[{"title":"对于开发人员","href":"/use-cases/developers","description":"无需离开 IDE，即可听写代码注释、文档、LLM 提示词和提交消息。","icon":"💻","accent":"from-blue-500 to-blue-600"},{"title":"对于产品团队","href":"/use-cases/product-teams","description":"面向产品团队的 AI 会议纪要。没有机器人加入会议，使用本地模型即可在您的电脑上转录，并整理出决策、行动项和负责人。","icon":"🗺️","accent":"from-violet-500 to-violet-600"},{"title":"对于Linux","href":"/use-cases/linux","description":"适用于 Linux 的语音转文字软件，提供本地或云端转录以及与会话匹配的设置指引。","icon":"🐧","accent":"from-orange-500 to-orange-600"},{"title":"面向 Google 文档用户","href":"/use-cases/google-docs","description":"从桌面端向当前 Google 文档语音输入，并直接在文档中检查可编辑文本。","icon":"✍️","accent":"from-purple-500 to-purple-600"},{"title":"Ubuntu 语音输入","href":"/use-cases/ubuntu","description":"在 Ubuntu 光标处将语音转换为可编辑文本，可选择云端或本地转写。","icon":"🎙️","accent":"from-orange-500 to-amber-600"},{"title":"对于作家","href":"/use-cases/writers","description":"口述你的草稿，拿回你自己的句子。书籍、文章和新闻通讯，都在你自己的电脑上转录。","icon":"✍️","accent":"from-purple-500 to-purple-600"},{"title":"面向 Microsoft Word 用户","href":"/use-cases/microsoft-word","description":"自动粘贴可将说出的文字放到 Word 当前活动光标处。否则请使用剪贴板。无需 Word 加载项。","icon":"⌨️","accent":"from-blue-500 to-blue-600"},{"title":"对于儿童","href":"/use-cases/kids","description":"面向家长和教育者的语音转文字软件，帮助 13 岁及以上的青少年说出想法，然后审阅并修改可编辑文本。","icon":"📝","accent":"from-emerald-500 to-emerald-600"},{"title":"对于学生","href":"/use-cases/students","description":"面向学生的免费语音转文字。录下课堂，拿到整理好的笔记，之后再口述论文，全部在自己的笔记本上完成，$0。","icon":"🎓","accent":"from-green-500 to-green-600"},{"title":"面向残障学生","href":"/use-cases/students-with-disabilities","description":"在您已经使用的桌面应用中口述作业，支持在线或离线使用。","icon":"📚","accent":"from-violet-500 to-violet-600"},{"title":"对于记者","href":"/use-cases/journalists","description":"面向记者和编辑的私密采访转录。在本地模式下，采访录音在本机转成带时间戳的逐字稿，不会上传给转录供应商。","icon":"📰","accent":"from-fuchsia-500 to-fuchsia-600"},{"title":"医疗听写","href":"/use-cases/medical","description":"临床医生的私人设备听写。将临床笔记和 SOAP 笔记听写到任何 EHR 中，无需将患者音频发送到云端。","icon":"🩺","accent":"from-blue-500 to-blue-600"},{"title":"面向牙医","href":"/use-cases/dentists","description":"将口述的临床记录转为可编辑文本。","icon":"🦷","accent":"from-cyan-500 to-blue-600"},{"title":"面向受 HIPAA 监管的团队","href":"/use-cases/hipaa-compliant","description":"面向受 HIPAA 监管的医疗团队的听写工具，提供本地转录路径，并可在 Business 套餐中申请适用于 OpenWhispr Cloud 的 BAA。","icon":"🛡️","accent":"from-emerald-500 to-teal-600"},{"title":"放射科","href":"/use-cases/radiology","description":"使用本地语音转文字，将口述影像所见输入报告的当前光标位置，让放射科医生专注查看图像。","icon":"🩻","accent":"from-violet-500 to-indigo-600"},{"title":"面向病理科医生","href":"/use-cases/pathology","description":"将大体描述和镜下所见直接听写到兼容的报告字段中，并在签发前逐字检查。","icon":"🔬","accent":"from-fuchsia-500 to-fuchsia-600"},{"title":"对于医生","href":"/use-cases/doctors","description":"转诊信、报告和病历记录，在您停止说话的那一刻就在您自己的电脑上打好。每位医生都免费。","icon":"⚕️","accent":"from-emerald-500 to-emerald-600"},{"title":"面向医生","href":"/use-cases/physicians","description":"一键将现病史、查体和评估内容输入到电子病历光标处。面向医生的免费设备端听写。","icon":"🩻","accent":"from-cyan-500 to-cyan-600"},{"title":"兽医听写","href":"/use-cases/veterinary","description":"兽医和兽医助理的私人设备听写。将 SOAP 笔记、出院说明和给畜主的电子邮件听写到任何兽医管理系统中，在农场出诊时也能离线使用。","icon":"🐾","accent":"from-emerald-500 to-emerald-600"},{"title":"法律听写","href":"/use-cases/lawyers","description":"面向律师和法律团队的私密听写。通过本地转录起草诉讼摘要、动议、备忘录和接案笔记，适用于受保密特权保护的工作。","icon":"⚖️","accent":"from-indigo-500 to-indigo-600"},{"title":"对于治疗师","href":"/use-cases/therapists","description":"治疗师的听写软件，用于会谈记录，提供 Cloud 与本地转录两种选择。","icon":"🧠","accent":"from-teal-500 to-teal-600"},{"title":"对于 ABA 治疗师","href":"/use-cases/aba-session-notes","description":"口述 ABA 会谈记录细节，然后检查并粘贴到您的记录系统中。","icon":"🗂️","accent":"from-amber-500 to-amber-600"},{"title":"法律文书","href":"/use-cases/legal-documents","description":"口述一条合同条款、一份证据开示答复或辩论状的一节，选一个本地 AI 模型，它就把这段口述起草成带标题和编号要点的结构化文本，撰写全程都在您自己的机器上。","icon":"📄","accent":"from-rose-500 to-rose-600"},{"title":"对于MacBook","href":"/use-cases/macbook","description":"听写功能，即使离线或使用电池也能持续工作，突破 macOS 30 秒的限制。","icon":"💻","accent":"from-gray-500 to-gray-700"},{"title":"对于 iPad","href":"/use-cases/ipad","description":"关于 OpenWhispr 在 iPad 上使用的直接解答：移动应用包含在 Pro 中，而 Free 版本从桌面开始使用。","icon":"📱","accent":"from-sky-500 to-sky-600"},{"title":"适用于 iPhone","href":"/use-cases/iphone","description":"一款适用于 iPhone 的听写应用。说出想法，获得清晰文本，并通过 Pro 在不同设备间继续工作。","icon":"📞","accent":"from-slate-500 to-slate-700"},{"title":"对于Windows","href":"/use-cases/windows","description":"在已打开的 Windows 应用中说话，即可在光标处获得可编辑的文本。","icon":"🪟","accent":"from-sky-500 to-sky-600"},{"title":"适用于 Windows 10","href":"/use-cases/windows-10","description":"专为 Windows 10 打造的语音转文字软件，无需升级到 Windows 11。","icon":"🪟","accent":"from-sky-500 to-sky-600"},{"title":"适用于 Windows 11","href":"/use-cases/windows-11","description":"适用于 Windows 11 的免费离线语音转文字。本地模型无需联网、无需 Copilot+ 硬件即可转写，费用为 0 美元。","icon":"🪟","accent":"from-sky-500 to-sky-600"},{"title":"面向 PC 用户","href":"/use-cases/pc","description":"适用于 Windows PC 写作的语音转文字软件。直接在您已使用的应用中听写，然后检查文本。","icon":"🖥️","accent":"from-sky-500 to-blue-600"},{"title":"笔记本电脑语音转文字应用","href":"/use-cases/laptop","description":"在笔记本电脑的活动文本框中说话，并直接在原处检查文字。","icon":"💻","accent":"from-sky-500 to-sky-600"},{"title":"对于电子邮件","href":"/use-cases/email","description":"通过讲话起草电子邮件。以思考的速度在 Gmail、Outlook 或任何电子邮件客户端中撰写消息。","icon":"📧","accent":"from-red-500 to-red-600"},{"title":"提高工作效率","href":"/use-cases/productivity","description":"在已打开的效率应用中，通过语音输入任务、笔记和回复。","icon":"⚡","accent":"from-emerald-500 to-emerald-600"},{"title":"无障碍使用","href":"/use-cases/accessibility","description":"为有 RSI、行动障碍或任何喜欢语音输入的用户提供免提打字。","icon":"♿","accent":"from-teal-500 to-teal-600"},{"title":"面向关节炎患者","href":"/use-cases/arthritis","description":"当关节炎让打字变得困难时，可以通过语音输入邮件、文档和消息，减少键盘操作。","icon":"🦴","accent":"from-purple-500 to-purple-600"},{"title":"面向 RSI 患者","href":"/use-cases/rsi","description":"将语音转换为当前应用中的可编辑文本，减少持续使用键盘。","icon":"🖐️","accent":"from-rose-500 to-rose-600"},{"title":"面向阅读障碍人士","href":"/use-cases/dyslexia","description":"面向阅读障碍人士的语音转文字工具。自然说话，即可将文字输入到您已经在使用的应用中。","icon":"🔤","accent":"from-violet-500 to-violet-600"},{"title":"面向听障人士","href":"/use-cases/hearing-impaired","description":"将录音语音转换为可阅读、搜索和再次查看的文本。可免费使用本地转写，或选择云端以获得更快速度。","icon":"👂","accent":"from-cyan-500 to-blue-600"},{"title":"离线使用","href":"/use-cases/offline","description":"无需互联网连接即可转录。完全在设备上运行 OpenAI Whisper 和 NVIDIA Parakeet 模型，以实现气隙工作流程。","icon":"✈️","accent":"from-amber-500 to-amber-600"},{"title":"对于内容创作者","href":"/use-cases/content-creators","description":"用语音口述标题、描述和字幕，无需打字，还能从已发布的视频中提取文字记录。","icon":"🎬","accent":"from-fuchsia-500 to-fuchsia-600"},{"title":"对于翻译人员","href":"/use-cases/translators","description":"以语音速度听写 99 多种语言的翻译。可在任何具有完全本地、私密处理的 CAT 工具中工作。","icon":"🌐","accent":"from-cyan-500 to-cyan-600"},{"title":"面向西班牙语兽医","href":"/use-cases/spanish-veterinary","description":"为兽医团队提供西班牙语听写。口述临床记录，检查转写文本，然后将其粘贴到您现有的 PIMS 中。","icon":"🐾","accent":"from-emerald-500 to-emerald-600"},{"title":"对于播客创作者","href":"/use-cases/podcasters","description":"在您自己的机器上免费、无限制地转录节目单集。将录音转换为文字稿，并将节目笔记语音输入到任何应用中。","icon":"🎙️","accent":"from-rose-500 to-rose-600"}],"oneTool":{"title":"一款工具，适配每种工作流","paragraph1":"OpenWhispr 是一款系统级听写工具——它能在任意应用的任意文本框中使用，覆盖 macOS、Windows 和 Linux。这种灵活性意味着它会主动适应你的工作流，而不是逼你去迁就它。","paragraph2":"开发者用它听写文档和 LLM 提示词，无需切换上下文。写作者以说话的速度起草整章内容。医疗专业人士借助完整的本地处理，让病历始终保持私密。任何受重复性劳损困扰的人，都能在不牺牲效率的前提下，让双手得到休息。","paragraph3":"无论你需要的是针对 Linux 或 Windows 等特定平台、某个特定职业的 AI 听写，还是离线转录这类专门需求，上述指南都会一步步告诉你 OpenWhispr 如何融入其中。"},"sections":{"workflow":{"eyebrow":"按工作流程探索","title":"为你的工作方式而设计"},"platform":{"eyebrow":"OpenWhispr 的更多使用方式","title":"选择平台或功能"}}},"useCaseAccessibility":{"breadcrumb":{"home":"Home","useCases":"Use Cases","accessibility":"For Accessibility"},"metadata":{"title":"Speech-to-Text Tools for Accessibility - OpenWhispr","description":"When typing is a barrier, use OpenWhispr to write hands-free in emails, documents, chats and forms. Choose Cloud for speed or local for offline work.","keywords":["speech-to-text tools for accessibility USA","accessibility speech-to-text","hands-free typing software","voice typing for RSI","dictation for motor disabilities","offline speech-to-text","local voice dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Accessibility speech-to-text software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source speech-to-text software for hands-free writing. Choose Cloud without a local model download, or local for device-only and offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text tools for accessibility","title":"When typing is a barrier, your voice keeps you writing.","body":"Write hands-free in the email, document, chat or form already open. Choose Cloud for speed, or local speech-to-text for device-only and offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"A man in a powered wheelchair speaks toward his laptop at a bright studio desk beside architectural drawings and a scale model"}},"trustBar":{"label":"Type by voice into:","names":["Gmail","Microsoft Word","Google Docs","Slack","Notion","Discord"]},"pain":{"label":"When typing takes too much","title":"The words are there. The keyboard is the barrier.","subtitle":"You need a way to get text into the app in front of you, without turning every correction into more keyboard work.","cards":[{"title":"Every correction asks more of your hands.","body":"When typing is painful or difficult, a transcript full of avoidable mistakes creates the same barrier twice. Use a custom dictionary for the names, acronyms and words you say often."},{"title":"A hold-to-talk key is not right for everyone.","body":"Choose toggle mode when holding a hotkey is uncomfortable. Start dictation with one press, speak, then stop when you are ready."},{"title":"Your tools should not require a new workflow.","body":"Put text into an email, document, chat or form where you can paste text. OpenWhispr uses the current writable cursor when automatic paste is available."},{"title":"Internet access is not always available.","body":"Choose a local model when you need device-only processing or an offline path. The transcription stays on your computer."}]},"features":{"label":"What changes","title":"Speak the sentence. Get the text where you need it.","subtitle":"Choose Cloud without a local model download, or a local model for device-only processing and offline work.","items":[{"image":{"label":"A woman speaks while sitting at a wooden table with an open laptop, in a sunlit room full of plants"},"title":"Cloud transcription without a model download","body":"Choose Cloud without downloading a local model. Audio is sent only to transcribe it, then immediately discarded."},{"image":{"label":"A man seated in a wheelchair speaks with his hands resting in his lap at a standing desk in an open brick-walled office"},"title":"Local transcription for offline and sensitive work","body":"Download a local model once and dictate without an internet connection. The audio stays on your device while it is transcribed."},{"image":{"label":"A woman speaks at a desktop monitor in a community office, one hand resting on a stack of printed papers"},"title":"Text in the app already open","body":"Press your hotkey, speak naturally, and paste the result into your document, message, email or web form."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Offline speech-to-text","value":"Local","footnote":"runs on your device without an internet connection"},{"label":"Cost for local speech-to-text","value":"$$0","footnote":"unlimited local AI models on every plan"},{"label":"Free Cloud allowance","value":"2,000","footnote":"words each week on the Free plan"}]},"howItWorks":{"label":"How it works","title":"Get the words out without returning to the keyboard.","subtitle":"Start free. Pick Cloud without a model download, or local for device-only and offline dictation.","image":{"label":"A woman speaks toward a raised laptop at a home desk, both hands resting on a padded wrist rest"},"steps":[{"title":"Download for free.","body":"Install OpenWhispr on Mac, Windows or Linux. Choose Cloud or download a local model."},{"title":"Dictate the thought.","body":"Use push-to-talk or toggle mode. Speak while the writable field you want to fill is active."},{"title":"Review and paste.","body":"Check the text, then put it into your email, document, message or form."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"The practical reasons","title":"10 reasons to choose OpenWhispr for accessibility","subtitle":"Compare modes, input controls, language support and plans before you choose a speech-to-text tool.","items":[{"title":"Toggle mode is available","body":"Start and stop dictation with separate presses when holding a hotkey is uncomfortable."},{"title":"Local models are unlimited and free","body":"Every plan includes unlimited local AI models, so local speech-to-text has no per-minute bill."},{"title":"It works offline","body":"Local transcription runs on your device without an internet connection."},{"title":"Cloud is a quick option","body":"Cloud works without a local model download, then immediately discards transcription audio."},{"title":"Dictate into the current app","body":"Put speech-to-text into an email, document, chat, form or any other place you can paste."},{"title":"Use your own words","body":"Add names, acronyms and recurring terms to the custom dictionary."},{"title":"100+ languages","body":"Speech-to-text supports more than 100 languages across every plan."},{"title":"Open source","body":"The MIT-licensed source is available to inspect and contribute to on GitHub."},{"title":"Choose the input rhythm","body":"Use push-to-talk for short bursts or toggle mode for longer dictation."},{"title":"Start without a subscription","body":"Free includes local models, a Cloud allowance and community support."}]},"faq":{"label":"Before you download","title":"Questions people ask before switching to voice input.","subtitle":"Straight answers on modes, cost and the day-to-day workflow.","items":[{"question":"Can I use this when typing is painful or difficult?","answer":"OpenWhispr turns spoken words into text you can paste into the app in front of you. It offers push-to-talk and toggle mode, so you can choose whether holding a hotkey works for you."},{"question":"Should I use Cloud or a local model?","answer":"Choose Cloud without a local model download. Choose a local model for device-only processing or offline work. Cloud audio is sent only to transcribe it, then immediately discarded."},{"question":"Will it work without internet access?","answer":"Yes. Local models run on your device and do not need an internet connection once downloaded."},{"question":"Can it handle names and specialised words?","answer":"Use the custom dictionary to add names, acronyms and recurring terms, then review the resulting text before sending or saving it."},{"question":"Does it only make a transcript?","answer":"Speech-to-text gives you text to review and paste where you need it. With your own API keys, you can also use AI writing assistance to reformat your words."},{"question":"What does Free include, and why upgrade?","answer":"Free includes unlimited local models, 2,000 Cloud words each week, five meeting-recording hours each month, 100+ languages and a custom dictionary. Pro adds unlimited Cloud transcription, more meeting time, sync, mobile and API/MCP access. Business adds unlimited meetings, Agent mode and chat over your data."}]},"reviews":{"label":"What users say","title":"The result people notice: less typing.","subtitle":"Three verbatim notes from people using OpenWhispr.","score":{"value":"$$0","label":"For unlimited local models","body":"Local speech-to-text is included on every plan. Choose paid plans for unlimited Cloud and additional workflow features."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Start for free.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for accessibility","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"People who want unlimited local speech-to-text and a practical Cloud allowance"}]},{"parts":[{"text":"People who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"People who need unlimited meetings, Agent mode, and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local models, plus limited Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meetings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your device. It works without an internet connection and is unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Enterprise includes team features and admin, SSO, SAML, SCIM, audit logs, retention controls and dedicated support."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Speech-to-text should fit the way you work.","paragraphs":["OpenWhispr is built for the moment when the keyboard is the hard part, not the thought. Speak naturally, review the text, and keep working in the app already open.","OpenWhispr is open source. You can inspect the code yourself. Local transcription stays on your device while it is transcribed.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Use your voice for the words that still need to be written.","body":"Download OpenWhispr, choose the mode that fits the moment, then use speech-to-text in the field you need to fill.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Unlimited local AI models are included on Mac, Windows and Linux."},"byline":{"authorLabel":"Written by","author":"Joshua Padoa","aiDisclosure":"Drafted with AI assistance","reviewerLabel":"Reviewed and fact checked by","reviewer":"Wesley van der Hoop","updatedLabel":"Last reviewed","dateModified":"2026-08-21"}}},"useCaseDevelopers":{"breadcrumb":{"home":"Home","useCases":"Use Cases","developers":"For Developers"},"metadata":{"title":"Free Local Dictation Software For Developers - OpenWhispr","description":"OpenWhispr is free, MIT-licensed dictation for developers. Talk the prompt through and it lands at your cursor, transcribed on your own machine.","keywords":["best dictation software for developers","dictation software for developers","speech to text for developers","voice dictation for programmers","open source dictation app","local speech to text","dictate prompts to AI coding tools"]},"schema":{"name":"OpenWhispr","applicationCategory":"Dictation software for developers","operatingSystem":"macOS, Windows, Linux","description":"Local-first, open-source dictation software for developers. Prompts, pull request descriptions and commit messages are transcribed on the developer's own machine. Free for unlimited local dictation.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation software for developers & engineers","title":"Talk the change through. Get the prompt typed.","body":"One hotkey, and the clean text lands where your cursor is: Cursor, a terminal, a GitHub review box. Pick a local model at setup and the audio never leaves your machine.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Person seated on the floor, reaching toward a laptop on a low table."}},"trustBar":{"label":"Dictate into any app, including:","names":["Cursor","VS Code","Claude Code","GitHub","Linear","Slack"]},"pain":{"label":"What you're putting up with","title":"So much of your typing isn't code.","subtitle":"Prompts, pull request descriptions, commit messages, review comments, standup updates. None of it is code. All of it is typing.","cards":[{"title":"The prompt takes longer than the run.","body":"An agent needs a paragraph of context to do the right thing, so you type the paragraph, and it finishes in under a minute. Say it instead and the paragraph lands in the composer."},{"title":"You already wired Whisper to a hotkey once.","body":"Then the OS update broke the paste, the venv rotted, or you never got the clipboard back. OpenWhispr is that script, finished and maintained."},{"title":"A privacy claim you can't check.","body":"Closed-source dictation asks you to believe a sentence on a marketing page. Ours is a repo you can read, and a local path you can test by pulling the cable out and carrying on."},{"title":"Built-in dictation has never heard of kubectl.","body":"macOS and Windows dictation guess at library names, CLI flags and service names. Put your stack in the custom dictionary to help the local Whisper model recognize those spellings, on the free plan."}]},"features":{"label":"What the hotkey actually does","title":"A paragraph of context. None of it typed.","subtitle":"Transcription runs on your machine. So can the cleanup. Neither costs anything.","items":[{"image":{"label":"Person speaking beside an open laptop at a shared desk."},"title":"Whisper runs on your machine, not ours","body":"Pick a local model at setup and Whisper or Parakeet transcribes the audio on your own machine, with no signal needed once the model is downloaded."},{"image":{"label":"Person gesturing while using a keyboard at a dimly lit desk."},"title":"One hotkey, any window","body":"The clean text pastes where your cursor already is: Cursor, VS Code, a terminal, a GitHub review box, Slack. Nothing to install per editor, and your clipboard is put back the way you left it."},{"image":{"label":"Person speaking at a workstation in an office at dusk."},"title":"The system prompt is yours to edit","body":"The prompt that cleans up your dictation ships as text you can open, rewrite, test against sample input and save."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Audio uploaded in local mode","value":"None","footnote":"the recording is transcribed on your own machine"},{"label":"Cost per developer","value":"$$0","footnote":"unlimited local dictation, every machine you own, forever"},{"label":"Setup takes","value":"A few steps","footnote":"download, pick a model, bind a hotkey, talk"}]},"howItWorks":{"label":"What the install involves","title":"Talking to your agent starts here.","subtitle":"No account, no card, and no telemetry unless you switch it on.","image":{"label":"Person speaking at a home desk with a keyboard, laptop and monitor."},"steps":[{"title":"Download it free.","body":"Mac, Windows or Linux. Setup offers OpenWhispr Cloud or local models. Pick local and it pulls them down once."},{"title":"Hit the hotkey and talk.","body":"Describe the change you want the way you would say it to a colleague at the next desk."},{"title":"Read it, then send it.","body":"The clean text is already at your cursor. Edit it, hit enter, and the recording is yours to keep or delete."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"What you can check yourself","title":"Ten reasons, and the repo is the citation for most of them.","subtitle":"Choosing dictation software as a developer comes down to what you'll take on trust. These are the parts you don't have to.","items":[{"title":"The audio is transcribed on your own disk","body":"A local Whisper or Parakeet model does the listening, on your machine. Drop the network and it keeps working, which is the shortest audit anyone has run on us."},{"title":"Free on every machine you own","body":"Unlimited local dictation at $0, forever, on the laptop, the desktop and the one in the office. There's no per-seat license to renew."},{"title":"MIT licensed, so fork it","body":"The desktop app is on GitHub under the MIT license. Read the transcription path before you install it, and change it if you want it to behave differently."},{"title":"It types into a terminal","body":"It pastes where your cursor is, including in a terminal. There's no editor plugin to install or maintain."},{"title":"Your identifiers, spelled your way","body":"Put library names, service names and the CLI tools you live in into the custom dictionary for sharper transcripts."},{"title":"The cleanup prompt is editable","body":"Tell it to leave code blocks alone, keep your shorthand, or stop it capitalising your service names. It's your prompt, and it ships as text."},{"title":"It runs with the wifi off","body":"A plane, a train, a customer's locked-down network. Local models need no signal once they're on disk."},{"title":"A CLI, and reading your notes is free","body":"npm install -g @openwhispr/cli, then openwhispr --local notes list reads them off your own machine with no login and no API key. Sync them to the cloud on a paid plan and Claude, ChatGPT and Cursor reach them over MCP."},{"title":"Telemetry is off unless you turn it on","body":"Analytics ship disabled out of the box. Recordings are files on your machine, deleted after 30 days by default, and Settings moves that or switches it off."},{"title":"100+ languages","body":"Standups, reviews and issue threads in the language your team actually speaks, transcribed on your own machine."}]},"faq":{"label":"Before you install it","title":"Six questions developers ask first.","subtitle":"Straight answers on what leaves the machine, what it types into, and what it won't do. If yours isn't here, ask us in Discord.","items":[{"question":"Does anything I dictate leave my machine?","answer":"Not the audio, once you pick a local transcription model at setup. Two more things worth knowing while you're in Settings. If you sign in, the AI cleanup that tidies your dictation runs through OpenWhispr Cloud until you point it at a local model too, so the text of what you said goes up even though the audio doesn't; signed out, it only leaves your machine if you've pointed cleanup at your own API key. And there's a cloud fallback for when a local model fails, which ships switched off and stays off until you turn it on."},{"question":"Does it work in a terminal, in Cursor and in VS Code?","answer":"Yes, and in anything else with a text field, because it pastes at your cursor rather than plugging into an editor. There's no extension to install per app. On a Mac it asks for the accessibility permission at setup, which is what lets it paste; without it the text goes to your clipboard and you paste it yourself. When it does paste for you, your previous clipboard contents are put back afterwards. On Linux, X11 works out of the box and a Wayland session needs ydotool installed before auto-paste works."},{"question":"Can I write code by voice with it?","answer":"Not the way Talon Voice does, and we'd rather say so. Talon is purpose-built for voice coding with language-specific commands and hands-free control of the whole machine. This is dictation in any text field, which in practice means the prose around the code: the prompt, the pull request description, the commit message, the review comment, the Slack reply."},{"question":"Does dictation software for developers get library names and CLI flags right?","answer":"That's what the custom dictionary is for, and it's free. Put your stack in it and the spellings are handed both to the local Whisper model doing the listening and to the AI model that tidies the transcript afterwards, so a service name doesn't get corrected back out after it was heard correctly. One exception worth knowing: the Parakeet model doesn't take the dictionary at the speech-to-text step, only the cleanup does."},{"question":"Why not just wire up whisper.cpp myself?","answer":"You can, and the parts that take the weekend aren't the transcription. They're the global hotkey on three platforms, the paste that works in a terminal and a browser, restoring the clipboard afterwards, model downloads and the cleanup pass that turns speech into something you'd send. All of that is written and MIT licensed on GitHub, so forking it is faster than starting."},{"question":"Is it actually free, and how does it compare to Wispr Flow?","answer":"Unlimited local dictation is $0 forever on Mac, Windows and Linux, and so is the custom dictionary. Our Pro is $8 a month, which makes OpenWhispr Cloud transcription unlimited and adds API and MCP access and the iPhone and iPad app."}]},"reviews":{"label":"What people tell us","title":"The people who found it before you did.","subtitle":"What people tell us.","score":{"value":"0","label":"Recordings kept on our servers","body":"With a local transcription model the audio never leaves your machine. Cloud audio is sent only to be transcribed, then never stored."},"items":[{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for developers","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Developers who want free, private, offline dictation and are fine with a local setup"}]},{"parts":[{"text":"Developers who want unlimited cloud transcription and MCP in their AI tools"}]},{"parts":[{"text":"Teams documenting and reviewing at volume"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local dictation, with the audio staying on your own machine"}]},{"parts":[{"text":"Unlimited cloud transcription plus API and MCP access, without managing local models"}]},{"parts":[{"text":"Unlimited recordings plus Agent mode and chat over your own notes"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"For individuals and teams"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with team features and admin, SSO, SAML and SCIM, audit logs, retention controls, and dedicated support."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"We wanted this for our own terminal first.","paragraphs":["OpenWhispr started as the thing we kept half-building for ourselves: a hotkey that turns a spoken paragraph into text, without shipping the paragraph to anybody.","We build in public. The desktop app is MIT licensed and on GitHub. When we say dictation can stay on your machine, you can check the code yourself.","And if it breaks the night before a release, you talk to the people who built it, in Discord."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"$$0 · open source · no extension to install","title":"Say the next prompt. Don't type it.","body":"Download OpenWhispr, pick a local model at setup, and the next paragraph you owe an agent gets spoken, cleaned and dropped at your cursor. No card, no seat to buy.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free forever on Mac, Windows & Linux."}}},"useCaseEmail":{"breadcrumb":{"home":"Home","useCases":"Use Cases","email":"For Email Writing"},"metadata":{"title":"AI Dictation Software for Email Writing - OpenWhispr","description":"OpenWhispr turns spoken thoughts into editable email drafts at your cursor. Review the wording, then send from the email client you already use.","keywords":["ai dictation software for email writing","voice to text for email","dictate emails","email dictation software","write emails by voice","speech to text for professional email"]},"schema":{"name":"OpenWhispr","applicationCategory":"AI dictation software for email writing","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source desktop dictation for writing professional email by voice at the active cursor, with local and Cloud transcription choices.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"AI dictation software for email writing","title":"Talk it through. Get an email draft to review.","body":"Dictate at the cursor in your email client. OpenWhispr turns your speech into editable text, then leaves the final send to you.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Man in a black sweater speaking and gesturing toward a laptop beside a window."}},"trustBar":{"label":"Dictate at your cursor in:","names":["Gmail","Outlook","Apple Mail","Thunderbird","Spark","Superhuman"]},"pain":{"label":"What slows down a professional email","title":"The reply is already clear. The keyboard is the delay.","subtitle":"Start with the point you need to make, keep the thread in view, and review the draft before anyone receives it.","cards":[{"title":"A blank reply still wants every keystroke.","body":"The answer is already in your head, but typing the greeting, context and ask turns a quick response into another task. Dictate the first pass at the email cursor."},{"title":"Tone changes with every recipient.","body":"A note to a teammate, a client and a manager cannot sound the same. Highlight rough wording and ask Voice Assistant to make it clearer, shorter or more formal before you send it."},{"title":"Names and acronyms create a correction pass.","body":"A clean paragraph is not useful if the client name, product code or project acronym is wrong. Add recurring terms to the custom dictionary so the transcript has the spellings you use."},{"title":"The email gets drafted in the wrong place.","body":"Writing in a separate voice app creates a copy, paste and context-switching loop. OpenWhispr puts text at the active writable cursor in the email client already open."}]},"features":{"label":"From thought to reviewed draft","title":"Speak the message. Keep control of the send.","subtitle":"Draft at the cursor, reshape selected wording by voice, and choose the transcription mode that fits the email.","items":[{"title":"Put the first draft at the email cursor","body":"Press the dictation hotkey, speak the subject or message, and let automatic paste place the text in the active writable field. Review it in the thread where it belongs.","image":{"label":"Woman with braided hair gesturing toward a laptop at a desk with books and plants behind her."}},{"title":"Edit selected wording with Voice Assistant","body":"Highlight a rough sentence, use the Voice Assistant hotkey, and ask for a shorter, clearer or more formal version. OpenWhispr safely replaces the selected text after the edit returns.","image":{"label":"Woman with short dark hair speaking and gesturing with both hands toward a laptop at a wooden table."}},{"title":"Choose Cloud or local before you dictate","body":"Use Cloud for top accuracy and speed without a model download. Choose a local model for device-only or offline email work.","image":{"label":"Man in a green shirt speaking toward a laptop at a table surrounded by plants."}}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Email audio sent to OpenWhispr","value":"None","footnote":"with a local model, transcription stays on your device"},{"label":"Cost to start writing by voice","value":"$$0","footnote":"unlimited local dictation on the Free plan"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first dictated email"}]},"howItWorks":{"label":"How it works","title":"From spoken thought to a reviewed email draft.","subtitle":"Install the desktop app, choose Cloud or local transcription, then dictate where the email will be sent.","image":{"label":"Woman in a blue shirt gesturing toward a desktop monitor at a wooden table with a keyboard and mug."},"steps":[{"title":"Download for free.","body":"Install on Mac, Windows or Linux. Choose Cloud for no model download or download a local model for offline use."},{"title":"Put the cursor in the email.","body":"Open a new message or reply, click the subject or body field, press your hotkey and speak."},{"title":"Review, then send it yourself.","body":"Check the recipient, facts, tone and wording. OpenWhispr provides editable text, not an autonomous send."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Why it earns a place in your email workflow","title":"10 reasons email writers choose OpenWhispr","subtitle":"Each reason answers a different question between the thought in your head and the message you are willing to send.","items":[{"title":"A spoken thought becomes editable email text","body":"Dictate the point while it is clear, and automatic paste puts the result at the active email cursor for review."},{"title":"Selected text can change without starting over","body":"Highlight the sentence and use Voice Assistant to shorten it, clarify it or change its tone in place."},{"title":"The email client stays on screen","body":"Gmail, Outlook, Apple Mail and other writable fields remain the workspace. The text goes to the cursor instead of a separate transcript window."},{"title":"The send stays yours","body":"OpenWhispr returns editable text. You check the recipient, facts and tone, then use the email client's own Send button."},{"title":"Local mode keeps sensitive dictation on the device","body":"Use a local speech-to-text model for confidential or offline email work. The recording does not leave your computer for transcription."},{"title":"Cloud avoids the model download","body":"Choose OpenWhispr Cloud for top accuracy and speed across hardware. The audio is discarded immediately after transcription."},{"title":"Names and acronyms get a standing hint","body":"Add client names, product terms and internal acronyms to the custom dictionary instead of correcting the same spelling in every message."},{"title":"One workflow covers 100+ languages","body":"OpenWhispr supports 100+ languages, so multilingual email drafts do not require a different dictation app."},{"title":"Free local dictation is not a trial","body":"Free includes unlimited local models plus 2,000 Cloud words each week. Pro adds unlimited Cloud transcription, sync, mobile, Agent mode and API/MCP access."},{"title":"Less first-pass typing, fewer repeated keystrokes","body":"Speak the long first pass, then keep the keyboard for names, facts and final edits. The workflow reduces composition keystrokes without pretending review disappears."}]},"faq":{"label":"Before you download","title":"Six questions before you dictate the next email.","subtitle":"Straight answers about email clients, review, terminology, processing modes and paid value.","items":[{"question":"Does it work in Gmail, Outlook, Apple Mail, Thunderbird, Spark and Superhuman?","answer":"OpenWhispr works system-wide instead of through a direct email integration. Put the cursor in a writable subject or message field, press the hotkey and dictate. With automatic paste enabled, the returned text goes to that active field."},{"question":"Will OpenWhispr send the email for me?","answer":"No. OpenWhispr returns editable text at your cursor. You remain responsible for checking the recipients, facts, tone and attachments, then pressing the email client's Send button."},{"question":"Does it produce an email draft or only a raw transcript?","answer":"Regular dictation converts your speech into editable text. Voice Assistant can take a spoken instruction or selected wording and produce a rewritten result, which you still review before sending."},{"question":"Can it handle client names, acronyms and other languages?","answer":"OpenWhispr supports 100+ languages and includes a custom dictionary for names, jargon and acronyms. Treat the dictionary as a strong spelling hint and check important details before sending."},{"question":"Should I use Cloud or a local model for email?","answer":"Choose Cloud for top accuracy and speed without downloading a model. Choose local for device-only or offline transcription. Cloud audio is sent only to transcribe it, then immediately discarded."},{"question":"What is free, and why would I upgrade?","answer":"Free includes unlimited local models, 2,000 Cloud words each week, 100+ languages and a custom dictionary. Pro adds unlimited Cloud transcription, Agent mode, sync, mobile, and API/MCP access. Business adds unlimited meeting recordings, chat over data, and priority support."}]},"reviews":{"label":"What users say","title":"The useful test is whether voice beats the keyboard.","subtitle":"The available testimonials are product-wide, not email-specific.","score":{"value":"3","label":"Desktop operating systems","body":"OpenWhispr runs on macOS, Windows and Linux."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for email writing","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"People who want unlimited local dictation and a 2,000-word weekly Cloud allowance"}]},{"parts":[{"text":"People who want unlimited Cloud transcription, sync, mobile, Agent mode, and API/MCP access"}]},{"parts":[{"text":"Teams that also need unlimited meeting recordings, chat over data, and priority support"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local email dictation at no cost, plus limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus Agent mode, mobile, sync, and API/MCP access"}]},{"parts":[{"text":"All Pro features plus unlimited meetings, chat over data, and priority support"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Email should sound like you before it reaches someone else.","paragraphs":["Professional email carries client details, internal decisions and unfinished thinking. OpenWhispr lets you dictate the draft at your cursor, then review it in the same email client.","We build in public. Every release and every line of code is on GitHub, where 5,000+ engineers have starred, forked and audited the project. When we say the dictation stays local, that's not a promise in a privacy policy. It's a property of the code, and anyone on your team can verify it.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"OpenWhispr team and community"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Talk through the next email. Review the draft. Then send.","body":"Download OpenWhispr, put the cursor in your email client and dictate the message that is already in your head.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseLinux":{"breadcrumb":{"home":"Home","useCases":"Use Cases","linux":"For Linux"},"metadata":{"title":"Linux Speech to Text Software for Voice Typing - OpenWhispr","description":"OpenWhispr is Linux speech to text software for voice typing, with local transcription or Cloud speed and session-aware setup guidance.","keywords":["linux speech to text software","linux voice typing","linux dictation software","offline speech to text linux","wayland speech to text","linux voice to text","open source speech to text linux"]},"schema":{"name":"OpenWhispr","applicationCategory":"Linux speech-to-text software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Linux speech-to-text software for voice typing. Choose Cloud for top accuracy and speed without a local model download, or local for offline, device-only transcription.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text software for Linux users","title":"Hold the hotkey. Get editable text.","body":"OpenWhispr is Linux speech to text software for producing editable text with your voice. Configure the delivery path for your session, then choose Cloud for speed or local for offline, device-only transcription.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Man speaking and gesturing at a desktop computer beside a window at dusk"}},"trustBar":{"label":"Set up for:","names":["Wayland","X11","GNOME","KDE","wtype","Clipboard"]},"pain":{"label":"What gets in the way of Linux voice typing","title":"The words are ready. Your desktop session is the question.","subtitle":"A useful Linux setup separates recognition from text delivery, then makes the limits visible before you rely on it.","cards":[{"title":"Recognition and delivery are different jobs.","body":"Getting words from speech is only half the workflow. The text still has to reach the focused field where you are writing. OpenWhispr treats that as a setup concern, not a promise hidden in the headline."},{"title":"Wayland is not just X11 with a new name.","body":"The app reads Wayland session signals and can see whether an Xwayland display is available. Within Wayland, it identifies GNOME, KDE and wlroots desktops. A setup that needs a helper should say so before you dictate a long note."},{"title":"A missing helper is a broken workflow.","body":"On Linux, text delivery can depend on wtype, ydotool or xdotool. OpenWhispr can surface installation guidance for the tools a session needs instead of leaving you to infer it from a silent failure."},{"title":"A voice workflow needs a real test.","body":"A short dictation is not enough. Test a real note in the focused field you use, then address any input permission or helper guidance before you make voice typing part of the day."}]},"features":{"label":"What Linux voice typing needs","title":"One speaking workflow. Clear choices at each layer.","subtitle":"Choose where transcription runs, configure the Linux session you have, then review the text in the work you were already doing.","items":[{"image":{"label":"Woman holding a small black device near her mouth at a desktop computer beside a bright window"},"title":"Choose Cloud when you want speed without a model download","body":"OpenWhispr Cloud is the top-accuracy, high-speed route with no local model download. Audio sent to Cloud is transcribed, then immediately discarded."},{"image":{"label":"Man seated with a laptop at a table beside a train window"},"title":"Choose local when offline, device-only work matters","body":"Every plan includes unlimited local AI models. Local transcription runs entirely on your device with no internet required, so you can keep working when a connection is not part of the setup."},{"image":{"label":"Man with glasses gesturing toward a desktop monitor at a wooden desk"},"title":"See Linux setup dependencies before they surprise you","body":"The Linux key listener reports a missing binary or input permission. The app can also flag when a required paste tool is unavailable, so you can test the real delivery path before depending on it."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Cloud transcription audio stored","value":"None","footnote":"audio sent to OpenWhispr Cloud is transcribed, then immediately discarded"},{"label":"Cost to use local models","value":"$$0","footnote":"unlimited local AI models are included on every plan"},{"label":"Free Cloud allowance","value":"2,000","footnote":"words of OpenWhispr Cloud transcription each week"}]},"howItWorks":{"label":"How it works","title":"Test the path from hotkey to editable text.","subtitle":"Start free, configure the Linux setup your session needs, and review the text before you make it part of real work.","image":{"label":"Woman speaking and gesturing at a desktop computer with a hand on the keyboard"},"steps":[{"title":"Download for Linux.","body":"Start with the desktop app. Local models are included, and Free also includes 2,000 OpenWhispr Cloud words each week."},{"title":"Set the hotkey and choose a transcription path.","body":"Choose Cloud for speed without a local download, or local for device-only transcription without internet. Then test the hotkey in your desktop session."},{"title":"Review the delivery path in the app you use.","body":"Confirm the text reaches the focused field as your session allows. If a paste helper or permission is missing, address that before you depend on the workflow."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons Linux users choose OpenWhispr","subtitle":"Linux dictation is not one checkbox. These are the decisions that determine whether voice typing is useful on the machine you actually run.","items":[{"title":"Local transcription is included","body":"Every plan includes unlimited local AI models. Choose that path when offline, device-only transcription is the requirement."},{"title":"Cloud avoids a local model download","body":"Cloud is the top-accuracy, high-speed option when you want a transcription path without first downloading a local model."},{"title":"Cloud transcription audio is not stored","body":"Audio sent to OpenWhispr Cloud is used to transcribe, then immediately discarded. That is separate from your choice to sync other data."},{"title":"Your desktop session is part of the setup","body":"OpenWhispr reads Wayland session signals and can see whether an Xwayland display is available. Within Wayland, it identifies GNOME, KDE and wlroots desktops as a concrete setup starting point."},{"title":"Missing input permission is surfaced","body":"The Linux key listener reports when it cannot access input devices. You can fix a known setup issue instead of guessing why a hotkey did not respond."},{"title":"Paste-tool guidance is specific","body":"When Linux text delivery needs help, OpenWhispr can identify a missing wtype, ydotool or xdotool path and show install commands for common distributions."},{"title":"A key listener is built for Linux","body":"The Linux key manager listens for key down and key up events, and includes a watchdog so a missed key release does not leave a held-key recording state stuck."},{"title":"Inspect the Linux path before you adopt it","body":"OpenWhispr is MIT-licensed. Inspect the session and text-delivery code before you choose it, rather than treating open source as a badge."},{"title":"Names and jargon have a place to go","body":"Every plan includes a custom dictionary for names, jargon and acronyms, so you can tune the terms that matter in your own voice workflow."},{"title":"Test before you reorganize your work","body":"Free local models let you test a real task in your app, desktop session and input environment before deciding whether more Cloud transcription is worth it."}]},"faq":{"label":"Before you configure it","title":"The questions Linux users ask before they rely on it.","subtitle":"Straight answers on delivery, desktop sessions, local models, Cloud transcription and what a paid plan changes.","items":[{"question":"Does OpenWhispr support Linux?","answer":"Yes. Linux is listed as an available desktop platform on every plan. Free is desktop-only, while paid plans also add the iPhone and iPad app."},{"question":"Will text always land at the cursor on Linux?","answer":"Text delivery depends on the desktop session and its available input tools. OpenWhispr checks Linux session signals and can show guidance when a supported paste helper is missing. Test the focused-field path in the apps you use before depending on it."},{"question":"What is different about Wayland and X11?","answer":"They are different Linux display-server environments, and they can need different text-delivery setup. OpenWhispr reads Wayland session signals and can see whether an Xwayland display is available. Within Wayland, it identifies GNOME, KDE and wlroots desktops so guidance can start from the session you actually have."},{"question":"Should I choose Cloud transcription or a local model?","answer":"Choose Cloud for top accuracy and speed without a local model download. Choose a local model for offline, device-only transcription. Cloud audio is sent only to transcribe it, then immediately discarded."},{"question":"Can I use it without internet?","answer":"Yes, when you choose a local model. Local AI models run entirely on your device with no internet required and are unlimited on every plan."},{"question":"What does a paid plan change for Linux voice typing?","answer":"Free includes unlimited local models, 2,000 OpenWhispr Cloud words each week and five hours of meeting recordings each month. Pro adds unlimited Cloud transcription, Agent mode, 20 meeting-recording hours, sync, personal API access, MCP integration and the iPhone and iPad app. Business adds unlimited meeting recordings, chat over your data and priority support."}]},"reviews":{"label":"What users say","title":"For people who would rather speak than type.","subtitle":"Three user quotes, kept as supplied. The Linux setup still deserves a real test in your own session.","score":{"value":"$$0","label":"Cost to test the local path","body":"Every plan includes unlimited local AI models. Test your session's hotkey and delivery path before you pay for more Cloud transcription."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for Linux voice typing","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Linux users who want local models plus a limited Cloud allowance"}]},{"parts":[{"text":"Linux users who want unlimited Cloud transcription, sync, mobile and API/MCP access"}]},{"parts":[{"text":"Teams that need unlimited meeting recordings, chat over data and priority support"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local models and 2,000 OpenWhispr Cloud words each week"}]},{"parts":[{"text":"Unlimited Cloud transcription plus sync, mobile and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings, chat over data and priority support"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device. It works without an internet connection and is unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Linux voice typing deserves an honest setup path.","paragraphs":["OpenWhispr started with a simple objection: needing to type less should not force you to give up the desktop apps and documents you already rely on.","We build in public. The app is MIT-licensed, so you can inspect the code behind local models, Cloud transcription and Linux setup choices before you make one.","Free includes community support. Pro adds email support, and Business adds priority support."],"image":{"label":"OpenWhispr team and community"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Let your voice carry more of the typing on Linux.","body":"Download OpenWhispr, configure your Linux session, and test whether editable text reaches the focused field where you actually write.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free on macOS, Windows and Linux. Paid plans add unlimited Cloud transcription, the iPhone and iPad app, sync and workflow tools."},"byline":{"authorLabel":"Written by","author":"Joshua Padoa","aiDisclosure":"Drafted with AI assistance","reviewerLabel":"Reviewed and fact checked by","reviewer":"Wesley van der Hoop","updatedLabel":"Last reviewed","dateModified":"2026-09-10"}}},"useCaseOffline":{"metadata":{"title":"离线听写 — 本地语音转文字 | OpenWhispr","description":"采用本地 Whisper 模型的离线听写软件。无需网络，数据不会发送到服务器，适合在飞机、偏远地区和敏感环境中使用。","keywords":["离线语音转文字","离线听写应用","无需联网转录","本地语音识别","隔离网络转录","离线 Whisper","nvidia parakeet","openai whisper"]},"breadcrumb":{"home":"首页","useCases":"使用场景","offline":"离线使用"},"hero":{"badge":"离线语音转文字","title":"无需联网即可转录","subtitle":"100% 离线语音转文字，由在本地设备上运行的 OpenAI Whisper 和 NVIDIA Parakeet 驱动。无需联网，数据永远不离开你的设备。","viewOnGitHub":"在 GitHub 上查看"},"whyOffline":{"title":"离线为何重要","paragraph1":"大多数语音转文字工具会把你的音频发送到云端服务器处理。这意味着它们在飞机上毫无用处，在火车上时灵时不灵，在网络不稳定的地区彻底失效。更糟的是，你的语音数据——包括你听写的一切内容——都要经过第三方服务器。","paragraph2":"OpenWhispr 采用了根本不同的方式。OpenAI Whisper 和 NVIDIA Parakeet 模型只需下载一次，便完全在你的电脑上运行。完成初始设置后，应用再也无需联系任何服务器。无论你连着光纤还是开着飞行模式，转录效果都一模一样。"},"howItWorks":{"title":"离线转录如何运作","download":{"title":"模型只需下载一次","body":"可从多种尺寸的 OpenAI Whisper 和 NVIDIA Parakeet 模型中选择。通过 WiFi 下载一次，模型便永久驻留在你的电脑上。模型越大越精准，模型越小越快速。"},"local":{"title":"通过 Whisper 与 Parakeet 本地处理","body":"转录使用 whisper.cpp 和 NVIDIA Parakeet 在你的 CPU/GPU 上运行。无服务器调用，无 API 密钥，无需联网。只靠你自己的硬件完成工作。"},"zeroData":{"title":"零数据传输","body":"你的音频在内存中处理，绝不写入磁盘，也绝不传输到任何地方。转录出的文字直接进入剪贴板。没有遥测，不会对你的语音数据做任何分析。"},"platforms":{"title":"全平台可用","body":"macOS、Windows 和 Linux。每个操作系统上都是同样的离线体验。带上你的笔记本电脑去任何地方，无需担心网络也能听写。"}},"scenarios":{"title":"离线听写大显身手的场景","flight":{"title":"在飞行途中","body":"六个小时不受打扰的飞行时间最适合写作。当基于云端的听写工具弹出“无连接”错误时，OpenWhispr 的表现与在地面上别无二致。起草邮件、撰写报告或捕捉灵感——全部用语音完成，全部离线。"},"remote":{"title":"偏远地区","body":"野外研究人员、记者和远程工作者常常身处手机信号和 WiFi 都无法覆盖的地方。OpenWhispr 毫不在意。在偏远的野外站点记录笔记，在小木屋里转录访谈，或者在露营时口述日记。只要你的笔记本电脑还有电，你就能听写。"},"privacy":{"title":"隐私敏感环境","body":"律师、医疗专业人员以及任何处理敏感信息的人，都需要值得信赖的听写工具。有了 OpenWhispr，你的语音数据在物理上根本无法接触互联网——它在本地处理，绝不存储。无需担心云端处理的 HIPAA 合规问题，也没有第三方服务器数据泄露的风险。"}},"whyChoose":{"title":"为何选择 OpenWhispr 进行离线使用","trulyOffline":{"title":"真正离线","body":"不是“离线但偶尔需要同步”——而是下载模型后真正零联网依赖。可在飞行模式下永久使用。"},"modelSizes":{"title":"多种模型尺寸","body":"从 tiny（快速、精度较低）到 large（较慢、精度最高）。为你的硬件和需求选择合适的取舍。下载多个并随意切换。"},"openSource":{"title":"免费且开源","bodyPrefix":"离线转录无需任何订阅费用。","link":"完整源代码","bodySuffix":"已在 GitHub 上公开。为图方便可选择 $8/月的 Pro。"},"languages":"99+ 种语言——全部离线可用","pushToTalk":"按住说话与切换模式"},"internalLinks":{"lead":"了解更多 OpenWhispr：","windows":"最佳 Windows 听写工具","sep1":"、","email":"邮件听写","sep2":"，以及","accessibility":"无障碍","sep3":"。或者","compare":"对比 OpenWhispr","suffix":"与其他听写工具。"},"cta":{"title":"用 OpenWhispr 走向离线","description":"下载应用，获取一个 Whisper 或 Parakeet 模型，便大功告成。从此以后，你再也不需要联网就能转录。","footer":"永久免费 · 100% 离线 · 无需账户"}},"useCaseTranslators":{"metadata":{"title":"翻译人员听写软件 | OpenWhispr","description":"使用本地 Whisper 听写 99 种以上语言的译文。私密、离线，适用于任何 CAT 工具。免费试用 OpenWhispr。","keywords":["面向译员的语音听写","翻译听写软件","翻译语音转文字","多语言听写应用","译员效率工具","口译员语音输入","语言专业人士听写","多语言语音转文字","nvidia parakeet","openai whisper"]},"breadcrumb":{"home":"首页","useCases":"使用场景","forTranslators":"面向译员"},"hero":{"badge":"面向译员与语言专业人士","title":"以语速口述翻译","subtitle":"别再用两种语言打字了。OpenWhispr 让你用 99+ 种语言自然地说出译文——配合 AI 清理，将你的输出整理为专业交付格式。私密、本地，并能在任何 CAT 工具中使用。","viewOnGitHub":"在 GitHub 上查看"},"problem":{"title":"译员为何需要语音听写","paragraph1":"翻译是耗费脑力的工作。你用一种语言阅读，揣摩含义与微妙之处，再用另一种语言产出文字。你最不需要的就是输出环节的阻力——切换键盘布局、四处寻找特殊字符、对抗那个根本读不懂目标语言的自动更正。","paragraph2":"大多数听写工具只为一种语言而生。它们在带口音的语音、专业术语以及译员每天身处的多语言混合工作流面前束手无策。OpenAI Whisper 和 NVIDIA Parakeet 改变了这一切——它们在海量多语言数据集上训练而成，能以接近母语者的精度处理 99+ 种语言。","callout":"OpenWhispr 为你提供由 Whisper 和 Parakeet 驱动的全系统语音听写。自然地说出你的译文，用智能体模式清理并整理输出，然后直接粘贴到你使用的任何工具中——Trados、memoQ、一个简单的文本编辑器，任何都行。"},"solutions":{"title":"OpenWhispr 如何帮助译员更快工作","languages":{"title":"99+ 种语言，由 Whisper 与 Parakeet 支持","body":"用任何源语言或目标语言听写，精度接近母语者。OpenAI Whisper 和 NVIDIA Parakeet 在海量多语言数据集上训练而成——它们能处理通用听写工具会漏掉的口音、地方方言和专业词汇。无需切换键盘布局。"},"agentMode":{"title":"面向翻译工作流的智能体模式","body":"设置针对你工作量身定制的自定义系统提示词：“清理并整理为专业译文”“整理为双语草稿”“去除填充词并润色以供发布”。输入你的原始口述，输出即可用于交付的格式化译文。"},"catTool":{"title":"适用于任何 CAT 工具","body":"Trados、memoQ、OmegaT、Memsource、Smartcat、Wordfast——或一个纯文本编辑器。OpenWhispr 通过剪贴板集成在全系统范围内运行。只要你能往里面粘贴文字，OpenWhispr 就能配合使用。无需插件，没有兼容性问题。"},"privacy":{"title":"私密的本地处理","body":"客户文档始终留在你的电脑上。借助本地 Whisper 和 Parakeet 处理，你的音频绝不离开你的电脑——无云端上传，无第三方访问。非常适合受保密协议约束的项目、机密法律文本和敏感的客户工作。"}},"scenarios":{"title":"真实翻译场景","literary":{"title":"文学与图书翻译","body":"你正在翻译一部小说，需要在目标语言中产出自然流畅的文笔。你不再一句一句地打字——在阅读与打字之间不停切换思维档位——而是读完一段，然后大声说出你的译文。智能体模式会把你的口述清理成精炼的文笔，在保留文学语感的同时去除停顿与说错重来的部分。","note":"常见译员工作流：自然地口述目标语言草稿，然后做一遍术语与文风校对，完成最终质检。"},"technical":{"title":"技术与法律文件翻译","body":"技术手册、法律合同、医学报告——这些文档要求术语精准、格式一致。自然地口述专业术语（Whisper 和 Parakeet 能很好地处理特定领域的词汇），然后用智能体模式配合诸如“整理为正式法律译文，保持段落结构”的提示词，确保输出始终一致。所有处理都在本地完成，因此机密文档绝不离开你的电脑。"},"subtitling":{"title":"字幕与本地化","body":"字幕翻译需要简短有力、迅速交付的文字。看视频，口述你译好的字幕行，再把它们粘贴到字幕编辑器里。当你需要产出数百条字幕片段时，说话比打字快得多。智能体模式可以帮助让字幕行保持精炼，并控制在字符数限制之内。"}},"whyChoose":{"title":"译员为何选择 OpenWhispr","accuracy":{"title":"多语言精度","body":"99+ 种语言，精度接近母语者。Whisper 和 Parakeet 专为多语言语音打造——它们能处理口音、方言以及语言之间的语码转换。"},"confidentiality":{"title":"客户机密","body":"本地处理意味着客户文档绝不离开你的电脑。无云端上传，无保密协议顾虑，无需担心数据隐私法规。"},"speed":{"title":"输出快 3 倍","body":"以每分钟 130 个词的速度说话，而不是每分钟打 40 个词。对于按字计费或赶截止日期的译员来说，这样的速度差异是革命性的。"},"linkAllUseCases":"所有使用场景","linkCompare":"对比替代方案"},"bottomCta":{"title":"以语速翻译","description":"别再一个字符一个字符地打出译文。用任何语言自然地说出来，让 AI 清理输出，然后直接粘贴到你的工具里。开源，且从设计上保护隐私。"}},"useCaseWindows":{"breadcrumb":{"home":"Home","useCases":"Use Cases","windows":"For Windows"},"metadata":{"title":"Free Dictation App for Windows - OpenWhispr","description":"OpenWhispr is a Windows dictation app that pastes editable text at your cursor. Use fast Cloud transcription or unlimited local models offline.","keywords":["dictation app for windows","windows dictation app","voice typing app for windows","speech to text windows","offline dictation windows","windows voice to text"]},"schema":{"name":"OpenWhispr","applicationCategory":"Windows dictation application","operatingSystem":"macOS, Windows, Linux, iOS","description":"A Windows dictation app that turns speech into editable text at your cursor, with fast Cloud transcription and unlimited local models for offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation app for Windows","title":"Speak at the cursor. Get editable text back.","body":"Press your OpenWhispr shortcut in a document, email or message. Choose fast Cloud transcription or a local model, then review the text pasted back where you started.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Man gesturing beside an open laptop at a table by an airport window"}},"trustBar":{"label":"Paste your text into:","names":["Microsoft Word","Microsoft Outlook","Microsoft Teams","Slack","Google Docs","OneNote"]},"pain":{"label":"What gets in the way on Windows","title":"Your thought is ready. The text box is still empty.","subtitle":"Built-in tools solve parts of the job. A dedicated dictation app should start fast, return editable text, and let you choose how speech is processed.","cards":[{"title":"The connection decides when you can dictate.","body":"Windows voice typing needs internet. Choose a local OpenWhispr model when you need dictation without a connection, and keep working in the same document or message."},{"title":"The dictation control moves from app to app.","body":"Word, Outlook and browser tools each start dictation differently. OpenWhispr uses one shortcut and a Windows paste path to return text to the field where you started."},{"title":"Misheard names become a second typing job.","body":"Add names, jargon and acronyms to the custom dictionary. The goal is editable text you can review, not a transcript that creates more correction work."},{"title":"The paste should not erase what you copied.","body":"OpenWhispr saves the clipboard before it pastes your transcript. If you have not copied something new, it restores the previous content afterward."}]},"features":{"label":"What you get","title":"One shortcut. Editable text in the Windows app already open.","subtitle":"Speak from the field you are using. OpenWhispr transcribes the words, pastes the text back, and leaves you with a draft you can edit before sending.","items":[{"image":{"label":"Woman gesturing toward a desktop monitor at a desk with an open notebook"},"title":"Text returns to the window where you started","body":"The Windows helper can restore the window captured at recording start before it pastes. Your document, reply or note stays the place where the work happens."},{"image":{"label":"Woman in a white coat looking at an open notebook beside a desktop computer"},"title":"Paste the transcript, then restore the clipboard","body":"OpenWhispr temporarily copies the transcript for the Windows paste. When the clipboard is still unchanged after the paste, your previous text, image or formatted content comes back."},{"image":{"label":"Man gesturing toward a desktop monitor beside a rain-covered window"},"title":"Choose Cloud or local on the same PC","body":"Use Cloud for top accuracy and speed without a model download. Use an unlimited local model for offline or device-only dictation. Neither choice changes where the editable text returns."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Speech sent to us in local mode","value":"None","footnote":"local transcription runs on your Windows PC"},{"label":"Cost to start dictating","value":"$$0","footnote":"unlimited local dictation is included on Free"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first dictated text"}]},"howItWorks":{"label":"How it works","title":"From spoken thought to editable Windows text in three steps.","subtitle":"Download the Windows app, choose fast Cloud transcription or an offline local model, then dictate from the field where the text belongs.","image":{"label":"Man gesturing toward a monitor with one hand on a keyboard at a desk by a window"},"steps":[{"title":"Download for Windows.","body":"Install OpenWhispr. Start on Free without entering card details."},{"title":"Put the cursor where the text belongs.","body":"Open the document, email, chat or note, press your OpenWhispr shortcut, and speak."},{"title":"Review the text in place.","body":"OpenWhispr pastes the transcript back into the Windows field so you can edit, send or keep writing."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons Windows users choose OpenWhispr","subtitle":"A dictation app has to fit the PC, the writing field and the way you want speech processed. Here is what OpenWhispr adds to that workflow.","items":[{"title":"One shortcut for the writing day","body":"Start from the document, email, message or note already open. You do not need a different dictation button for each destination."},{"title":"Install it for everyone, or just for you","body":"The Windows installer adds desktop and Start menu shortcuts. On a work PC where you cannot install for all users, choose \"Only for me\"."},{"title":"Your previous clipboard can come back","body":"After pasting, OpenWhispr restores the clipboard content it replaced when you have not copied something new in the meantime."},{"title":"Local dictation works without internet","body":"Download a local model once, then dictate on the PC when the network is unavailable or device-only processing is the right fit."},{"title":"Cloud avoids a local model download","body":"Choose Cloud when you want top accuracy, speed and consistent performance without asking the Windows PC to run the speech model."},{"title":"Names and jargon get their own dictionary","body":"Add the people, products and acronyms you type every day so review starts with the terms that matter to your work."},{"title":"Dictation supports 100+ languages","body":"Use the same Windows app when your work moves between languages instead of changing to a separate dictation tool."},{"title":"Free is useful before you upgrade","body":"Free includes unlimited local models, 2,000 Cloud words a week and five meeting-recording hours a month."},{"title":"Pro expands the workflow beyond one PC","body":"Pro adds unlimited Cloud transcription, Agent mode, device sync, mobile access and API or MCP access."},{"title":"The Windows behavior is open source","body":"Inspect how the app captures shortcuts, pastes text and packages its Windows helpers. The implementation is public on GitHub."}]},"faq":{"label":"Before you download","title":"Questions Windows users ask before switching.","subtitle":"Straight answers about the cursor workflow, internet, privacy, corrections and plans.","items":[{"question":"How is OpenWhispr different from Windows voice typing?","answer":"Windows voice typing is built into Windows and uses online speech recognition. OpenWhispr is a separate desktop app with a shared shortcut and paste workflow, plus a choice between Cloud transcription and local models that work offline."},{"question":"Does it integrate with Word, Outlook, Teams or Slack?","answer":"OpenWhispr does not need a native integration to return text. It copies the transcript and uses the Windows paste path, so choose the text field first and review what lands there before you send it."},{"question":"Do I need internet to dictate?","answer":"Not with a local model. Local transcription runs on the PC and is unlimited on every plan. Cloud needs a connection and is the no-download choice for top accuracy and speed."},{"question":"What happens to my audio?","answer":"With a local model, transcription stays on your device. With OpenWhispr Cloud, audio is sent only for transcription and discarded immediately afterward. Choose the mode that fits the text you are writing."},{"question":"Can it learn names and technical terms?","answer":"Yes. The custom dictionary is included on every plan, so you can add names, jargon and acronyms that recur in your Windows writing."},{"question":"Is the Windows dictation app free?","answer":"Yes. Free includes unlimited local models, 2,000 Cloud words a week and five hours of meeting recordings a month. Pro adds unlimited Cloud transcription, Agent mode, sync, mobile and API or MCP access."}]},"reviews":{"label":"What users say","title":"Built for people who would rather speak than retype.","subtitle":"Real comments from people using OpenWhispr for local models, AI chats and offline work.","score":{"value":"1","label":"Shortcut across your Windows writing","body":"Start from the text field already open, dictate once, and review the editable result there."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for Windows users","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Windows users starting with unlimited local dictation and a limited Cloud allowance"}]},{"parts":[{"text":"Windows users who want unlimited Cloud transcription, Agent mode, sync, mobile and API or MCP access"}]},{"parts":[{"text":"Teams that need unlimited meeting recordings, chat over data and priority support"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Useful Free access with unlimited local dictation and 2,000 Cloud words a week"}]},{"parts":[{"text":"Unlimited Cloud transcription plus Agent mode and cross-device workflow features"}]},{"parts":[{"text":"Unlimited meeting recordings plus chat over data and priority support"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"A Windows dictation app you can inspect.","paragraphs":["OpenWhispr starts with the field already under your cursor. Speak, receive editable text, and keep the document or message as the place where the work happens.","The source code and release history are public on GitHub. You can inspect the Windows shortcut listener, paste path and packaged helpers, including the code that keeps local dictation on the PC.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"OpenWhispr team and community"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Speak the next sentence. Review it at the cursor.","body":"Download OpenWhispr, open the field where the words belong, and speak. Review the editable text there before you send it.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseWriters":{"breadcrumb":{"home":"Home","useCases":"Use Cases","writers":"For Writers"},"metadata":{"title":"Free Speech To Text Software For Writers - OpenWhispr","description":"OpenWhispr is free, open-source speech to text software for writers. It writes down what you said instead of drafting for you, on your own machine.","keywords":["speech to text software for writers","dictation software for writers","voice to text for authors","speech to text for novelists","dictation app for bloggers","voice typing for writing"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech to text software for writers","operatingSystem":"macOS, Windows, Linux, iOS","description":"Local-first, open-source speech to text software for writers. Your dictation is written down in your own words rather than redrafted for you, and in local mode it is transcribed on your own machine. Free on macOS, Windows and Linux.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech to text software for writers & authors","title":"Talk the draft out. Get your words, not an AI's.","body":"OpenWhispr writes down what you actually said, ums trimmed and nothing invented. Take the local route at setup and the manuscript never leaves your machine.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Person speaking at a desk with printed pages and a laptop"}},"trustBar":{"label":"Write into:","names":["Scrivener","Ulysses","iA Writer","Obsidian","Google Docs","Microsoft Word"]},"pain":{"label":"What you're choosing between","title":"You said the line. It came back as somebody else's.","subtitle":"You want the draft down at the speed you talk. You do not want it back sounding like everyone else's.","cards":[{"title":"The draft comes back in nobody's voice.","body":"OpenWhispr's cleanup mode is instructed to keep your wording and return the cleaned transcript. It does not compose an unsaid sentence. When you dictate a draft, the page begins with what you said."},{"title":"Dictate a line of dialogue, get an answer back.","body":"Talk to an assistant and it answers you. Say \"Where were you last night?\" into OpenWhispr and that is what lands on the page: your detective's question, not a reply to it. The cleanup model is told the speaker is never talking to it, so a question in your draft stays a question in your draft."},{"title":"You cannot see what got changed.","body":"A polished transcript is hard to trust when you cannot compare it with the words you dictated. OpenWhispr keeps both. Expand any dictation in your history and the raw transcript is sitting there word for word, with its own copy button. When cleanup changed nothing, the entry says so."},{"title":"An unfinished manuscript in someone else's account.","body":"Choose local transcription when you want your manuscript handled on your own machine. It needs no upload."}]},"features":{"label":"What lands on the page","title":"Faster drafts. Still your prose.","subtitle":"Local dictation is unlimited on every plan, runs on your own machine, and gives back the sentences you said.","items":[{"image":{"label":"Person speaking beside a laptop and printed pages"},"title":"It types what you said, at your cursor","body":"Hold the hotkey, talk, and the words land in whatever you are writing in. Scrivener, Word, Obsidian, a browser text box: it pastes straight at the cursor, so there is no plugin to install and no separate window to copy out of."},{"image":{"label":"Person using a laptop beside an open notebook"},"title":"The raw transcript, kept beside the clean one","body":"Open the history to compare the finished text with the raw transcript from a dictation. Expand the entry and copy the raw words. Prefer no cleanup? One toggle in Settings turns it off and the transcript is what lands at your cursor."},{"image":{"label":"Person speaking at a laptop beside an open book"},"title":"Your character names, spelled your way","body":"Put the names, places and invented words from your book in the custom dictionary. Local Whisper and cloud transcription use those spellings as hints. Type Aerdwyn once and those routes lean on it instead of guessing phonetically."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Your recordings on our servers","value":"None","footnote":"audio is never stored, and on the local route it never leaves your device"},{"label":"Cost per 100,000 words","value":"$$0","footnote":"local dictation is unlimited on every plan, including Free"},{"label":"Setup","value":"Your pace","footnote":"download, pick a model, choose a hotkey"}]},"howItWorks":{"label":"Download to first paragraph","title":"Talk it through once. Then edit, not rewrite.","subtitle":"No account to open, no plugin to install, nothing to upload. Choose a local model at setup and the whole thing runs on your own machine, offline.","image":{"label":"Person seated at a desk with a closed laptop and printed pages"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Skip the account if you want to. Take the local route at setup and it puts both the transcribing and the cleanup on your own machine."},{"title":"Hold the hotkey and talk.","body":"Open Scrivener, Word, Obsidian or anything else that takes text, hold your hotkey, and say the paragraph. It lands at the cursor when you let go."},{"title":"Read it back, then write.","body":"If a line reads wrong, open the history and check it against the raw transcript."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"What it will and will not do","title":"Ten reasons writers keep their own words.","subtitle":"Picking speech to text software is a decision about your voice before it is anything else. The cleanup controls are in Settings, and the prompt that governs cleanup ships with the app.","items":[{"title":"It transcribes, it does not draft.","body":"The cleanup prompt is instructed to keep your voice, wording, formality and intent, and to keep technical terms and proper nouns exactly as spoken. It removes fillers, fixes grammar and punctuation, and breaks up run-ons. It does not write you a new sentence."},{"title":"The AI only composes when you ask it to.","body":"Composing is a separate mode. It runs when you address the assistant out loud by whatever name you gave it, or when you hold the separate assistant hotkey. Say nothing to it and nothing is generated. Its own instructions forbid adding content that was not spoken or requested."},{"title":"Questions stay questions.","body":"Assistants answer you. This one is told the speaker is never talking to it, so a rhetorical question in an essay or a line of dialogue in a novel comes out on the page as you said it."},{"title":"Deliberate bad grammar survives, if you want it to.","body":"Cleanup fixes grammar and breaks up run-on sentences, which is wrong for a stream-of-consciousness passage or a character who talks in fragments. Turn cleanup off for that stretch and the raw transcript is what you get. Either way both versions are saved."},{"title":"You can turn cleanup off.","body":"One toggle in Settings, labeled Enable text cleanup. Off means no cleanup model is in the path and the transcript is what lands at your cursor."},{"title":"Your invented vocabulary goes to the transcriber.","body":"The custom dictionary is passed to the model as the spellings to use, on a local Whisper model and on cloud transcription alike, so character names, places and coinages come back closer to how you wrote them."},{"title":"Unlimited local dictation, on every plan.","body":"Local transcription is unlimited and free on Free, Pro and Business alike. A novel is not a metered event. The word allowance in the pricing table applies to cloud transcription only."},{"title":"It works offline.","body":"Take the local route at setup and there is no network call to make. A retreat with one bar of signal, or seat 24B on a red-eye. The draft still comes out."},{"title":"100+ languages, and you can switch mid-draft.","body":"Dictate in the language you think in and switch between languages without changing a setting. Useful if your dialogue is not in the language your narration is."},{"title":"It writes into whatever you already use.","body":"Dictation pastes at the cursor rather than into a proprietary editor, so Scrivener, Ulysses, iA Writer, Obsidian, Google Docs and Word all work with no plugin to install."}]},"faq":{"label":"Before you install it","title":"The questions writers ask first.","subtitle":"Straight answers on voice, privacy and cost.","items":[{"question":"Will it rewrite my prose?","answer":"It cleans, it does not compose. Cleanup is instructed to keep your voice, wording, formality and intent, and to keep proper nouns and jargon exactly as spoken. What it does change is fillers, false starts, stutters and repetitions, and it fixes grammar and punctuation and breaks up run-on sentences. If your prose runs long on purpose, turn cleanup off in Settings and you get the raw transcript. Either way both versions are saved."},{"question":"How do I check what it changed?","answer":"Open the history in the control panel and expand any dictation. Underneath the finished text you get the raw transcript, word for word, with a copy button. If cleanup made no change at all, the entry tells you so."},{"question":"Will it get my characters' and places' names right?","answer":"Add them to the custom dictionary in Settings. Those spellings are passed to the transcriber as the ones to use, on a local Whisper model and on cloud transcription alike. It is a strong hint rather than a hard rule, and the local Parakeet model does not take a dictionary at all."},{"question":"Is my unpublished manuscript private?","answer":"Choose local transcription at setup to process audio on your own machine without upload. Cloud transcription sends audio to be transcribed, then immediately discards it. Cloud backup can sync your transcriptions and notes across devices when enabled. Team-space and shared notes can sync separately."},{"question":"Is it really free for a whole book?","answer":"Yes. Local dictation is unlimited on the Free plan and every plan above it, with no word cap and no card. The 2,000 words a week you see in the pricing table is the Free tier's cloud transcription allowance, which is a different thing."},{"question":"Does it work in Scrivener, Ulysses or Word?","answer":"It works anywhere you can type. Dictation pastes at the cursor rather than living in its own editor, so there is nothing to install into your writing app and nothing to copy out afterwards. Mac, Windows and Linux, in 100+ languages."}]},"reviews":{"label":"What people tell us","title":"Local models. No WiFi. Better than typing.","subtitle":"From our Discord, GitHub and inboxes. They write code, not chapters, so read them for the tool rather than the trade.","score":{"value":"Raw transcript","label":"Available to check","body":"Cleanup is instructed to keep your wording and return only the cleaned transcript, with no commentary or answers. When a raw transcript is available, open it in history and copy it to check."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for writers","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Writers drafting a book who want unlimited, offline dictation for free"}]},{"parts":[{"text":"Writers who also want cloud transcription without managing a model"}]},{"parts":[{"text":"Editorial teams and agencies who want unlimited recordings and priority support"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local dictation at no cost. Upgrade for unlimited Cloud transcription and workflow features."}]},{"parts":[{"text":"Unlimited cloud transcription when you would rather not run a model"}]},{"parts":[{"text":"Everything in Pro plus unlimited meeting recordings, agent mode and priority support"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"More: unlimited meeting recordings, Agent mode and chat over your data"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Contact us about Enterprise."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"The line you got right out loud, first time. We built for that.","paragraphs":["OpenWhispr started with a simple objection: getting your own sentences onto the page faster should not mean handing them to something that rewrites them into its own voice.","The cleanup prompt ships with the app. You can read exactly how it is told to keep your wording.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Nothing on the page that you did not say.","body":"Download OpenWhispr, take the local route at setup, and your next chapter gets talked out on your own machine, in your own sentences.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free to download on Mac, Windows and Linux."}}},"security":{"metadata":{"title":"本地语音转文字的隐私与安全 | OpenWhispr","description":"OpenWhispr 保护您的语音隐私：在设备上本地处理，不存储音频，不使用您的数据训练 AI，并提供完整的用户控制。"},"breadcrumb":{"home":"首页","security":"安全性"},"hero":{"title":"OpenWhispr 的安全性","subtitle":"您的语音数据十分私密。这里清楚说明我们如何处理这些数据、存储哪些内容，以及您拥有哪些控制权。"},"howItWorks":{"title":"OpenWhispr 的工作原理","intro":"OpenWhispr 是一款适用于 macOS、Windows 和 Linux 的桌面应用，可将语音转换为文字。您可以自行决定音频的处理方式——共有三种模式，选择最适合您需求的一种即可。","local":{"title":"本地模型——一切都不离开您的设备","body":"您可以直接在本机运行开源的 Whisper 模型。选择此选项后，您的音频将完全在设备本地处理，绝不会传输到任何地方。如果您重视让数据远离互联网，那么这就是适合您的选项。"},"byok":{"title":"使用您自己的 API 密钥","body":"您可以连接来自 OpenAI 等提供商的自有 API 密钥。您的音频会通过您自己的账户直接发送给该提供商进行转录。您掌控与提供商之间的关系，并适用其隐私政策。"},"cloud":{"title":"OpenWhispr Cloud","body":"我们的托管云端转录服务会在我们这端完成处理。音频会发送到我们的 API，转录后再返回。我们和我们的提供商都不会存储您的音频——它会被实时处理后即时丢弃。我们在后台使用第三方 AI 提供商，并可能在它们之间切换，但我们始终确保没有任何提供商会用您的数据进行训练。"},"openSource":{"label":"开源桌面应用：","body":"OpenWhispr 桌面应用完全开源，已在 GitHub 上提供。您可以确切查看音频是如何被采集、处理和传输的。云端 API 服务则是独立的，由我们托管。"}},"dataStorage":{"title":"数据处理与存储","audio":{"title":"音频录音","body":"没有人会存储您的音频。使用本地模型时，音频绝不离开您的设备。使用您自己的 API 密钥时，音频会直接发送至您选择的提供商，且不会被保留。使用 OpenWhispr Cloud 时，音频会被实时处理后即时丢弃——我们和我们的提供商都不会保留它。"},"transcriptions":{"title":"转录文本","body":"转录文本存储在您设备的本地。如果您选择启用云端同步，它们也会存储在我们的数据库中，以便您在多台设备上访问。您可以随时关闭同步。"},"account":{"title":"账户数据","body":"如果您创建账户，我们会存储您的电子邮箱、姓名和身份验证凭据。密码经过哈希处理，绝不以明文形式存储。支付处理完全由 Stripe 负责。"},"diagnostic":{"title":"诊断数据","body":"诊断报告默认关闭。您必须在设置中明确启用它。启用后，我们会收集匿名的性能指标以改进产品。其中绝不会包含任何音频或转录内容。"}},"training":{"title":"AI 训练政策","intro":"我们对 AI 训练立场坚定：您的数据不会被用于训练任何 AI 模型。绝无例外。","cloud":{"label":"OpenWhispr Cloud：","body":"我们云服务背后使用的每一家提供商都已退出使用您的数据进行模型训练。我们可能会出于质量和速度考量在提供商之间切换，但「不训练」的承诺始终适用。"},"byok":{"label":"使用您自己的密钥：","body":"如果您使用自己的 API 密钥，您的提供商的训练政策将取决于您在其平台上的账户设置。大多数主流提供商都提供了退出训练的选项——请查阅您的提供商文档。"},"self":{"label":"OpenWhispr 本身：","body":"我们不会使用您的转录文本、音频或任何用户生成的内容来训练模型。"},"local":{"label":"本地模型：","body":"完全无需担心训练问题。您的音频始终留在您的设备上，绝不会发送到任何地方。"}},"infrastructure":{"title":"基础设施安全","intro":"对于选择启用云端功能的用户，我们遵循业界标准的安全实践：","encryption":{"title":"加密","body":"所有传输中的数据均使用 TLS 加密。我们云端数据库中静态存储的数据也已加密。与第三方提供商的 API 连接均使用安全、经过身份验证的端点。"},"authentication":{"title":"身份验证","body":"我们支持电子邮箱/密码和 Google OAuth 登录。会话通过安全、短时效的令牌进行管理，并会自动刷新。凭据绝不会与第三方共享。"},"dependencies":{"title":"依赖项管理","body":"我们使用自动化工具监测依赖项中的已知漏洞，并及时应用补丁。我们的开源代码库意味着社区可以审查我们的供应链。"}},"controls":{"title":"您的控制权","intro":"由您决定 OpenWhispr 如何处理您的数据。您在设置过程中选择处理模式，并可随时在设置中更改。","mode":{"title":"选择您的处理模式","body":"可选择本地模型、您自己的 API 密钥或 OpenWhispr Cloud。如果您希望音频绝不离开您的设备，请选择本地模型。"},"sync":{"title":"启用或关闭云端同步","body":"可让转录文本仅保留在本地，或在多台设备间同步。由您决定。"},"diagnostic":{"title":"选择启用诊断报告","body":"诊断数据收集默认关闭。如果您愿意帮助我们改进 OpenWhispr，可以将其打开。"},"delete":{"title":"删除您的数据","body":"您可以在应用内删除单条转录、清除所有本地数据，或彻底删除您的账户。"}},"providers":{"title":"云端提供商","cloud":{"title":"OpenWhispr Cloud","body":"使用 OpenWhispr Cloud 时，您的音频会通过我们的托管 API 处理。我们会在后台使用第三方 AI 提供商，并可能为提高质量或速度而更换提供商；我们会按照《数据处理附录》提前通知。所有提供商均已选择不使用您的数据进行训练。"},"byok":{"title":"使用您自己的密钥","body":"使用您自己的 API 密钥时，您的音频会直接发送至您选择的提供商。其隐私政策和训练设置将依据您在其平台上的账户而定。"},"tableIntro":"我们在整个平台上使用的服务：","headerProvider":"提供商","headerPurpose":"用途","headerOptOut":"退出训练","rows":[{"provider":"OpenAI","purpose":"云端转录和 AI（通过 OpenWhispr Cloud 或 BYOK）","optOut":"我们的账户已选择退出"},{"provider":"Groq","purpose":"云端 AI 推理和转录故障转移（通过 OpenWhispr Cloud 或 BYOK）","optOut":"我们的账户已选择退出"},{"provider":"Parasail","purpose":"代理功能的云端 AI 推理","optOut":"零数据保留"},{"provider":"Mem0","purpose":"代理功能的 AI 记忆","optOut":"不会使用您的数据进行训练"},{"provider":"Exa","purpose":"代理功能的网页搜索","optOut":"不会使用您的数据进行训练"},{"provider":"Neon","purpose":"数据库和身份验证","optOut":"不适用"},{"provider":"Vercel","purpose":"网站托管和分析","optOut":"不适用"},{"provider":"Stripe","purpose":"支付处理","optOut":"不适用"},{"provider":"Resend","purpose":"事务性电子邮件","optOut":"不适用"},{"provider":"Google","purpose":"OAuth 登录","optOut":"不适用"},{"provider":"Slack","purpose":"内部支持沟通","optOut":"不适用"}],"footnote":"根据我们的《数据处理附录》，在新增或替换处理客户数据的提供商之前，我们至少提前 30 天通知。当前列表发布在 trust.openwhispr.com，本页面也会更新以反映变更。"},"vulnerability":{"title":"漏洞报告","intro":"如果您在 OpenWhispr 中发现安全漏洞，我们希望能够了解。作为一个开源项目，我们重视来自社区的负责任披露。","howToReport":"如何报告","emailPrefix":"请发送邮件至","emailSuffix":"并附上问题描述。","steps":"如有可能，请附上复现步骤。","ack":"我们将在 48 小时内确认收到您的报告，并随时向您通报修复进展。","disclosure":"在我们获得合理的修复时间之前，请勿公开披露漏洞。"},"contact":{"title":"有疑问？","intro":"如果您对我们的安全实践有疑问，或想进一步了解我们如何处理您的数据，请联系我们：","teamName":"OpenWhispr 团队","securityLabel":"安全：","generalLabel":"一般咨询：","privacyLink":"隐私政策","termsLink":"服务条款","sourceLink":"源代码"}},"models":{"metadata":{"title":"Whisper 模型规格、Parakeet 等 | OpenWhispr","description":"比较 Whisper 从 Tiny 到 Large-v3 的模型规格，以及 Orukeet、NVIDIA Parakeet、Groq 和 OpenAI。可在本地私密运行，或使用自己的 API 密钥。","keywords":["whisper 模型","nvidia parakeet","语音转文字模型","whisper tiny base small medium large turbo","groq 转录","本地 AI 模型","语音转文字提供商","openai whisper","parakeet tdt","whisper.cpp","sherpa-onnx","orukeet","oruk"]},"breadcrumb":{"home":"首页","models":"模型"},"badges":{"recommended":"推荐","fastest":"最快"},"hero":{"titlePrefix":"模型与","titleHighlight":"提供商","subtitle":"OpenWhispr 支持多种语音转文字引擎和 AI 提供商。为了隐私可全部本地运行，为了速度可使用云端 API，也可两者混用。"},"localTranscription":{"title":"本地转录模型","subtitle":"完全在您的本机上运行语音转文字。无需联网，数据不离开您的设备。","whisper":{"meta":"通过 whisper.cpp · 99+ 种语言 · GGML 格式","headerModel":"模型","headerSize":"大小","headerSpeed":"速度","headerQuality":"质量","headerBestFor":"最适合","models":[{"id":"tiny","name":"Tiny","size":"75 MB","speed":"最快","speedLabel":"最快","quality":"基础","description":"轻量级模型，适合快速草稿和低配硬件"},{"id":"base","name":"Base","size":"142 MB","speed":"快速","speedLabel":"快速","quality":"良好","recommended":true,"description":"对大多数用户而言速度与准确度的最佳平衡"},{"id":"small","name":"Small","size":"466 MB","speed":"中等","speedLabel":"中等","quality":"更佳","description":"更高的准确度，适合专业转录"},{"id":"medium","name":"Medium","size":"1.5 GB","speed":"较慢","speedLabel":"较慢","quality":"高","description":"高准确度，适合多语言和技术性内容"},{"id":"large-v3","name":"Large v3","size":"3 GB","speed":"最慢","speedLabel":"最慢","quality":"最佳","description":"在全部 99+ 种语言中实现最高准确度"},{"id":"turbo","name":"Turbo","size":"1.6 GB","speed":"快速","speedLabel":"快速","quality":"良好","description":"针对速度优化的 Large-v3 架构——以 Base 级延迟实现接近 Large 的准确度"}]},"parakeet":{"meta":"通过 sherpa-onnx · 25 种语言 · ONNX INT8","sotaBadge":"顶尖英语表现","languagesList":"Parakeet 支持的语言：英语、保加利亚语、克罗地亚语、捷克语、丹麦语、荷兰语、爱沙尼亚语、芬兰语、法语、德语、希腊语、匈牙利语、意大利语、拉脱维亚语、立陶宛语、马耳他语、波兰语、葡萄牙语、罗马尼亚语、俄语、斯洛伐克语、斯洛文尼亚语、西班牙语、瑞典语、乌克兰语。","models":[{"id":"parakeet-tdt-0.6b-v3","name":"Parakeet TDT 0.6B v3","size":"680 MB","speed":"快速","speedLabel":"快速","quality":"出色","languages":25,"description":"顶尖的英语准确度，并支持 25 种语言的多语言识别。采用 INT8 量化，实现高效的本地推理。"}]},"orukeet":{"meta":"由 Oruk 出品 · 通过 sherpa-onnx · 25 种语言 · ONNX INT8","defaultBadge":"全新安装的默认模型","setupGuide":"Orukeet 设置指南","languagesList":"Orukeet 支持的语言：英语、保加利亚语、克罗地亚语、捷克语、丹麦语、荷兰语、爱沙尼亚语、芬兰语、法语、德语、希腊语、匈牙利语、意大利语、拉脱维亚语、立陶宛语、马耳他语、波兰语、葡萄牙语、罗马尼亚语、俄语、斯洛伐克语、斯洛文尼亚语、西班牙语、瑞典语、乌克兰语。","models":[{"id":"orukeet-v0.1.0","name":"Orukeet v0.1.0","size":"672 MB","speed":"快速","speedLabel":"快速","quality":"出色","recommended":true,"languages":25,"description":"由 Oruk 基于 Parakeet 微调的多语言模型。与 Parakeet 共用同一 sherpa-onnx 运行时，无需额外安装，也是全新安装时默认启用的模型。"}]},"languagesCount":"{count} 种语言","fullyLocal":"100% 本地"},"cloudTranscription":{"title":"云端转录提供商","subtitle":"使用您自己的 API 密钥实现云端驱动的转录。自带密钥让一切尽在您的掌控。","getApiKey":"获取 API 密钥","custom":{"title":"自定义端点","label":"任何兼容 OpenAI 的 API","descPrefix":"连接 Ollama、自托管的 Whisper 服务器、LocalAI，或任何提供兼容 OpenAI","descSuffix":"端点的服务。"},"providers":[{"name":"OpenAI","keyUrl":"https://platform.openai.com/api-keys","models":[{"id":"gpt-4o-mini-transcribe","name":"GPT-4o Mini Transcribe","speed":"极快","speedLabel":"极快","description":"快速、准确的转录"},{"id":"gpt-4o-transcribe","name":"GPT-4o Transcribe","speed":"标准","speedLabel":"标准","description":"OpenAI 最准确的转录"},{"id":"whisper-1","name":"Whisper","speed":"标准","speedLabel":"标准","description":"原始的 Whisper API 端点"}]},{"name":"Groq","recommended":true,"keyUrl":"https://console.groq.com/keys","models":[{"id":"whisper-large-v3-turbo","name":"Whisper Large v3 Turbo","speed":"超快","speedLabel":"超快","description":"216 倍实时速度——目前最快的云端转录"}]},{"name":"Mistral","keyUrl":"https://console.mistral.ai/api-keys","models":[{"id":"voxtral-mini-latest","name":"Voxtral Mini","speed":"快速","speedLabel":"快速","description":"来自 Mistral AI 的多语言转录"}]}]},"openWhisprCloud":{"title":"OpenWhispr Cloud","subtitle":"零配置的转录和 AI 处理。无需 API 密钥——登录即用。","point1":{"title":"托管转录","body":"我们自动路由至最快的提供商"},"point2":{"title":"内置 AI 文本清理","body":"格式整理、标点修正及去除口头语"},"point3":{"title":"提供免费方案","body":"升级至 Pro 即可无限使用"}},"aiReasoning":{"title":"AI 文本处理","subtitle":"转录完成后，OpenWhispr 可以清理您的文本——修正语法、去除口头语并整理输出格式。可从云端或本地 AI 模型中选择。","cloud":{"title":"云端提供商（BYOK）","providers":[{"name":"OpenAI","models":["GPT-5.2","GPT-5 Mini","GPT-5 Nano","GPT-4.1","GPT-4.1 Mini","GPT-4.1 Nano"]},{"name":"Anthropic","models":["Claude Opus 4.5","Claude Sonnet 4.5","Claude Haiku 4.5"]},{"name":"Google","models":["Gemini 3 Pro","Gemini 3 Flash","Gemini 2.5 Flash Lite"]},{"name":"Groq","models":["Qwen3 32B","GPT-OSS 120B","GPT-OSS 20B","LLaMA 3.3 70B","LLaMA 3.1 8B","Mixtral 8x7B"]}]},"local":{"title":"本地模型（通过 llama.cpp）","footnote":"本地推理模型通过 llama.cpp 运行，并采用 GGUF 量化。所有处理都留在您的本机上。","models":[{"provider":"Qwen","recommended":"Qwen3 8B","others":"0.6B、1.7B、4B、8B、32B 各版本"},{"provider":"Mistral","recommended":"Mistral 7B Instruct","others":"Q4 和 Q5 量化版本"},{"provider":"Meta LLaMA","recommended":"LLaMA 3.2 3B","others":"1B、3B、8B 各版本"},{"provider":"OpenAI OSS","recommended":"GPT-OSS 20B","others":"开源旗舰"}]}},"howItWorks":{"title":"各部分如何协同工作","step1":{"title":"1. 您开口说话","body":"按住热键，对着任意应用自然地说话"},"step2":{"title":"2. 模型转录","body":"Orukeet、Whisper、Parakeet 或某个云端提供商将语音转换为文字"},"step3":{"title":"3. AI 清理","body":"可选的 AI 处理会修正语法、去除口头语并整理文本格式"}},"cta":{"title":"免费试用每一个模型","body":"下载 OpenWhispr，选择最适合您的模型。本地模型完全免费。云端提供商则使用您自己的 API 密钥。","githubButton":"GitHub"}},"download":{"metadata":{"title":"下载 OpenWhispr — 免费语音输入软件","description":"免费下载 OpenWhispr：在 macOS、Windows 和 Linux 本地运行的开源语音输入软件。私密、离线，无需云端。","keywords":["下载 OpenWhispr","OpenWhispr macOS","OpenWhispr Windows","OpenWhispr Linux","语音听写下载","免费语音转文字应用"]},"breadcrumb":{"home":"首页","download":"下载"},"referral":{"title":"首月免费！","welcome":"欢迎使用 OpenWhispr！您的推荐优惠已应用到您的账户。","nextStepsTitle":"接下来：","step1":"点击下方按钮下载 OpenWhispr","step2":"安装并登录您的账户","step3":"升级到 Pro——结账时将自动应用您的优惠"},"latestLabel":"最新版","hero":{"title":"下载 OpenWhispr","subtitle":"支持 macOS、Windows、Linux 和 iOS。选择与您设备匹配的版本。"},"autoDownloadNotice":"您的下载即将自动开始……","loading":"加载中……","recommendedLabel":"为您的设备推荐：","allDownloads":{"title":"全部下载","subtitle":"选择与您系统匹配的安装包。","releaseNotes":"查看更新说明","installGuide":"安装指南","uninstall":"卸载说明","systemRequirements":"系统要求","downloadHelp":"下载帮助"},"platforms":{"macos":{"label":"macOS","description":"适用于 Apple Silicon 和 Intel 的 DMG 版本","options":[{"label":"苹果电脑 (Apple Silicon)","meta":"DMG"},{"label":"苹果机（Intel）","meta":"DMG"}]},"windows":{"label":"Windows","description":"适用于 Windows 10 及更高版本的安装程序","options":[{"label":"Windows（安装程序）","meta":"EXE"}]},"linux":{"label":"Linux","description":"AppImage、Debian、RPM 和 tarball","options":[{"label":"Linux AppImage (x86_64)","meta":"AppImage"},{"label":"Linux DEB (amd64)","meta":"Debian 软件包"},{"label":"Linux RPM (x86_64)","meta":"RPM封装"},{"label":"Linux tar.gz (x64)","meta":"便携式压缩包"}]},"ios":{"label":"iPhone 和 iPad","description":"App Store 上的通用应用","optionLabel":"iOS 版 OpenWhispr","optionMeta":"App Store · 需要 iOS 17"}}},"contact":{"metadata":{"title":"联系我们 | OpenWhispr","description":"联系 OpenWhispr 团队。无论是技术支持、合作洽谈、媒体咨询，还是一般问题，欢迎与我们取得联系。","keywords":["OpenWhispr 联系方式","联系 OpenWhispr","OpenWhispr 技术支持","OpenWhispr 销售","OpenWhispr 帮助"]},"breadcrumb":{"home":"首页","contact":"联系我们"},"hero":{"title":"联系我们","subtitle":"我们期待您的来信。欢迎通过以下任意渠道与我们联系。"},"cards":{"team":{"title":"想为您的团队引入 OpenWhispr？","description":"在整个组织中部署语音转文字，享受批量定价与优先支持。","linkText":"联系我们的销售团队"},"talk":{"title":"与团队对话","description":"有问题、反馈，或只是想打个招呼？随时给我们留言。"},"help":{"title":"需要帮助？","description":"漏洞反馈、功能建议或技术支持——我们的社区随时为您服务。","linkText":"在 GitHub 上提交 issue"}}},"contactSales":{"metadata":{"title":"企业 AI 会议笔记与语音输入 | OpenWhispr","description":"联系销售团队，了解企业定价、SSO，以及私密本地 AI 会议笔记和语音输入软件的定制方案。"},"hero":{"title":"联系我们的销售团队","subtitle":"为您的组织量身定制方案,提供专属支持、SSO 和团队管理功能。"},"success":{"title":"我们已收到您的咨询","body":"我们的团队将在 24 小时内与您联系。","backHome":"返回首页"},"form":{"name":{"label":"姓名","placeholder":"您的全名"},"email":{"label":"邮箱","placeholder":"work@company.com"},"company":{"label":"公司","placeholder":"公司名称"},"companySize":{"label":"公司规模","placeholder":"选择规模"},"companySizeOptions":{"1-10":"1-10 名员工","11-50":"11-50 名员工","51-200":"51-200 名员工","201-1000":"201-1,000 名员工","1000+":"1,000 名以上员工"},"platform":{"label":"邮箱和日历平台","placeholder":"您的团队使用什么?"},"platformOptions":{"google":"Google(Gmail / Google Calendar)","microsoft":"Microsoft(Outlook / Teams)","other":"其他"},"message":{"label":"还有什么我们可以帮您解答的吗?","placeholder":"请告诉我们您的使用场景、需求或任何疑问……"},"submit":"发送咨询","submitting":"发送中……","errors":{"nameRequired":"请填写姓名","emailInvalid":"请输入有效的邮箱地址","companyRequired":"请填写公司名称","companySizeRequired":"请选择您的公司规模","platformRequired":"请选择您的平台","alreadySubmitted":"您最近已经提交过了。我们会尽快与您联系!","generic":"出了点问题,请重试。","network":"网络错误。请检查您的网络连接后重试。"}},"features":{"title":"企业版功能","managedApiCredits":"托管 API 额度","ssoCompliance":"SSO 与合规","teamFeatures":"团队功能与管理后台","cloudSync":"跨设备云同步"},"responseTime":{"value":"< 24 小时","label":"平均响应时间"},"emailDirect":{"prefix":"更喜欢用邮件?直接联系我们:"}},"legalPrivacy":{"metadata":{"title":"OpenWhispr 隐私政策 | 本地或云端，不出售数据","description":"在设备上或在云端使用 OpenWhispr，由您选择。开源，且我们绝不出售您的数据。了解我们收集哪些信息、由谁处理，以及您享有的权利。"},"breadcrumb":{"home":"首页","privacy":"隐私政策"},"hero":{"title":"隐私政策","subtitle":"了解 OpenWhispr 如何处理您的数据。","lastUpdated":"最后更新：2026年10月1日"},"overview":{"title":"概述","paragraph1":"OpenWhispr 由 Gizmo Labs Inc.（特拉华州公司，地址：221 W 9th Street, Wilmington, Delaware 19801, USA）发布并运营，该公司对本政策所述的个人信息负责（即其“控制者”）。OpenWhispr 是一款以隐私为核心打造的桌面与移动端听写与会议笔记应用。您可以选择语音的处理方式：在您的设备上处理，或通过 OpenWhispr Cloud 处理；在桌面应用中，还可以通过您使用自己的 API 密钥连接的服务商处理。您可以随时在设置中更改，本政策将说明每种选择意味着什么。","paragraph2Prefix":"本政策准确说明了 OpenWhispr 收集哪些数据、数据存放在哪里、还有谁能看到这些数据，以及您如何控制或删除它们。如需以通俗语言了解我们如何处理您的数据以及我们绝不会做什么，请查看我们的","securityLink":"安全页面","paragraph2Suffix":"。"},"dataGlance":{"title":"数据一览","intro":"我们收集哪些数据、数据流向何处的速览摘要。完整说明见下方。","headerCategory":"类别","headerWhere":"存放位置","headerWho":"还有谁能看到","headerOptional":"是否可选?","rows":[{"category":"账户与账单","where":"我们的云端数据库","who":"Neon、Stripe","optional":"登录和购买方案所必需"},{"category":"语音与音频","where":"您的设备（桌面应用默认保留听写音频 30 天）","who":"OpenWhispr Cloud 提供商，或您自己的提供商（BYOK）；本地模式下无人","optional":"由您选择模式"},{"category":"转录内容","where":"设备本地 SQLite；如已备份则存于我们的云端数据库（桌面应用）","who":"Neon（如已备份）；用于清理的 OpenWhispr Cloud AI 提供商，或您自己的提供商（BYOK）","optional":"桌面应用的云端备份默认关闭"},{"category":"笔记与智能体对话","where":"设备本地 SQLite；如已备份、共享或位于团队空间，则存于我们的云端数据库","who":"我们云端的笔记：Neon 和 OpenAI（搜索索引）；Mem0（智能体记忆）；Exa（智能体网络搜索）；OpenWhispr Cloud AI 次级处理者，或您自己的提供商（BYOK）","optional":"可选；付费方案的移动端备份默认开启"},{"category":"日历","where":"仅限本设备","who":"Google 或 Microsoft（您连接的日历）；如您向智能体询问事件，则还包括您的 AI 服务商","optional":"可选连接"},{"category":"会议检测信号","where":"仅存于设备内存","who":"无人","optional":"可禁用"},{"category":"服务日志（云端请求）","where":"我们的云端数据库","who":"Neon","optional":"对 OpenWhispr Cloud 请求而言不可选；随您的账户一并删除"},{"category":"移动应用分析、归因与购买","where":"供应商服务器（移动应用）","who":"AppsFlyer、RevenueCat、Superwall、Sentry","optional":"分析与归因：可通过“Usage Analytics”（使用情况分析）开关关闭；在 iOS 上，仅当您允许跟踪时才会进行归因。购买处理为应用内购买所必需"},{"category":"网站分析与广告","where":"您的浏览器（Cookie）及供应商的服务器","who":"Google（Analytics 4、Ads）、Meta；Vercel（无 Cookie 统计）","optional":"选择加入（欧洲经济区、英国、瑞士），其他地区为选择退出（广告遵循 GPC）"},{"category":"B2B 访客识别","where":"您的浏览器（脚本）","who":"RB2B","optional":"仅限美国访客；默认启用（可选择退出，并遵循 GPC）"},{"category":"推荐计划","where":"我们的云端数据库","who":"Neon；Resend（邀请邮件）","optional":"可选"},{"category":"Insights 与排行榜","where":"您的设备；如您同步或加入，则存于我们的云端数据库","who":"排行榜：您工作区的成员，或使用相同工作邮箱域名的人","optional":"同步和排行榜在您开启前均为关闭状态"},{"category":"支持与销售消息","where":"我们的邮件和支持工具","who":"Google Workspace、Linear、Slack","optional":"仅在您联系我们时"},{"category":"营销邮件","where":"我们的邮件服务商","who":"Resend","optional":"一键退订"}]},"collect":{"title":"我们收集的信息","account":{"title":"账户与账单","intro":"当您创建 OpenWhispr 账户时，我们仅收集管理账户所需的最少信息：","emailLabel":"邮箱/密码注册：","emailBody":"邮箱、姓名和经过哈希处理的密码(我们绝不以明文存储您的密码)","googleLabel":"Google、Microsoft 或 Apple 登录：","googleBody":"来自该服务商的邮箱、姓名和基本资料。我们绝不会获取您在该服务商的密码。","paidLabel":"付费方案：","paidBody":"由 Stripe 处理的账单信息。我们会存储您的客户 ID 和订阅状态，绝不存储您完整的卡号。","sessionCookies":"用于身份验证和自动刷新会话的会话记录，包括每个已登录会话的 IP 地址以及浏览器或设备类型，我们用这些信息保障您的账户安全","websiteCookiesPrefix":"用于分析和广告的网站 Cookie，取决于您的 Cookie 选择（在欧洲经济区、英国和瑞士为选择加入，在其他地区为选择退出）。请查看我们的","cookiePolicyLink":"Cookie 政策","websiteCookiesSuffix":"。"},"voice":{"title":"语音、音频和转录","intro":"桌面应用提供三种转录模式；移动应用提供本地处理和 OpenWhispr Cloud。您可以选择最适合自己需求的一种，并随时在设置中更改。","localLabel":"本地处理。","localBody":"OpenWhispr 可使用本地开源模型，在您的设备上完全完成语音转文字。选择此模式时，不会有任何音频被传输到任何地方。","byokLabel":"自带 API 密钥。","byokBody":"如果您提供自己的第三方转录服务商 API 密钥，您的音频将直接发送至该服务商，在您的账户下进行处理。届时适用其隐私政策。我们不会查看、存储或代理这些音频。","cloudLabel":"OpenWhispr Cloud。","cloudBodyPrefix":"我们的托管转录服务。音频会发送至我们的 API，或在实时转录时直接发送至我们的某一语音转文字服务商，经实时处理，且不会被我们存储。我们会为每次转录保留一条使用记录（字数、服务商、模型和时长），用于运行服务和计费，但不会保留文本本身，除非您开启云端备份。我们不会使用您的数据训练 AI 模型。请查看我们的","securityLink":"安全页面","cloudBodySuffix":"，了解训练数据选择退出选项；当前的服务商列表见下方的次级处理者表。","retentionLabel":"音频保留（桌面应用，默认开启）。","retentionBody":"桌面应用默认会将录制的听写音频在您的设备上保留 30 天，以便您回顾或重新处理，之后自动删除。您可以在设置中选择更短或更长的期限（最长 90 天），或将其关闭。保留的音频绝不会同步到 OpenWhispr Cloud；只有当您选择重新转录某段录音时，它才会离开您的设备，并发送至您所选的转录服务。会议录音不会被保留（见下方“系统音频捕获”）。","transcribedLabel":"转录的文本","transcribedBody":"存储在您设备上的本地 SQLite 数据库中。在桌面应用中，如果您开启云端备份（默认关闭，适用于付费方案），您的转录内容——最终文本以及清理前的原始文本——也会存储在我们的云端数据库中，以便在您的所有设备上显示。移动应用仅在您的设备上保留听写历史。","dictionaryLabel":"自定义词典。","dictionaryBody":"如果您添加单词或短语（技术术语、人名、行话）——或 OpenWhispr 从您的更正中学习到它们——这些内容会存储在您的设备上，并作为提示发送给您的转录和 AI 清理服务商。如果开启了云端备份，您的词典也会同步到我们的云端数据库。它们不会用于其他任何用途。"},"notes":{"title":"笔记和 AI 智能体对话","paragraph1":"笔记存储在您设备上的本地 SQLite 数据库中。一条笔记可以包含标题、正文内容、AI 增强版本(如果您使用增强功能),以及——对于会议笔记而言——从您的日历邀请中提取的会议记录和参会者名单。","syncLabel":"云端备份与共享（可选）。","syncBody":"如果您开启云端备份（适用于付费方案），您的笔记、文件夹、词典和片段——以及在桌面应用中的智能体对话和转录内容——会存储在我们的云端数据库（Neon，静态加密）中，以便您在多台设备间访问。备份在桌面应用中默认关闭，在移动应用中默认开启，您可以随时将其关闭。您共享的笔记以及团队空间中的笔记，无论此设置如何，都会存储在我们的云端。当笔记存于我们的云端时，我们会将其文本发送给 OpenAI，为其建立搜索索引。音频绝不会同步。Insights 同步是另一项需单独开启的功能，它会上传不含内容的计数数据（字数、每次听写的日期和时间、时长和模型类型）。","agentLabel":"AI 智能体(可选)。","agentBody":"当您使用 AI 智能体时，您的提示词以及任何工具结果（它读取的笔记、它查询的日历事件、网络搜索、您要它处理的剪贴板内容）会被发送至您配置的 AI 服务商；如果您在本地模型上运行它，这些内容则会保留在您的设备上。无论您使用哪种模型，智能体执行的网络搜索都会经由 OpenWhispr 发送至我们的搜索服务商 Exa；如果开启了云端备份，笔记搜索将由我们的云端响应，云端会使用 OpenAI 处理该搜索。通过 OpenWhispr Cloud，智能体还会对您的对话保留长期记忆，由我们的服务商 Mem0 存储，以便它回忆先前的上下文；该记忆会在您删除账户时被删除。","searchLabel":"本地语义搜索。","searchBody":"系统会使用一个在本地运行的小型嵌入模型，在您的设备上为笔记建立搜索索引，该设备端索引绝不会离开您的设备。存储在我们云端的笔记也会在云端建立索引，如上所述。"},"calendar":{"title":"日历（可选）","intro":"连接日历是可选的。您可以连接 Google Calendar 或 Microsoft Outlook / Microsoft 365 日历；在 Mac 上，也可以让 OpenWhispr 读取“日历”App。OpenWhispr 会读取您的日历列表和即将到来的事件（只读——我们绝不会创建、编辑或删除任何内容），并利用这些数据来：","use1":"在预定会议即将开始时通知您","use2":"根据事件名称为您的会议笔记建议标题","use3":"将日历邀请的参会者附加到该笔记中","cached":"即将到来的事件会与您的笔记和转录一起缓存在本地。连接直接在您的设备与日历服务商之间进行，我们不会将您的日历存储在我们的服务器上、共享它，或将其用于广告或 AI 训练。只有当您向 AI 智能体询问事件时，事件详情才会到达我们的服务器和您的 AI 服务商；如果会议笔记已备份或共享，其标题和参会者名单会随笔记同步。","disconnectPrefix":"在 OpenWhispr 设置中断开连接即可删除本地缓存，或随时通过您的","permissionsLink":"Google 账户权限","disconnectSuffix":"或 Microsoft 账户的应用权限撤销访问权限。","limitedUseLabel":"有限使用。","limitedUsePrefix":"OpenWhispr 对从 Google API 接收的信息的使用和传输，在传递给任何其他应用时，均遵守","userDataPolicyLink":"Google API 服务用户数据政策","limitedUseSuffix":",包括其中的有限使用要求。"},"detection":{"title":"会议检测信号","intro":"OpenWhispr 可以检测会议何时进行，从而主动提示为您记笔记。为此，它仅在您的设备上监控两件事：","signal1":"您的电脑上是否正在运行已知的会议应用(Zoom、Teams、Webex、FaceTime)","signal2":"是否有任何应用正在使用您的麦克风(在 macOS、Windows 和 Linux 上)","body":"OpenWhispr 不会捕获那些应用正在做什么、它们在录制什么，或谁在通话中。它只知道一个非此即彼的事实：“某个会议应用已打开”或“麦克风正在使用中”。这些信号仅保存在内存中，绝不会发送到任何地方（如果您开启调试日志，应用的本地日志文件会记录检测到的是哪个会议应用）。您可以在设置中禁用会议检测。","systemAudioLabel":"系统音频捕获（可选）。","systemAudioBody":"OpenWhispr 可以同时录制您的麦克风和来自其他应用的音频，从而转录会议双方的内容。该功能适用于 macOS 14.2 或更高版本（您须先授予 macOS 的“系统音频录制”权限），以及 Windows 和 Linux。会议音频（无论来自您的麦克风还是其他应用）会在您录制时进行处理，且不会被保留。为区分不同说话人，应用可能会在您的电脑上写入一份临时工作副本，并在处理完成后将其删除。"},"accessibility":{"title":"基于辅助功能的功能","intro":"当您首次使用 OpenWhispr 粘贴转录内容时，它会请求您操作系统的辅助功能权限。我们将其用于两件事：","autopasteLabel":"自动粘贴：","autopasteBody":"将转录的文本粘贴到您当前聚焦的任何文本框中，这样您就无需手动复制粘贴。","correctionLabel":"纠错学习：","correctionBody":"在粘贴后最长 30 秒内，OpenWhispr 会观察聚焦的文本框，留意您对它所粘贴文本进行的修改。如果您修正了一个拼错的单词，OpenWhispr 会将更正后的单词添加到您的自定义词典（见上文），从而让未来的转录更加准确。它所观察的文本绝不会发送给 OpenWhispr 或任何第三方。您可以在设置中通过“从纠正中自动学习”将此功能关闭。","revoke":"您可以随时在操作系统设置中撤销辅助功能权限，这将同时禁用上述两项功能。"},"diagnostic":{"title":"服务日志（OpenWhispr Cloud）","intro":"当您使用 OpenWhispr Cloud 时，我们的服务器会记录每个请求的技术细节，以便我们运行服务、计费、提供支持和修复问题：","metric1":"音频时长、大小和格式","metric2":"语音转文字服务商、模型、处理时间、语言和字数","metric3":"AI 模型服务商、模型、处理时间和 token 数量","metric4":"往返耗时和错误信息，包括来自我们服务商的错误消息","metric5":"客户端信息（客户端类型和应用版本）","body":"这些记录与您的账户以及每个应用会话的随机 ID 相关联，且绝不包含您的音频或您听写的文本。我们使用它们来运行、支持和改进 OpenWhispr，并了解哪些营销渠道带来了活跃用户；它们绝不会被出售或与广告商共享。无论您的设置如何，这些记录都会针对 OpenWhispr Cloud 请求生成——桌面应用中的“使用分析”设置目前无法将其关闭。"},"referral":{"title":"推荐计划(可选)","intro":"如果您参与推荐计划，我们会收集：","item1":"您的专属推荐码以及您发送的邀请","item2":"您邀请的人的邮箱地址(上传至我们的服务器，以便我们追踪转化并防止滥用)","item3":"对您推荐链接的访问（IP 地址、用户代理、来源 URL）","item4":"某次推荐是否带来了注册或订阅","body":"只有当您主动发起邀请时，我们才会向收件人发送邮件。推荐数据仅用于归属奖励和防止滥用——别无他用。"},"mobile":{"title":"移动应用：分析、归因与应用内购买","intro":"我们的移动应用包含少量桌面应用所没有的第三方软件开发工具包（SDK）。它们帮助我们了解应用的使用情况以及哪些营销有效，并处理应用内购买。我们会利用它们收集的数据来改进产品，并衡量和拓展我们的营销。它们都不会收到您的音频、转录内容、笔记或智能体对话。","appsflyerLabel":"AppsFlyer（归因与应用分析）。","appsflyerBody":"告诉我们哪些广告和推广活动带来了安装与订阅，以及应用的使用情况，以便我们改进产品并合理使用营销预算。它会收到设备标识符和广告标识符、IP 地址、设备与操作系统信息，以及安装、打开应用、完成教程、查看升级页面、完成转录（仅记录来源和服务商，绝不包含文本）、开始订阅等应用事件，还有少量非内容属性，例如展示了哪个升级页面、购买了哪款产品。在 iOS 上，只有当您在 Apple 询问时允许跟踪，AppsFlyer 才会运行；如果您拒绝，则不会向它发送任何数据。AppsFlyer 可能也会用于衡量我们桌面应用的营销效果；在这种情况下，它收到的是同类数据，绝不包含听写内容。","revenuecatLabel":"RevenueCat（应用内购买）。","revenuecatBody":"记录并验证通过 App Store 或 Google Play 完成的购买，使您的方案在各设备之间保持同步，并让我们了解订阅收入。它会收到您账户的随机计费标识符（而非您的邮箱地址）、商店购买收据、您的订阅状态、设备类型和操作系统，以及 IP 地址。","superwallLabel":"Superwall（升级页面）。","superwallBody":"为应用中的升级页面（付费墙）提供支持，并让我们测试哪些设计和价格效果最好，从而改进升级体验。它会收到同一个随机计费标识符、设备、操作系统和应用版本、您的方案与使用状态（方案、试用状态、已用及剩余字数、处理模式），以及展示、关闭、购买等升级页面事件。","sentryLabel":"Sentry（崩溃与错误报告）。","sentryBody":"收集崩溃报告、错误信息和应用日志信息，以便我们修复缺陷。它被配置为不发送个人标识符，且仅在发生错误时捕获一段简短的屏幕回放，并对屏幕上的文本进行遮蔽。错误报告遵循与 AppsFlyer 相同的“Usage Analytics”开关设置。","controlsLabel":"您的控制权。","controlsBody":"在应用的“Privacy & Data”（隐私与数据）设置中关闭“Usage Analytics”（使用情况分析），即可停止收集分析数据。在 iOS 上，您还可以随时在“设置 → 隐私与安全性 → 跟踪”中更改跟踪许可。RevenueCat 和 Superwall 是完成和管理应用内购买所必需的，无法单独关闭。"}},"use":{"title":"我们如何使用您的信息","intro":"我们将所收集的数据用于以下目的。如果您位于欧洲经济区或英国，每项目的的法律依据在“国际用户”部分说明。","delivery":{"title":"服务交付","item1":"提供转录、笔记、同步、共享、团队和智能体功能","item2":"处理付款并管理订阅","item3":"发送服务消息并交付软件更新","item4":"提供客户支持"},"improvement":{"title":"安全与改进","item1":"使用服务日志来运行、支持和改进 OpenWhispr","item2":"修复缺陷并开发新功能","item3":"保障账户安全，防止欺诈和滥用，包括对我们推荐计划的滥用","item4":"履行法定义务，例如保存税务记录和回应合法请求"},"growth":{"title":"营销与增长","item1":"通过邮件向您发送有关我们自有方案的产品资讯和优惠（可随时退订）","item2":"衡量哪些广告和推广活动带来了访问、安装和订阅，并向访问过我们网站的人展示我们的广告","item3":"测试并改进我们的升级页面和定价，并了解订阅与续订情况","item4":"识别从美国访问我们网站的个人和公司，并邀请部分用户参加反馈通话"},"neverLabel":"我们绝不会","neverBody":"将您的转录内容、笔记或音频用于广告、向第三方营销，或训练我们自己的 AI 模型。绝无例外。"},"sharing":{"title":"我们如何共享信息","intro":"我们不会出售、交易或出租您的个人信息。我们仅与下列服务提供商共享信息，或在本节所述的其他有限情形下共享，以及在您选择时与其他 OpenWhispr 用户共享：您共享的笔记、您加入的团队空间，或您加入的排行榜——排行榜会向使用相同工作邮箱域名的人显示您的姓名和活动计数（绝不包括您的转录内容），并且还会向您工作区的成员显示您的邮箱地址。","subprocessors":{"title":"次级处理者","intro":"以下是为我们处理个人数据的服务提供商，用于运营 OpenWhispr Cloud、我们的应用、网站、支持和营销，另包括您的设备所连接的登录、日历和下载服务。每家服务提供商仅接收实现所示用途所需的信息。Google 和 Meta（用于广告）、RB2B 以及各登录服务提供商也会根据其自身的隐私政策，将其接收的数据用于其自身目的。服务提供商发生变更时，我们会更新此表；受我们《数据处理附录》约束的企业客户，将按该附录的规定获得变更的提前通知。","headerService":"服务","headerPurpose":"用途","headerDataCategory":"数据类别","rows":[{"service":"Neon","purpose":"我们的主数据库：账户、云端备份与同步、使用记录和服务日志","dataCategory":"账户信息、已同步的笔记、转录与对话、使用记录、服务日志"},{"service":"Stripe","purpose":"支付处理与订阅管理","dataCategory":"账单信息、姓名、客户邮箱"},{"service":"Resend","purpose":"服务邮件与营销邮件发送","dataCategory":"邮箱地址、姓名、邮件内容，以及用于决定您收到哪些邮件的方案、地区、活动和团队信息（例如有多少同事使用 OpenWhispr）"},{"service":"Vercel","purpose":"网站和 API 托管，以及无 Cookie 的网站统计","dataCategory":"网络和 API 请求（包括 IP 地址和传输中的内容）、匿名化页面指标"},{"service":"Google（Tag Manager、Analytics 4、Ads）","purpose":"网站分析、广告衡量与再营销（按您的 Cookie 选择）","dataCategory":"假名化使用数据及设备/Cookie 标识符（见 Cookie 政策）；购买时还包括广告点击 ID、订单 ID 和金额"},{"service":"Meta（Pixel、Conversions API）","purpose":"衡量我们的 Meta 广告（按您的 Cookie 选择）","dataCategory":"页面浏览和下载，附带 Cookie 和浏览器 ID、IP 地址及用户代理；购买时还包括经哈希处理的邮箱地址和账户 ID"},{"service":"RB2B","purpose":"用于销售跟进的 B2B 网站访客识别（仅限美国访客；默认启用，可选择退出，遵循 GPC）","dataCategory":"根据您的 Cookie 选择，将设备和网络标识符与公开职业资料（姓名、公司、工作电子邮件）进行匹配（请参阅 Cookie 政策）"},{"service":"Google、Microsoft、Apple","purpose":"使用您现有的账户登录；Google 和 Microsoft 日历（如已连接）","dataCategory":"电子邮件和个人资料；日历列表和事件（发送至您的设备）"},{"service":"OpenAI","purpose":"OpenWhispr Cloud 背后的云端语音转文本、语言模型和笔记搜索索引（使用我们的 API 密钥）","dataCategory":"语音音频（临时）、提示词、文本、存储在我们云端的笔记文本"},{"service":"Groq","purpose":"OpenWhispr Cloud 背后用于故障转移的云端 AI 推理（使用我们的 API 密钥）","dataCategory":"提示词、文本"},{"service":"Deepgram","purpose":"OpenWhispr Cloud 背后部分转录模型的实时语音转文本（使用我们的 API 密钥）","dataCategory":"语音音频（流式传输，临时）"},{"service":"Parasail","purpose":"OpenWhispr Cloud 文本清理、格式化、翻译及 AI 智能体背后的默认语言模型","dataCategory":"听写文本、提示词、智能体消息、笔记内容"},{"service":"Mem0","purpose":"OpenWhispr Cloud AI 智能体的长期记忆","dataCategory":"智能体消息和回复，与您的账户 ID 关联"},{"service":"Exa","purpose":"AI 智能体的网络搜索","dataCategory":"智能体编写的搜索查询"},{"service":"AppsFlyer","purpose":"面向产品、营销与增长的安装归因及应用使用分析（移动应用；也可能用于衡量我们桌面应用的营销）。可通过“Usage Analytics”开关关闭；在 iOS 上仅当您允许跟踪时才会运行","dataCategory":"设备标识符与广告标识符、IP 地址、设备/操作系统信息、应用事件（安装、打开、教程、查看升级页面、完成转录、订阅；完成转录仅含来源与服务商）。绝不包含音频或文本"},{"service":"RevenueCat","purpose":"移动应用的应用内购买验证与订阅状态","dataCategory":"随机计费标识符、商店购买收据、订阅状态、设备类型与操作系统、IP 地址"},{"service":"Superwall","purpose":"移动应用的应用内升级页面（付费墙）、定价及升级页面实验","dataCategory":"随机计费标识符、设备/操作系统/应用版本、方案与使用状态、升级页面与购买事件"},{"service":"Sentry","purpose":"移动应用的崩溃与错误报告（错误报告须经“Usage Analytics”开关同意）","dataCategory":"崩溃报告、错误信息、应用日志信息、发生错误时的遮蔽屏幕回放；不含个人标识符"},{"service":"Expo (EAS Update)","purpose":"移动应用的应用更新推送；应用启动时会向 Expo 的服务器查询是否有更新版本","dataCategory":"平台与应用运行时版本、用于判断更新是否已下载的随机令牌、IP 地址。不含账户或内容数据"},{"service":"GitHub、Hugging Face","purpose":"桌面应用更新及设备端 AI 模型下载","dataCategory":"IP 地址、应用版本和平台。不含账户或内容数据"},{"service":"Google Workspace（Gmail）、Linear、Slack","purpose":"处理支持请求和销售咨询，以及向我们团队发送内部提醒","dataCategory":"您的消息和附件、姓名、邮箱地址、您提供给我们的详细信息，以及登录邮箱验证等账户事件"},{"service":"lemlist、Hetzner","purpose":"通过邮件邀请部分用户参加反馈通话（自 2026 年 9 月起暂停）：联系人信息补充、发送和托管","dataCategory":"姓名、工作邮箱、账户活跃程度，以及 lemlist 为我们查找到的公司和公开职业信息（例如公司规模和 LinkedIn 资料）"}]},"byok":{"title":"您连接的第三方 AI 服务商(BYOK)","intro":"您也可以使用自己的 API 密钥直接连接这些提供商。在该 BYOK 路径中，您的数据通过您的账户直接发送给提供商；OpenWhispr 不是您发送内容的数据控制者，并适用该提供商的隐私政策和账户设置。其中一些提供商（例如 OpenAI 和 Groq）也会在使用我们自有密钥时作为 OpenWhispr Cloud 的次级处理者运行；请参阅上方的次级处理者表。BYOK 路径是独立的，仅在您提供自己的密钥时适用。通过我们的 API 或 MCP 连接器连接到您账户的应用也是如此：它们可以按照其自身条款，读取和更改您允许的笔记、转录内容和其他数据。","transcriptionLabel":"云端转录：","transcriptionBody":"OpenAI、Groq、xAI、Mistral、Google Gemini、Deepgram、AssemblyAI、Corti、Tinfoil、自定义端点","agentLabel":"AI 智能体 / 推理：","agentBody":"OpenAI、Anthropic、Google Gemini、Groq、Corti、Tinfoil、自定义端点，以及您组织自有的 AWS Bedrock、Azure OpenAI 或 Google Vertex AI"},"local":{"title":"本地处理(不涉及任何第三方)","body":"OpenWhispr 内置了可选的设备端语音转文字和 AI 模型。当您使用这些模型时，您的音频、转录内容和智能体提示词完全在您的设备上处理，不会发送至 OpenWhispr 或任何第三方——但您要求智能体执行的网络搜索、在开启备份时由我们云端响应的笔记搜索（见上文“AI 智能体”），以及您选择备份或共享的任何内容除外。下载模型时，您的 IP 地址会被共享给托管该模型的网站（见上表）。"},"legal":{"title":"法律与业务转让","body":"如法律、法院命令要求，或为保护我们的权利和安全，我们可能会披露信息。如发生合并、收购或出售，您的信息可能会在与本政策一致的条款下转移给新的所有者。"}},"retention":{"title":"数据保留与删除","localLabel":"本地数据","localBody":"（笔记、转录、智能体对话、保留的音频）会保留在您的设备上，直到您将其删除（保留的音频也会自动删除，见下文）。卸载桌面应用不会删除这些数据：请使用设置中的“重置应用数据”，在删除账户时选择“同时清除此设备上的所有数据”，或自行删除应用的数据文件夹。在 iPhone 上，删除该应用会移除其本地数据。","audioLabel":"音频保留（桌面应用）","audioBody":"默认会在您的设备上保留听写音频 30 天，之后自动删除；您可以选择 1 至 90 天，或将其关闭。","cloudLabel":"账户与云端数据","cloudBody":"（您的账户、已备份和共享的内容、Insights、智能体记忆和使用记录）会在您的账户存续期间保留。当您删除账户时，我们会立即删除这些数据，最迟在 30 天内完成；您放入团队空间的笔记属于该工作区，并会保留在该工作区中。您从我们云端删除的项目会在 30 天后被永久移除。我们会在税务和会计法律要求的期限内保留发票和付款记录。删除账户目前还不会将您从我们的营销邮件列表中移除，因此请使用退订链接或联系我们。支持邮件会保留至不再需要用于帮助您或证明我们如何处理您的请求为止。","calendarLabel":"日历本地缓存","calendarBody":"会在您从 OpenWhispr 设置中断开日历连接时立即删除。","diagnosticLabel":"服务日志","diagnosticBody":"在您的账户存续期间保留，并在您删除账户时删除。"},"security":{"title":"安全","item1":"所有云端连接均使用 TLS 1.2 或更高版本。","item2":"我们的云端数据库(Neon)采用静态加密。","item3":"账户密码由我们的登录系统进行哈希处理；我们绝不以明文存储密码。","apiKeysLabel":"API 密钥","apiKeysPrefix":"您为第三方服务商输入的密钥会在您的设备上加密（AES-256），加密密钥保存在您操作系统的安全凭据存储中（macOS 上的“钥匙串”、Windows 上的“凭据管理器”、Linux 上的 Secret Service）。如果该存储不可用，应用会改用 Electron 内置的加密存储；仅当两者都不可用时（例如在某些 Linux 环境中），密钥才会以未加密形式保存在应用的数据文件夹中（参见","apiKeysLink":"issue #532","apiKeysSuffix":"）。","diskEncryption":"您的本地 SQLite 数据库未被 OpenWhispr 加密——它依赖于您操作系统的磁盘加密(macOS 上的 FileVault、Windows 上的 BitLocker、Linux 上的 LUKS)。我们建议您保持磁盘加密处于启用状态。","morePrefix":"如需了解更多关于我们安全实践的内容，请查看我们的","securityLink":"安全页面","moreSuffix":"。"},"rights":{"title":"您的权利与选择","intro":"对于您的个人信息，您享有以下权利：","accessLabel":"访问与可携带性：","accessBody":"在应用内将任意笔记导出为 Markdown 文件，或发送邮件至 support@openwhispr.com，获取我们所持有的关于您的个人信息的完整副本，包括已同步的笔记、转录历史和账户数据。","correctionLabel":"更正：","correctionBody":"在应用的账户设置中更改您的姓名，以及您的密码（如果您使用密码登录）。如需更正您的邮箱地址或我们持有的其他任何信息，请给我们发送邮件。","deletionLabel":"删除：","deletionBody":"在应用内删除单条笔记和转录、清除您的历史记录，或删除您的账户及其云端数据——也可以给我们发送邮件，由我们为您处理。","optoutLabel":"选择退出与同意：","optoutBody":"随时更改您的网站 Cookie 选择（网站页脚中的“Cookie 设置”），在移动应用中关闭“Usage Analytics”（使用情况分析）或在 iPhone 设置中关闭跟踪，关闭云端备份或 Insights 同步，退出排行榜，断开日历连接，或通过任意邮件中的链接退订营销邮件。撤回同意不影响撤回前已进行的处理。您还可以随时反对直接营销或基于我们正当利益的处理，或删除您的账户。"},"international":{"title":"国际用户","intro":"Gizmo Labs Inc. 位于美国。我们的服务器和大多数服务提供商位于美国，因此您的信息在美国处理；当您从其他国家或地区使用 OpenWhispr 时，您是将信息发送给位于美国的我们。如果企业客户对 OpenWhispr 的使用涉及从欧洲经济区、英国或瑞士传输个人数据，我们的《数据处理附录》将适用欧盟委员会的标准合同条款（附带英国附录和瑞士修订）。您可以向我们索取一份副本。","gdpr":{"title":"欧洲经济区和英国用户（GDPR）","body":"我们的法律依据如下：合同——用于提供您的账户和您使用的功能、收取付款、提供支持以及回复销售咨询；正当利益——用于保障 OpenWhispr 的安全并防止滥用；利用服务日志、崩溃报告和无 Cookie 的网站统计来运行、支持和改进 OpenWhispr；测试我们的升级页面和价格，并了解哪些营销渠道带来了活跃用户；发送您要求发送的推荐邀请；通过邮件向您介绍我们自己的产品（相关规则允许我们向现有用户发送此类邮件，直至其退订，即所谓的“软性选择加入”（soft opt-in））；以及邀请部分用户参加反馈通话；同意——用于网站分析和广告 Cookie、移动端广告归因以及排行榜；以及法定义务——用于保存税务记录和回应合法请求。您可以随时撤回同意、反对基于正当利益的处理（对于直接营销，您始终可以反对）、要求我们限制处理，并向您居住地或工作地的数据保护机构提出投诉。"},"ccpa":{"title":"加利福尼亚州用户(CCPA)","body":"您有权知晓我们收集哪些个人信息，并有权访问、更正和删除这些信息。我们不出售个人信息。根据加利福尼亚州法律，我们网站的广告 Cookie 和 RB2B 访客识别可能被视为用于定向广告的“共享”；您可以通过网站页脚中的“您的隐私选择”或发送全球隐私控制（Global Privacy Control）信号来选择退出。我们不会因您行使这些权利而区别对待您。"}},"changes":{"title":"本政策的变更","intro":"我们可能会不时更新本隐私政策。我们将通过以下方式就重大变更通知您：","item1":"在我们的网站上发布更新后的政策","item2":"向注册用户发送邮件通知","item3":"更新本页顶部的“最后更新”日期","body":"在变更生效后继续使用 OpenWhispr,即表示您接受更新后的政策。"},"contact":{"title":"联系我们","intro":"如果您对本隐私政策有任何疑问、顾虑或请求——包括行使您的任何权利——请联系我们：","teamName":"OpenWhispr 团队","emailLabel":"邮箱：","subjectLine":"邮件主题：隐私咨询","responsePrefix":"我们力争在 48 小时内回复与隐私相关的咨询，并在一个月内完成请求的处理。如需以通俗语言了解我们如何处理您的数据以及我们绝不会做什么，请查看我们的","securityLink":"安全页面","responseSuffix":"。"},"legalNotice":"本文件的英文版本为权威版本；如任何译文与英文版本不一致，以英文版本为准。本地化版本仅为方便阅读而提供，在作为依据前应经过法律顾问审阅。"},"legalTerms":{"metadata":{"title":"服务条款 | OpenWhispr","description":"规范您使用 OpenWhispr 软件与服务的条款。开源许可、订阅与数据所有权。","keywords":["OpenWhispr 服务条款","OpenWhispr 条款","条款与条件","软件许可","订阅条款","数据所有权"]},"breadcrumb":{"home":"首页","terms":"服务条款"},"header":{"title":"服务条款","subtitle":"您与 OpenWhispr 之间的协议。","lastUpdated":"最后更新：2026年9月11日"},"overview":{"title":"概述","p1":"本服务条款（以下简称\"条款\"）是 Gizmo Labs Inc.（一家以 OpenWhispr 名义经营的特拉华州公司，以下简称\"OpenWhispr\"、\"我们\"）与您（以下简称\"客户\"、\"您\"）之间的合同。本条款规范您对 OpenWhispr 软件、服务及网站的使用。","p2":"使用 OpenWhispr 即表示您接受本条款。我们对您个人数据的处理受我们的隐私政策约束，该政策以引用方式纳入本条款。"},"definitions":{"title":"定义","intro":"为清晰起见，本文档中以下术语具有特定含义：","service":"服务 — 指 OpenWhispr 桌面应用、OpenWhispr Cloud API、openwhispr.com 网站，以及我们提供的任何相关软件或服务。","customer":"客户 / 您 — 指使用本服务的个人或组织。","account":"账户 — 指您已注册的 OpenWhispr 账户。","customerContent":"客户内容 — 指您通过本服务创建、上传或处理的音频录音、转录文本、笔记、提示词、文件及其他内容。","customerData":"客户数据 — 指客户内容，以及有关您或您组织的任何账户信息。","subscription":"订阅 — 指您已注册的付费方案。","subprocessor":"次级处理者 — 指我们用于运营本服务的第三方服务。现行清单载于我们的隐私政策。"},"service":{"title":"本服务","intro":"OpenWhispr 是一款桌面口述、转录与会议记录工具，并配有可选的 AI 智能助理和 Google Calendar 集成。您可自行选择音频的处理方式：","local":"本地处理（默认）。语音转文字完全在您的设备上使用本地开源模型运行。任何音频都不会离开您的电脑。","byok":"自带 API 密钥。您连接来自第三方转录或 AI 提供方的自有 API 密钥。您的音频将直接发送至该提供方。详见第 6 条。","cloud":"OpenWhispr Cloud。我们的托管式转录服务。音频发送至我们的 API，实时处理后即被丢弃。现行提供方清单及训练退出选项请见我们的安全页面。","openSource":"开源桌面应用。OpenWhispr 桌面应用依据 MIT 许可证发布。完整源代码可在 GitHub 上获取。OpenWhispr Cloud API 与 openwhispr.com 网站为专有，由我们运营。","noSla":"无服务等级协议（SLA）。OpenWhispr 目前不为本服务提供 SLA。我们力求提供可靠的服务，但不作任何正常运行时间的承诺。需要 SLA 的企业客户可联系 sales@openwhispr.com 获取附加条款。"},"account":{"title":"您的账户","intro":"您可以使用电子邮件和密码创建 OpenWhispr 账户，也可以通过 Google 登录创建。部分服务功能需要账户。","bullet1":"每位个人或每个组织限一个账户","bullet2":"请妥善保管您的凭据。在您账户下发生的一切由您负责。","bullet3":"若您怀疑账户遭到未授权访问，请立即告知我们","bullet4":"对于严重违反本条款的行为，或在长期不活跃后，我们可能暂停或关闭账户","age":"年龄要求。OpenWhispr 面向 13 岁及以上用户。若您未满 18 岁，即表示您确认您的父母或法定监护人已阅读本条款，并代您同意本条款。"},"acceptableUse":{"title":"可接受使用","intro1":"OpenWhispr 桌面应用依据 MIT 许可证开源 — 任何人均可在该许可证下自由使用、修改和分发。以下可接受使用规则适用于您使用本服务的专有部分（OpenWhispr Cloud、网站、我们的 API）之时，以及适用法律对您可对任何版本所做之事加以限制之时。","intro2":"请仅将 OpenWhispr 用于合法目的。您不得使用本服务：","bullet1":"在法律要求的情况下，未经任何个人知情和同意而录制其内容（见第 8 条 — 录音同意义务由您承担，而非由我们承担）","bullet2":"转录您无合法权利录制或处理的内容（受版权保护的广播、您并非参与方的机密通话等）","bullet3":"生成、分发或存储非法、有害、诽谤、骚扰、仇恨性或侵犯他人权利的内容","bullet4":"制作深度伪造、声音克隆，或任何意图欺骗或冒充他人的内容","bullet5":"对本服务的专有部分进行逆向工程、反编译，或试图提取其源代码。桌面应用采用 MIT 许可证，您可在 GitHub 上阅读其全部内容；而 OpenWhispr Cloud API 与网站则不然。","bullet6":"使用 OpenWhispr Cloud API 生成的输出来训练、开发或改进竞争性的人工智能或机器学习产品。此项限制仅适用于 OpenWhispr Cloud — 采用 MIT 许可证的桌面应用不受此限制。","bullet7":"绕过速率限制、用量上限或访问控制","bullet8":"以扰乱其他用户的方式发送垃圾信息、抓取数据，或使用自动化工具访问本服务","bullet9":"未经我们书面许可，转售 OpenWhispr Cloud 的访问权限或对其再授权","bullet10":"未经所有被录制者同意即上传或转录音频。您有责任遵守录音发生地适用的录音及隐私法律。","bullet11":"未经我们书面同意，转售、再授权或将 OpenWhispr 转录服务或 OpenWhispr API 作为独立服务提供，或利用其构建竞争性的转录服务","bullet12":"超出适用于 API 和 CLI 访问的速率限制或用量上限。为保护本服务，我们可自行决定限流、暂停或吊销 API 密钥或云端访问权限；标记为测试版（beta）的功能可能随时变更或暂停，恕不另行通知。","outro":"若您滥用 OpenWhispr，我们可能暂停或关闭您的账户，情节严重者我们将寻求法律救济。"},"byok":{"title":"自带密钥（BYOK）","p1":"OpenWhispr 允许您通过提供自有 API 密钥来连接第三方转录和 AI 提供方（如 OpenAI、Anthropic、Google Gemini、Deepgram、AssemblyAI、Groq、Mistral 或自定义端点）。现行受支持提供方清单载于我们的隐私政策，并可能随时间变化。","p2":"当您使用 BYOK 时，您的音频和提示词将直接发送至您所选择的提供方。我们不会查看、存储或代理该流量。您的关系存在于您与该提供方之间 — 该提供方与您的关系亦是如此：","bullet1":"除我们的条款外，您还受该第三方提供方条款的约束","bullet2":"您需自行负责您的 API 费用、速率限制、数据处理，以及对底层提供方政策的遵守","bullet3":"我们不对您所连接的任何第三方提供方的行为、错误、停机或费用承担责任"},"aiOutputs":{"title":"AI 输出与用户核实","p1":"本服务使用机器学习来生成转录文本、AI 增强笔记、智能助理回复及其他输出。与所有 AI 系统一样，OpenWhispr 可能产生不准确、不完整、有偏见或完全虚构的结果（\"幻觉\"）。","p2":"在依赖 AI 生成的内容或据此采取行动之前，您有责任对其进行审阅和核实。我们不对任何 AI 输出的准确性或完整性作出保证。","p3":"AI 输出并非专业建议。请勿将其作为法律、医疗、财务、安全关键或其他高风险决策的唯一依据。"},"customerContent":{"title":"客户内容与录音同意","ownTitle":"您拥有您的客户内容","ownBody":"OpenWhispr 不对您通过本服务创建的音频录音、转录文本、笔记或任何内容主张所有权。它们归您所有。","licenseTitle":"为运营本服务而授予的有限许可","licenseBody":"您授予我们一项有限许可，以便在为提供本服务所需的范围内处理您的客户内容 — 例如转录您提交的音频、存储您保存的笔记，或在您的各设备间同步内容。此许可仅限于为您运营本服务，并在您删除内容或关闭账户时终止。","noTrainingTitle":"绝不使用您的数据进行训练","noTrainingBody":"我们绝不会使用客户内容 — 包括音频、转录文本、笔记或 AI 智能助理提示词 — 来训练任何人工智能或机器学习模型。这是一项合同承诺，而不仅仅是隐私承诺。","consentTitle":"录音同意由您负责","consentBody":"您声明并保证，您对每一个使用 OpenWhispr 处理的音频源都拥有录制、转录和处理的合法权利。您全权负责遵守您或任何其他参与者所在每一司法管辖区的适用窃听、双方同意及录音披露法律。","indemnity":"对于因您创建、上传或处理的客户内容而引起的任何索赔 — 包括您未经同意录制他人的索赔 — 您将对 OpenWhispr 作出赔偿（另见第 12 条关于赔偿的规定）。"},"ip":{"title":"知识产权","desktop":"桌面应用 — 依据 MIT 许可证发布。完整源代码见 GitHub。","cloud":"OpenWhispr Cloud 与 openwhispr.com — 为专有，依据本条款向您授权（而非出售）。","customerContent":"客户内容 — 归您所有（见第 8 条）。","trademarks":"商标 — OpenWhispr 名称、徽标及相关标识归 Gizmo Labs Inc. 所有。未经我们书面许可，您不得使用。","feedback":"反馈许可 — 若您建议功能、报告漏洞或以其他方式与我们分享反馈，我们可自由使用该反馈，无需补偿或署名。"},"billing":{"title":"方案、计费与退款","freeTier":"免费版 — 包含每周 2,000 字的 OpenWhispr Cloud 转录额度。","paidPlans":"付费方案 — 按月或按年预先计费。除非您在下一计费日前取消，否则方案将自动续订。","pricingChanges":"定价变更 — 任何价格变更生效前，我们将提前 30 天通知。","processor":"支付处理方 — 所有付款由 Stripe 处理。我们绝不会看到您完整的卡号。","failedPayments":"付款失败 — 若付款失败，我们可能在宽限期后暂停本服务。","taxes":"税费 — 您需负责任何适用税费。","guarantee":"30 天退款保证。若您对新的付费订阅不满意，请在首次付款后 30 天内联系 support@openwhispr.com，我们将为您退款。退款将在 5–7 个工作日内退回至您的原付款方式。本保证仅适用于首次订阅 — 不适用于续订或定制企业服务。"},"privacy":{"title":"隐私","body":"我们对您个人数据的处理受我们的隐私政策约束，该政策以引用方式纳入本条款。隐私政策说明了 OpenWhispr 收集哪些数据、我们如何使用这些数据、还有哪些方处理这些数据，以及如何控制或删除这些数据。如有需要，可提供数据处理附录 — 请联系 support@openwhispr.com。"},"liability":{"title":"免责声明、责任与赔偿","asIsTitle":"\"按现状\"免责声明","asIsBody":"本服务\"按现状\"且\"按可用情况\"提供，不附带任何明示或默示的保证 — 包括对适销性、特定用途适用性或不侵权的任何保证。我们不保证转录的准确性、AI 输出的准确性（见第 7 条），也不保证服务不中断的可用性。","betaBody":"Beta 或预览功能按现状提供，可能在不另行通知的情况下变更或移除，并被排除在任何保证或服务承诺之外。","limitTitle":"责任限制","limitBody":"在法律允许的最大范围内，我们因本条款或本服务而产生或与之相关的总责任，以下列两者中较高者为上限：(a) 您在索赔发生前 12 个月内向我们支付的费用，或 (b) 100 美元。对于间接、附带、后果性、特殊或惩罚性损害 — 包括利润损失、数据丢失或业务中断 — 我们概不负责。","indemnityTitle":"赔偿","indemnityByYou":"由您赔偿。对于因以下原因引起的任何第三方索赔，您将赔偿、抗辩并使 Gizmo Labs Inc. 免受损害：(a) 您的客户内容，(b) 您对本服务的滥用，(c) 您违反本条款或任何法律，(d) 您未经授权对任何个人进行录音或转录（见第 8 条），或 (e) 您通过 BYOK 连接的任何第三方 AI 提供方（见第 6 条）。","indemnityByUs":"由我们赔偿。对于声称本服务本身在依照本条款使用时侵犯美国专利、版权或商标的任何第三方索赔，我们将赔偿、抗辩并使您免受损害。此义务不包括因您的修改、您将本服务与非 OpenWhispr 软件组合、Beta 功能、BYOK 提供方或客户内容而引起的索赔。我们的总赔偿义务以上述责任限制为上限。"},"termination":{"title":"期限、终止与变更","terminationTitle":"终止","terminationBody":"您可随时通过关闭账户或卸载桌面应用来停止使用 OpenWhispr。对于违反本条款、未付款或长期不活跃的情形，我们可能暂停或终止您的访问权限 — 除严重违约情形外，我们将提供合理通知。","dataTitle":"您的数据将会如何处理","dataBody":"您设备上的本地数据在您删除之前始终归您控制。卸载 OpenWhispr 会移除本地应用数据目录。云同步数据将在账户删除后 30 天内删除（与我们的隐私政策一致）。","surviving":"存续条款。以下条款在终止后继续有效：客户内容所有权（第 8 条）、知识产权（第 9 条）、付款义务（第 10 条）、免责声明与责任限制（第 12 条）、赔偿（第 12 条）、适用法律（第 14 条），以及本存续规则。","changesTermsTitle":"本条款的变更","changesTermsBody":"我们可能不时更新本条款。对于重大变更，我们将在其生效前至少 30 天通过电子邮件和/或应用内通知告知您。在生效日期之后，您继续使用 OpenWhispr 即表示您接受更新后的条款。","changesServiceTitle":"本服务的变更","changesServiceBody":"我们可随时新增、修改或停用功能。对于削减您所依赖功能的重大变更，我们将提供合理通知。Beta 功能可能随时变更或移除，恕不另行通知。"},"legal":{"title":"法律与一般条款","governingTitle":"适用法律与管辖地","governingBody":"本条款受美国特拉华州法律管辖，不考虑其法律冲突原则。任何法律诉讼均须在位于特拉华州的州法院或联邦法院提起（公司法事务由衡平法院管辖；联邦事务由美国特拉华州地区法院管辖）。","disputeBody":"在提起任何法律诉讼之前，您同意先联系 support@openwhispr.com，以尝试非正式地解决争议。","exportTitle":"出口管制","exportBody":"您不得以违反美国出口管制、OFAC 制裁的方式使用本服务，亦不得在任何受禁运国家/地区使用（目前包括古巴、伊朗、朝鲜、叙利亚，以及乌克兰的克里米亚、顿涅茨克和卢甘斯克地区）。","forceMajeureTitle":"不可抗力","forceMajeureBody":"对于因超出合理控制范围的事件所导致的延误或失败 — 自然灾害、战争、流行病、互联网中断、第三方服务故障或政府行为 — 任何一方均不承担责任。","dmcaTitle":"DMCA","dmcaBody":"若您认为本服务上的内容侵犯了您的版权，请向 support@openwhispr.com 发送 DMCA 删除通知，主题行注明\"DMCA Notice\"，并包含 17 U.S.C. § 512(c)(3) 所要求的信息。我们将依据 DMCA 对有效通知作出回应。","confidentialityTitle":"保密","confidentialityBody":"双方同意保护对方在合作期间共享的非公开商业信息，未经许可不予披露，但法律要求的情形除外。","publicityTitle":"宣传","publicityBody":"除非您通过发送邮件至 support@openwhispr.com 选择退出，否则我们可在营销材料中将您标识为客户。","noticesTitle":"通知","noticesBody":"我们通过电子邮件（发送至您账户上的地址）或应用内通知向您发出通知。您则通过电子邮件向 support@openwhispr.com 向我们发出通知。","otherTitle":"其他条款","severability":"可分割性 — 若本条款的任何条款被认定为不可执行，其余条款仍完全有效。","assignment":"转让 — 未经我们书面同意，您不得转让本条款。我们可在合并、收购，或出售全部或几乎全部资产时转让本条款。","waiver":"弃权 — 我们未执行任何条款，不构成对我们日后执行该条款权利的放弃。","entireAgreement":"完整协议 — 本条款连同隐私政策（以引用方式纳入）构成您与 OpenWhispr 之间就本服务达成的完整协议，并取代任何先前的协议。"},"contact":{"title":"联系我们","intro":"对本条款有疑问？欢迎联系：","company":"Gizmo Labs Inc.","dba":"以 OpenWhispr 名义经营","email":"电子邮件：support@openwhispr.com","subject":"邮件主题：服务条款咨询","seeAlso":"另请参阅我们的隐私政策和安全页面。"},"legalNotice":"本文件的英文版本为权威版本；如任何译文与英文版本不一致，以英文版本为准。本地化版本仅为方便阅读而提供，在作为依据前应经过法律顾问审阅。"},"legalCookies":{"metadata":{"title":"Cookie 政策 | OpenWhispr","description":"OpenWhispr 如何使用 Cookie。分析和营销类 Cookie 默认关闭，仅在您同意后才会加载。"},"breadcrumb":{"home":"首页","cookies":"Cookie 政策"},"hero":{"title":"Cookie 政策","subtitle":"openwhispr.com 如何使用 Cookie，以及您如何始终掌握主动权。","lastUpdated":"最后更新：2026年9月17日"},"whatThisCovers":{"title":"本政策涵盖范围","body":"本政策涵盖 openwhispr.com 上的 Cookie 及类似技术，是对我们隐私政策的补充。OpenWhispr 桌面应用默认在本地处理音频，相关说明见隐私政策。"},"ourApproach":{"title":"我们的做法","body":"OpenWhispr 坚持隐私优先。我们采用 Google Consent Mode（高级模式）：Google 的分析和广告代码会在每个页面加载，但它们仅会根据您所在地区和您的选择来设置 Cookie（详见下方分地区说明）。在尚未允许 Cookie 时，这些代码仍会向 Google 发送匿名、无 Cookie 的信号——包括您的 IP 地址以及您已拒绝同意这一情况——仅用于汇总性、不可识别个人身份的统计（该数据会传送至位于美国的 Google）。我们通过 Vercel 收集的使用指标不依赖 Cookie，也不会识别您的身份，因此无论您是否同意都会运行。您可随时通过页脚的 Cookie 设置更改您的选择。"},"categories":{"title":"Cookie 类别","headerCategory":"类别","headerWhatItDoes":"作用","headerConsent":"是否需要同意？","rows":[{"category":"严格必要","whatItDoes":"登录、会话安全、识别您所在地区，以及记住您的 Cookie 选择。","consent":"否——始终启用"},{"category":"分析","whatItDoes":"通过 Google Analytics 4 进行汇总性使用情况统计。","consent":"在欧洲经济区/英国需同意；在美国默认开启（可选择退出）"},{"category":"营销","whatItDoes":"通过 Google Ads 和 Meta 衡量我们的广告效果、记住您的访问来源，并通过 RB2B 识别美国企业访客，以便我们跟进。","consent":"在欧洲经济区/英国需同意；在美国默认开启（可选择退出，GPC 信号会自动为您退出）"}]},"cookies":{"title":"我们可能设置的 Cookie","headerCookie":"Cookie","headerProvider":"提供方","headerPurpose":"用途","headerType":"类型","headerDuration":"有效期","rows":[{"cookie":"ow_consent","provider":"OpenWhispr","purpose":"记录您的 Cookie 选择和政策版本的副本（在代码加载前读取）","type":"第一方","duration":"12 个月"},{"cookie":"ow-geo","provider":"OpenWhispr","purpose":"记录您所在地区，以应用正确的默认设置和提示","type":"第一方","duration":"1 天"},{"cookie":"会话 Cookie","provider":"OpenWhispr（Better Auth）","purpose":"保持您的登录状态","type":"第一方","duration":"每个会话"},{"cookie":"ow_attribution","provider":"OpenWhispr","purpose":"记录访问来源（广告点击 ID、广告系列标签、引荐来源），以便了解注册来自哪个渠道。仅在同意营销时使用。","type":"第一方","duration":"90 天"},{"cookie":"_ga, _ga_","provider":"Google（Analytics 4）","purpose":"区分重复访问，以便统计使用情况","type":"第三方","duration":"最长 24 个月"},{"cookie":"_gcl_au","provider":"Google Ads","purpose":"转化与广告效果衡量","type":"第三方","duration":"最长 90 天"},{"cookie":"_fbc, _fbp","provider":"Meta","purpose":"用于衡量我们 Meta 广告效果的广告点击 ID 和浏览器 ID（仅在投放 Meta 广告时）","type":"第三方","duration":"最长 90 天"},{"cookie":"RB2B 标识符","provider":"RB2B","purpose":"将美国企业访客与其公开职业资料（姓名、公司、工作电子邮件）进行匹配，以便我们联系。仅限美国访客。仅在同意营销时使用。","type":"第三方","duration":"由 RB2B 的脚本设置；请参阅 RB2B 的隐私政策"},{"cookie":"dub_id, ow_affiliate_consent","provider":"OpenWhispr / Dub","purpose":"记住向你推荐 OpenWhispr 的创作者，并通过 Dub 将符合条件的付款与其佣金关联。仅在你允许营销时使用。","type":"第一方","duration":"90 天"}],"note":"Google Tag Manager 本身不会设置任何 Cookie。只有在 Google、Meta 或 RB2B 标签触发后，上述第三方 Cookie 才会出现——在欧洲经济区/英国，须在您同意后触发；在美国等选择退出地区，则默认触发（除非您选择退出）。此外，RB2B 仅针对美国访客触发，绝不会在其他地区触发。即使关闭 Cookie，Google 的标签仍可能发送匿名、无 Cookie 的信号（包括您的 IP 和同意状态），且仅用于汇总衡量。我们会将您的选择保存在 ow_consent Cookie 中（在标签加载前读取的精简副本），并将完整内容——您的选择、随机的非识别 ID、时间和地区——保存在浏览器的本地存储中。我们的服务器不会保留副本，因此清除浏览器数据后会再次询问您。"},"managing":{"title":"管理您的选择","body":"您可随时使用 Cookie 设置来接受、拒绝或细化您的选择；您的浏览器也可以拦截或删除 Cookie。如果您的浏览器发送 Global Privacy Control（GPC）信号，我们会将其视为退出广告类 Cookie——如需同时关闭分析类 Cookie，请使用 Cookie 设置。","button":"管理 Cookie 偏好"},"byRegion":{"title":"分地区说明","items":["欧洲经济区和英国：在您同意之前，不会设置任何分析或广告类 Cookie，且您可随时撤回同意。（代码会先加载以发送匿名、无 Cookie 的信号——详见“我们的做法”——但在您同意之前不会在您的设备上存储任何内容。）","美国：分析和广告类 Cookie 默认开启；您可随时选择退出。请使用页脚的您的隐私选择，或发送 Global Privacy Control（GPC）信号——GPC 会关闭广告类 Cookie；如需同时关闭分析类 Cookie，请使用 Cookie 设置。","其他地区：我们会向您显示提示，且您可随时调整偏好。"]},"changes":{"title":"本政策的变更","body":"当我们的 Cookie 使用方式发生变化时，我们会更新本页面及“最后更新”日期。有疑问？请发送电子邮件至 support@openwhispr.com。"},"legalNotice":"本文件的英文版本为权威版本；如任何译文与英文版本不一致，以英文版本为准。本地化版本仅为方便阅读而提供，在作为依据前应经过法律顾问审阅。"},"changelog":{"metadata":{"title":"更新日志 — OpenWhispr 新功能速览","description":"查看 OpenWhispr 发布的每一个版本、改进和缺陷修复。内容自动同步自 GitHub 发布记录。","keywords":["OpenWhispr 更新日志","OpenWhispr 版本发布","OpenWhispr 更新","OpenWhispr 新功能","OpenWhispr 发布说明"]},"breadcrumb":{"home":"首页","changelog":"更新日志"},"hero":{"title":"更新日志","subtitle":"每一个版本、改进和缺陷修复 —— 直接同步自我们的 GitHub 发布记录。"},"empty":{"prefix":"暂无发布记录。请稍后再来查看，或前往以下平台查看发布记录：","githubLink":"GitHub","suffix":"。"},"prerelease":"预发布版","viewOnGitHub":"在 GitHub 上查看 →"},"home":{"metadata":{"title":"OpenWhispr：免费开源语音输入软件","description":"OpenWhispr 是可靠的开源语音输入软件。借助快速、准确的 AI 语音转文字，轻松完成听写、转录和会议笔记记录。","keywords":["语音转文字","听写应用","语音识别","Whisper AI","nvidia parakeet","开源听写","语音听写","AI 转录"]},"breadcrumb":{"home":"首页"},"video":{"name":"OpenWhispr 演示 — AI 语音听写","description":"看看 OpenWhispr 的实际效果:按下快捷键,自然地说话,你的文字便会即时出现在任何应用中。由 OpenAI Whisper 和 NVIDIA Parakeet 驱动。"},"schema":{"name":"OpenWhispr","description":"开源 AI 语音听写。借助 OpenAI Whisper 和 NVIDIA Parakeet,即时将语音转为文字。私密、跨平台、免费。","offerDescription":"核心功能免费且开源。另有可选的 Pro 套餐。","featureList":["通过 OpenAI Whisper 和 NVIDIA Parakeet 实现本地语音转文字","支持 99 种以上语言","全系统通用的按键说话快捷键","用于文本整理的 AI 智能体模式","完全开源","跨平台(macOS、Windows、Linux、iOS)","支持离线使用"]},"hero":{"title":"你的语速比打字快 3 倍。","subtitle":"开源语音转文字助手，隐私保护融入设计。\n在任何应用中都能更快输入。","availability":"支持 macOS、Windows、Linux 和 iOS"},"privacy":{"eyebrow":"开源，隐私至上","title":"您的数据始终属于您。","subtitle":"OpenWhispr 是开源软件。我们这样处理您的数据：","principles":[{"headline":"您的语音绝不会传到我们这里","description":"使用本地听写时，开放模型会直接在您的电脑上将语音转成文字。任何内容都不会发送给我们或其他任何人。"},{"headline":"云端是您的选择，而非默认设置","description":"开启云端后，音频会被发送去转写；文字返回的那一刻，音频就会被删除。我们绝不会保留录音。"},{"headline":"我们不会用您说的话训练模型","description":"您的转写内容不会用于训练模型或改进产品。如果将来发生变化，也一定会由您主动选择加入，并事先征得您的同意。"},{"headline":"别只听我们怎么说","description":"OpenWhispr 在 GitHub 上以 MIT 许可证开源。您可以亲自核实本页面上的每一项说明。"}],"visuals":{"retention":"数据保留","training":"数据训练","offByDefault":"默认关闭","public":"公开","github":"GitHub","contributors":"+214 位贡献者"}},"socialProof":{"title":"用户怎么说","description":"来自开发者、写作者和团队使用 OpenWhispr 的真实反馈。"},"faqSection":{"title":"常见问题","description":"下载之前,大家通常想了解的内容。"},"cta":{"title":"免费试用","descriptionLead":"支持 macOS、Windows、Linux 和 iOS。","descriptionEmphasis":"开源,本地运行。"},"faqs":[{"question":"隐私性如何?会有人看到我的转录内容吗?","answer":"使用本地处理时,你的语音绝不离开设备 —— 100% 私密。即便使用云端处理,音频也是即处理即丢弃。我们绝不存储、出售你的数据,也不会用它训练模型。OpenWhispr 是开源的,你可以亲自验证这一点。"},{"question":"我能用自己的 API 密钥代替 OpenWhispr Cloud 吗?","answer":"当然可以。你可以自带 OpenAI、Anthropic、Google Gemini、Groq 或任何兼容服务商的密钥。你掌控自己的费用,数据关系直接面向服务商,除了密钥本身的限制外没有任何额外限制。"},{"question":"OpenWhispr 真的免费吗?","answer":"是的 —— 核心应用开源且永久免费。本地 AI 模型在你的设备上运行,毫无限制。自带 API 密钥同样没有任何限制。我们还提供自有的内置云服务 OpenWhispr Cloud —— 每个免费账户每周可享 2,000 字,无需 API 密钥,也无需任何设置。Pro 套餐($8/月)取消这一上限,并新增移动端、会议录制和邮件支持。"},{"question":"它支持哪些平台和应用?","answer":"OpenWhispr 可在 macOS、Windows 和 Linux 上运行。它适用于任何应用程序 —— Slack、VS Code、Chrome、Word、Notion、你的终端,凡是能接受文本输入的,统统支持。只需按下快捷键、开口说话,文字便会出现在光标所在之处。"},{"question":"支持哪些语言?","answer":"OpenWhispr 支持 100 多种语言,包括英语、西班牙语、法语、德语、中文、日语、阿拉伯语等等。你可以设定首选语言,也可以使用自动检测,在不同语言间随时切换。"},{"question":"本地处理和云端处理有什么区别?","answer":"本地处理直接在你的电脑上运行 AI 模型 —— 完全私密、可离线使用,但速度取决于你的硬件。云端处理则将音频发送到服务器进行转录 —— 在大多数机器上更快、更准确,但需要联网。"},{"question":"我该如何自定义快捷键?","answer":"打开设置,你可以将任意按键或组合键设为录音快捷键。按住录音和按键切换两种模式都受支持。在 Windows 上,Ctrl+Shift+Space 这类组合快捷键同样可用。"},{"question":"它真的是开源的吗?我能参与贡献吗?","answer":"百分之百。OpenWhispr 采用 MIT 许可 —— 个人和商业用途均免费。完整源代码都在 GitHub 上。我们欢迎社区提交贡献、反馈缺陷和提出功能建议。"}],"compliance":{"title":"合规与认证","subtitle":"安全团队会要求的一切，我们都已准备就绪。","certifications":[{"status":"合规","title":"HIPAA","description":"无论使用本地处理还是云端处理，我们都会妥善保护您的健康信息。"},{"status":"合规","title":"AICPA SOC 2 & ISO 27001","description":"已通过独立验证。无论安全审查需要查看什么，我们都已备妥。","standards":["SOC 2 TYPE II","ISO 27001"]}]}},"homeBento":{"section":{"title":"你能得到什么","subtitle":"语音听写、AI 整理、离线模式,以及 100 多种语言。"},"cards":{"models":{"headline":"你的电脑,你的模型。","description":"运行本地语音转文字模型,或自带 API 密钥。全部免费。"},"voiceInstructions":{"headline":"用语音下达指令。","description":"说一句「帮我整理一下」或「给 Mike 起草一封邮件」,它便会照办无误。"},"dictionary":{"headline":"它会记住你的用词。","description":"AI 可能听漏的人名、行话和术语 —— 都能从你的修正中自动学习。"},"languages":{"headline":"说任何语言。","descriptionPrefix":"100 多种语言,支持自动检测。对话中途即可切换,","descriptionSuffix":"。"},"apps":{"headline":"适配你所有的应用。","description":"ChatGPT、Claude、Cursor、Slack、Docs —— 只要能接受文本,就能用。"},"offline":{"headline":"可离线使用。","description":"没有 wifi,也没问题。OpenWhispr 离线运行,让你随时随地都能转录。"}},"modelPicker":{"speechToText":"语音转文字","aiModels":"AI 模型","cloud":"云端","local":"本地","custom":"自定义","endpointUrl":"端点 URL","apiKey":"API 密钥","optional":"(可选)","modelName":"模型名称","model":"模型","autoDiscovered":"已从端点自动发现","openAiCompatibleNote":"兼容任何与 OpenAI 兼容的 API —— Ollama、LM Studio、vLLM 等等。","descriptions":{"noApiKeyNeeded":"无需 API 密钥","mostAccurate":"最准确","fastAndAccurate":"快速又准确","originalWhisper":"原版 Whisper","highAccuracy":"高准确度","realtime216x":"216 倍实时速度","multilingual":"多语言","fastest":"最快","fast":"快速","balanced":"均衡","accurate":"准确","best":"最佳","highestQuality":"最高质量","fastReasoning":"快速推理","latestFlagship":"最新旗舰","fastAndEfficient":"快速又高效","context1M":"100 万上下文","mostCapable":"能力最强","flagshipReasoning":"旗舰推理","ultraFast":"超快速","context131K":"13.1 万上下文","throughput500":"500 词元/秒","throughput280":"280 词元/秒","recommended":"推荐","compact":"精简","lightweight":"轻量","powerful":"强大"}},"agentMode":{"youSaidLabel":"你说","youSaid":"\"嘿 OpenWhispr,帮我整理一下这封邮件 —— 我要告诉 mike 会议改到周四下午三点,另外提醒他准备好幻灯片\"","outputLabel":"输出","outputGreeting":"你好 Mike,","outputBody":"简单同步一下 —— 我们的会议已改到周四下午三点。另外提醒一下,记得把幻灯片准备好。","outputClosing":"谢谢!"},"dictionary":{"addPlaceholder":"添加一个词或短语…","autoLearns":"从你的修正中自动学习"},"languages":{"more":"另有 100 多种"}},"homeSocial":{"title":"用户怎么说","description":"来自开发者、写作者和团队的 OpenWhispr 真实使用反馈。","translatedFromJapanese":"译自日语"},"pricing":{"metadata":{"title":"价格：提供永久免费方案 - OpenWhispr","description":"OpenWhispr 的本地语音输入和转录永久免费。需要云端转录、更多会议时长和团队工具时，可随时升级。","keywords":["openwhispr 价格","openwhispr 价格方案","语音转文字价格","听写应用价格","免费听写应用","语音转文字方案"],"ogImageAlt":"OpenWhispr 价格方案"},"breadcrumb":{"home":"首页","pricing":"价格"},"hero":{"title":"永久免费。需要更多功能时再升级。","subtitle":"本地听写无限量，人人免费。付费方案另含无限量云端转录、更多会议录制时长以及团队协作工具。"},"billingToggle":{"ariaLabel":"计费周期","monthly":"按月","yearly":"年度的","savings":"免费 2 个月"},"perUserMonth":"/用户/月","perMonthShort":"/月","billedYearly":"$${total}/用户按年计费","billedMonthly":"按月计费","moreInfo":"了解更多","cta":{"download":"下载","getStarted":"立即开始","contactSales":"联系销售"},"tiers":{"free":{"subtitle":"适合所有人","features":["## OpenWhispr Cloud","每周 2,000 词转录额度","每月 5 小时会议录制","## 永久免费","无限量本地 AI 模型 {i}","使用自己的 API 密钥，无限量云端转录","支持 100 多种语言","自定义词典","零数据留存","社区支持"]},"pro":{"subtitle":"适合个人与团队","includesPrefix":"包含全部 Free 功能 +","features":["无限量云端转录","智能体模式","每月 20 小时会议录制","多设备同步","个人 API 访问权限","在所有应用中集成 MCP","iPhone 和 iPad 应用","邮件支持"]},"business":{"subtitle":"适合个人与团队","badge":"最受欢迎","includesPrefix":"包含全部 Pro 功能 +","features":["无限量会议录制","与你的数据对话","优先支持"]},"enterprise":{"subtitle":"面向高级安全与支持需求","priceDisplay":"定制","includesPrefix":"包含全部 Business 功能 +","features":["团队功能与管理后台","SSO、SAML 与 SCIM","审计日志与留存控制","专属支持"]}},"featureInfo":{"Unlimited local AI models":{"title":"内置本地模型","sections":[{"label":"语音转文字","items":["OpenAI Whisper（全部规格）","NVIDIA Parakeet"]},{"label":"AI 写作助手","items":["Qwen","Llama","Mistral","Gemma","GPT-OSS"]}]}},"compareSection":{"title":"对比所有方案","description":"逐项功能并列对比。本地听写在每个方案中都无限量且免费。"},"enterpriseBanner":{"title":"需要 SSO、合规或管理控制？","description":"我们的 Enterprise 方案新增 SSO 与 SAML、SCIM 配置、审计日志、组织级留存控制、定制数据处理协议以及专属支持。","cta":"联系我们"},"socialProof":{"title":"深受爱说话胜过爱打字的人喜爱","description":"来自开发者、写作者和团队的 OpenWhispr 真实使用反馈。"},"faqSection":{"title":"价格相关问题","description":"关于方案与计费的直接解答。"},"comparison":{"included":"包含","notIncluded":"不包含","selectPlanAriaLabel":"选择一个方案进行对比","caption":"逐项功能对比 OpenWhispr 各方案","plansHeader":"方案","aboutLabel":"关于 {label}","groups":[{"title":"语音转文字","rows":[{"label":"本地 AI 模型","tooltip":"完全在你的设备上运行，无需联网——在每个方案中均无限量且免费。","values":["无限量","无限量","无限量","无限量"]},{"label":"OpenWhispr Cloud 转录","tooltip":"音频发送至 OpenWhispr Cloud，无需下载本地模型即可获得顶级准确度与速度，且转录后绝不留存。","values":["每周 2,000 词","无限量","无限量","无限量"]},{"label":"使用自己的 API 密钥","tooltip":"使用你自己的 OpenAI、Anthropic、Gemini 或 Groq 密钥进行无限量云端转录，不受 OpenWhispr 任何限制。","values":["","","",""]},{"label":"语言","values":["100+","100+","100+","100+"]},{"label":"自定义词典","tooltip":"教会 OpenWhispr 你的人名、行话和缩写，让转录更精准。","values":["","","",""]}]},{"title":"会议与工作流","rows":[{"label":"会议录制","values":["每月 5 小时","每月 20 小时","无限量","无限量"]},{"label":"发言人标注","tooltip":"自动识别并标注谁说了什么——无论是会议还是上传的音频。","values":["","","",""]},{"label":"多设备同步","values":["","","",""]},{"label":"移动配套应用","tooltip":"用手机进行听写。","values":["","","",""]},{"label":"API 与 MCP 访问","tooltip":"将 OpenWhispr 接入你自己的工具和 Agent 工作流。","values":["","","",""]},{"label":"命令行界面","tooltip":"从终端脚本化操作 OpenWhispr。本地 CLI 在每个方案中均可驱动桌面应用；付费方案另含由 OpenWhispr Cloud API 支持的远程 CLI。","values":["本地","本地与远程","本地与远程","本地与远程"]},{"label":"智能体模式","tooltip":"让 AI 自动对你的转录内容执行操作。","values":["","","",""]},{"label":"与你的数据对话","tooltip":"跨笔记、文件和转录内容提问。","values":["","","",""]}]},{"title":"隐私与安全","rows":[{"label":"零数据留存","tooltip":"云端音频转录后立即丢弃。我们绝不存储、出售你的数据，也绝不用其训练模型。","values":["","","",""]},{"label":"开源（MIT）","values":["","","",""]},{"label":"SSO 与 SAML","values":["","","",""]},{"label":"SCIM 配置","tooltip":"从你的身份提供商自动配置和注销团队成员。","values":["","","",""]},{"label":"审计日志","tooltip":"可导出的账户与管理活动日志，用于安全与合规审查。","values":["","","",""]},{"label":"组织级留存控制","tooltip":"由管理员配置的数据留存与自动删除策略，覆盖整个组织。","values":["","","",""]},{"label":"合规与 DPA","tooltip":"为你的组织提供定制数据处理协议与合规审查。","values":["","","",""]}]},{"title":"团队与支持","rows":[{"label":"团队管理与管理后台","values":["","","基础版","高级版"]},{"label":"支持","values":["社区","邮件","优先","专属"]},{"label":"支持平台","values":["macOS · Windows · Linux","macOS · Windows · Linux · iOS","macOS · Windows · Linux · iOS","macOS · Windows · Linux · iOS"]}]}]},"faqs":[{"question":"免费方案真的永久免费吗？","answer":"是的。在 macOS、Windows 和 Linux 上以 100 多种语言进行无限量本地听写，价格为 $0——无需信用卡，也没有试用倒计时。Free 方案还包含每周 2,000 个 OpenWhispr Cloud 词额度。付费方案是可选的升级项，而非倒计时。"},{"question":"本地转录和云端转录有什么区别？","answer":"本地转录在你自己的设备上运行——完全私密、可离线使用，且无限量免费。云端转录在 OpenWhispr Cloud 上运行，可在任何硬件上获得顶级准确度。Free 方案每周含 2,000 个云端词额度；Pro、Business 和 Enterprise 则不限量。"},{"question":"可以按年付费吗？","answer":"可以。Pro 为每月 $8，或每年 $80（免费两个月）。Business 为每用户每月 $16，或每用户每年 $160。按年计费平摊到每月更划算。"},{"question":"你们为学生、军人或非营利组织提供折扣吗？","answer":"是的。如果 OpenWhispr 是你学业、课堂教学、服役经历、非营利组织工作的一部分，或是你日常使用的辅助工具，你就可以享受 12 个月 6 折优惠。将简单的证明材料发送至 support@openwhispr.com，我们会把你的专属折扣码发给你。","linkText":"在帮助中心查看详情","linkHref":"https://docs.openwhispr.com/help/account/discounts-and-promo-codes"},{"question":"开始使用需要信用卡吗？","answer":"不需要。下载即可使用 Free 方案，无需信用卡，也没有账户门槛。只有当你升级到 Pro 或 Business 时，才需要填写付款信息。"},{"question":"可以随时更改或取消方案吗？","answer":"可以。随时升级、降级或取消。如果取消，你仍可在 Free 方案上享受无限量本地听写——你的私密本地转录永不停摆。"},{"question":"可以用自己的 API 密钥代替付费吗？","answer":"当然可以。在任何方案中，你都可以使用自己的 OpenAI、Anthropic、Gemini 或 Groq 密钥进行无限量云端转录，不受 OpenWhispr 任何限制——你掌控成本，数据关系也直接保留在提供商一方。"},{"question":"你们会用我的语音数据训练 AI 模型吗？","answer":"不会。本地转录绝不离开你的设备，云端音频转录后立即丢弃。我们绝不存储、出售你的数据，也绝不用其训练——而且由于 OpenWhispr 是开源的，你可以自行验证。"},{"question":"你们如何处理安全与合规？","answer":"音频在传输过程中加密，并在每个方案中以零数据留存的方式处理。Enterprise 另增 SSO 与 SAML、SCIM 配置、审计日志、组织级留存控制以及定制数据处理协议。详情请访问我们的信任中心 trust.openwhispr.com。"},{"question":"Enterprise 方案新增哪些功能？","answer":"Enterprise 新增 SSO 与 SAML、SCIM 配置、审计日志、组织级数据留存控制、定制数据处理协议、团队管理与管理控制，以及专属支持——并为你的组织量身定制托管式入门引导。请联系我们进行配置。"}]},"blogHub":{"metadata":{"title":"语音听写软件博客：指南与技巧 | OpenWhispr","description":"学习语音听写、开源语音转文字和 AI 会议笔记。提供 Whisper、说话人分离及隐私优先工具的实用指南。","keywords":["语音听写博客","语音转文字指南","Whisper AI 文章","听写技巧","语音技术博客","openwhispr 博客"]},"breadcrumb":{"home":"首页","blog":"博客"},"hero":{"title":"博客","subtitle":"深入解读语音听写、OpenAI Whisper、NVIDIA Parakeet 与语音转文字技术的指南、对比和深度文章。"},"readArticle":"阅读全文 →","posts":[{"title":"符合 HIPAA 的 AI 会议记录工具：BAA 究竟覆盖什么","description":"哪些 AI 会议记录工具会签署业务伙伴协议，哪些把 AI 功能排除在外，以及本地转录为何改变整个判断。","date":"2026年8月11日","tag":"合规"},{"title":"Google Docs Voice Typing：完整指南","description":"如何使用 Google Docs 语音输入：设置、命令、修复、限制以及如何使用 OpenWhispr 在任何应用程序中进行口述。","date":"2026 年 6 月 26 日","tag":"指导"},{"title":"如何做好会议记录：一份实用指南","description":"一套真正能用起来的会议记录方法——记下决策、行动事项和待解决的问题，快速整理并妥善分享。附赠一份可直接复制粘贴的免费模板。","date":"2026年6月1日","tag":"指南"},{"title":"本地说话人分离：我们如何打造 100% 隐私的会议记录工具","description":"全面拆解 OpenWhispr 本地说话人分离流程的技术细节——pyannote 分段、CAM++ 嵌入、sherpa-onnx 以及设备端声纹档案。音频数据绝不离开你的设备。","date":"2026年4月15日","tag":"技术"},{"title":"2026 年 Windows 上最好用的听写工具","description":"全面对比 Windows 上每一款可用的听写与语音转文字工具——从内置的 Win+H 到 AI 驱动的工具，客观比较。","date":"2026年3月15日","tag":"对比"},{"title":"2026 年 Linux 上最好用的听写工具","description":"全面对比 Linux 上每一款可用的听写与语音转文字工具——客观比较。","date":"2026年2月14日","tag":"对比"},{"title":"为什么选择本地 AI：你的语音数据不该离开你的设备","description":"监管态势、硬件能力，以及在本地处理语音数据的隐私理由——附真实的数据泄露统计与合规要求。","date":"2026年2月12日","tag":"隐私"},{"title":"为什么开源对语音工具至关重要","description":"为什么语音软件应当可审计——真实的隐私事件、开放模型生态（OpenAI Whisper、NVIDIA Parakeet），以及挑选开源语音工具时该关注什么。","date":"2026年2月10日","tag":"思想洞察"},{"title":"本地转录 vs 云端转录：隐私、速度与准确度全面对比","description":"对本地与云端转录的客观对比——真实定价数据、隐私政策、准确度基准测试，以及各自适用的场景。","date":"2026年2月8日","tag":"指南"},{"title":"Whisper 模型尺寸详解：Tiny、Base、Small、Medium 与 Large 对比","description":"你该用哪个 Whisper 模型？全部 7 个模型的完整解析，涵盖参数量、显存需求、速度基准测试与准确度权衡。","date":"2026年2月5日","tag":"技术"},{"title":"Whisper AI 工作原理：完整指南","description":"深入剖析 Whisper 语音识别模型的技术细节——架构、训练数据、模型尺寸，以及它究竟如何转录语音。","date":"2026年2月3日","tag":"技术"}]},"blogWindowsTools":{"metadata":{"title":"2026 年最佳 Windows 语音听写工具 | OpenWhispr","description":"真实对比 2026 年最佳 Windows 语音听写软件：OpenWhispr、Dragon、Wispr Flow 等，帮你找到免费、隐私且开源的选择。","keywords":["Windows 最佳听写应用","Windows 语音转文字","Windows 语音识别","Windows 听写软件","Windows 语音输入","2026 Windows 语音转文字","whisper windows","dragon 听写替代方案","Windows 语音听写"],"ogTitle":"2026 年 Windows 最佳听写工具盘点","ogDescription":"全面对比 2026 年适用于 Windows 的听写与语音转文字工具。"},"schema":{"title":"2026 年 Windows 最佳听写工具盘点","description":"全面对比 2026 年适用于 Windows 的听写与语音转文字工具。"},"breadcrumb":{"home":"首页","blog":"博客","article":"Windows 最佳听写工具"},"layout":{"title":"2026 年 Windows 最佳听写工具盘点","subtitle":"Windows 的听写选择从未如此丰富——从内置的 Win+H，到边说边改写的 AI 工具，应有尽有。这里是 2026 年所有值得认真考虑的方案，诚实对比。","date":"2026 年 3 月 15 日","tag":"博客"},"sections":[{"id":"direct-answer","title":"直接答案"},{"id":"fact-check","title":"事实核查速览"},{"id":"openwhispr","title":"OpenWhispr"},{"id":"windows-voice-typing","title":"Windows Voice Typing"},{"id":"dragon-professional","title":"Dragon Professional"},{"id":"wispr-flow","title":"Wispr Flow"},{"id":"typeless","title":"Typeless"},{"id":"aqua-voice","title":"Aqua Voice"},{"id":"microsoft-365-dictation","title":"Microsoft 365 Dictation"},{"id":"talon-voice","title":"Talon Voice"},{"id":"google-docs-voice-typing","title":"Google Docs Voice Typing"},{"id":"comparison","title":"横向对比"},{"id":"recommendations","title":"我们的推荐"},{"id":"sources","title":"参考来源"}],"directAnswer":{"leadStrong":"2026 年 Windows 上最佳的听写工具，取决于你的需求。","bodyBeforeLink":"对于大多数想要精致、隐私优先、并带 AI 文本整理体验的用户来说，","bodyAfterLink":"提供了最完整的方案：由 whisper.cpp 驱动的按键说话听写、全系统级热键支持、99+ 种语言，以及可选的 AI 排版——全部开源且免费。如果想要零安装的起步方案，Windows Voice Typing（Win+H）已经内置其中。如果预算不是问题，且需要针对英语的最高训练精度，Dragon Professional 仍是传统标杆，售价 $500–$669。希望使用 AI 自动编辑的云端优先用户，可以考虑 Wispr Flow 或 Typeless。而对于需要完全免手操作电脑的开发者，Talon Voice 则自成一档。","lastUpdated":"最后更新：2026 年 3 月。工具可用性与功能声明的每一项主要论点均至少经过两个来源交叉核实。"},"factCheck":{"title":"事实核查速览（双来源）","item1Lead":"OpenWhispr 通过 whisper.cpp 在本地运行 Whisper：","item1Body":"功能集与本地运行时均有公开文档记载。","item1Link1":"openwhispr.com","item1Link2":"whisper.cpp","item2Lead":"Windows Voice Typing 需要联网：","item2Body":"转录通过 Azure Speech Services 在云端完成。Copilot+ PC 增加了本地 AI 语法纠正（\"Fluid Dictation\"），但核心转录仍在云端。","item2Link1":"Microsoft 支持","item2Link2":"Windows Central","item3Lead":"Dragon Professional 仍在销售，但已进入维护模式：","item3Body":"Microsoft 于 2022 年 3 月完成了对 Nuance 价值 197 亿美元的收购，并将 Nuance 技术重心放在 Azure 和 Microsoft 365 上，而非 Dragon。","item3Link1":"Nuance Dragon","item3Link2":"Wikipedia","item4Lead":"云端优先工具以隐私换取便利：","item4Body":"Wispr Flow、Typeless 和 Aqua Voice 都会将音频发送到云端服务器进行处理。","item4Link1":"Wispr Flow","item4Link2":"Typeless","item4Link3":"Aqua Voice","item5Lead":"OpenWhispr 的定位：","item5Body":"OpenWhispr 是 Windows 上唯一一款开源、可离线、带 AI 整理的全系统级听写工具。","item5Link1":"OpenWhispr","item5Link2":"Whisper 上游项目"},"labels":{"pros":"优点","cons":"缺点","pricing":"价格：","bestFor":"最适合："},"comparisonTable":{"title":"横向对比","headerTool":"工具","headerOpenSource":"开源","headerOffline":"离线","headerLanguages":"语言","headerAiCleanup":"AI 整理","headerSystemWide":"全系统级","headerPrice":"价格","rows":[{"tool":"OpenWhispr","openSource":"full","offline":"full","languages":"99+","aiCleanup":"full","systemWide":"full","price":"免费（Pro $8/月）"},{"tool":"Windows Voice Typing","openSource":"none","offline":"none","languages":"44","aiCleanup":"partial","systemWide":"full","price":"免费（随 Windows）"},{"tool":"Dragon Professional","openSource":"none","offline":"full","languages":"6","aiCleanup":"none","systemWide":"full","price":"$$500–$669 一次性"},{"tool":"Wispr Flow","openSource":"none","offline":"none","languages":"100+","aiCleanup":"full","systemWide":"full","price":"免费版 / $15/月"},{"tool":"Typeless","openSource":"none","offline":"none","languages":"100+","aiCleanup":"full","systemWide":"full","price":"免费版 / $12/月"},{"tool":"Aqua Voice","openSource":"none","offline":"none","languages":"49","aiCleanup":"full","systemWide":"full","price":"免费版 / $8/月"},{"tool":"Microsoft 365 Dictation","openSource":"none","offline":"none","languages":"~55","aiCleanup":"partial","systemWide":"none","price":"随 M365 提供"},{"tool":"Talon Voice","openSource":"partial","offline":"full","languages":"英语","aiCleanup":"none","systemWide":"full","price":"免费 / Patreon $25/月"},{"tool":"Google Docs Voice Typing","openSource":"none","offline":"none","languages":"100+","aiCleanup":"none","systemWide":"none","price":"免费"}]},"openwhispr":{"title":"1. OpenWhispr","intro":"OpenWhispr 是一款开源、跨平台的桌面听写应用，基于 Electron 和 whisper.cpp 构建。它提供全系统级的按键说话语音转文字——按住热键、开口说话、松开按键，转录后的文本就会出现在光标所在的任何位置。它还包含一个 AI 智能体模式，可借助 LLM（OpenAI、Anthropic、Google Gemini 或本地模型）整理、排版并重组你听写的文本。在 Windows 上，它以标准 .exe（搭配 NSIS 安装程序）安装，并在 x64 硬件上原生运行。","pros":["精致的图形界面，配全系统级按键说话热键","AI 智能体模式，自动整理与排版文本","通过 Whisper 与 NVIDIA Parakeet 模型支持 99+ 种语言","完全开源——可从源码构建，也可使用预编译二进制文件","搭配本地模型完全离线——音频绝不离开你的设备"],"cons":["基于 Electron——资源占用比原生应用更高","AI 整理功能需要 API 密钥或 Pro 订阅","较大的 Whisper 模型需要不错的硬件才能实现实时性能"],"pricingValue":"核心功能免费，可选 Pro $8/月","bestForValue":"想要完整、私密的听写体验，并带 AI 文本整理的用户"},"windowsVoiceTyping":{"title":"2. Windows Voice Typing（Win+H）","intro":"Windows Voice Typing 是 Microsoft 内置的听写功能，按 Win+H 即可在全系统范围内调用。它使用基于云端的 Azure Speech Services 进行转录，并包含自动标点。在 Copilot+ PC 上，\"Fluid Dictation\" 增加了由本地 AI 驱动的语法纠正和填充词去除。旧版的 Windows Speech Recognition 已被 Voice Access 取代，后者同时提供听写与完整的免手操作 PC 控制。","pros":["零成本、零配置——内置于每台 Windows 10/11 PC","全系统级——在整个操作系统的任意文本框中都可用","支持自动标点和 44 种语言","Copilot+ PC 上的 Fluid Dictation 可纠正语法并去除填充词"],"cons":["需要联网——所有音频都会发送到 Microsoft 的云端","对清晰英语语音的精度普遍报告为 85–90%","无 AI 文本重排或整理（Copilot+ PC 除外）","非开源——专有云服务"],"pricingValue":"免费（随 Windows 提供）","bestForValue":"不想安装任何东西时，进行快速、随手的听写"},"dragon":{"title":"3. Dragon Professional v16","intro":"Dragon Professional 是 Windows 听写领域的传统重量级选手。它最初名为 Dragon NaturallySpeaking，由 Nuance 开发（Nuance 已被 Microsoft 以 197 亿美元的交易收购，该交易于 2022 年 3 月完成）。版本 16 是最新发布版，可完全离线运行，配合训练好的用户配置文件，英语精度可达 99%+。然而 Dragon 实际上已进入维护模式——Microsoft 尚未宣布新版本，并已将 Nuance 技术重心转向 Azure 和 Microsoft 365。","pros":["Nuance 声称训练好的英语配置文件精度高达 99%","完全离线——识别完全在设备本地运行","面向工作流的深度语音命令自定义","法律和医疗专业人士的行业标准"],"cons":["极其昂贵——视零售商而定为 $500–$669","实际上处于维护模式——没有新的开发","仅支持 6 种语言（英语、德语、法语、西班牙语、荷兰语、意大利语）","仅限 Windows，不支持 Mac 或 Linux"],"pricingValue":"$$500–$669 一次性（升级约 $350）","bestForValue":"需要离线、最高英语精度的法律和医疗专业人士","assessmentLead":"诚实评价：","assessmentBody":"Dragon 数十年来一直是金标准，但其前景并不明朗。自收购 Nuance 以来，Microsoft 未对 Dragon 进行投入。售价 $500–$669，却投资于一款可能不会获得实质性更新的软件。在大多数使用场景下，如今基于 Whisper 的现代工具已能以极低的成本达到甚至超越 Dragon 的精度。"},"wisprFlow":{"title":"4. Wispr Flow","intro":"Wispr Flow 是一款基于云端的 AI 听写工具，可在 Windows、macOS、iOS 和 Android 上使用。它在 AI 后处理方面表现出色——自动去除填充词、纠正语法，并根据你正在听写的应用调整语气（在 Slack 中更随意，在邮件客户端中更正式）。它支持 100+ 种语言并自动检测，还包含一个用于语音触发快捷方式的片段库。","pros":["出色的 AI 自动编辑，支持按应用调整语气","100+ 种语言并自动检测","跨平台（Windows、macOS、iOS、Android）","所有套餐均符合 HIPAA 资格"],"cons":["无离线模式——没有网络就完全无法使用","所有音频都会发送到云端服务器进行处理","价格偏贵，$15/月（按年付为 $12/月）","免费版限每周 2,000 字"],"pricingValue":"免费（每周 2,000 字），Pro $15/月","bestForValue":"想要最佳 AI 自动编辑，且不介意云端处理的用户"},"typeless":{"title":"5. Typeless","intro":"Typeless 是一款基于云端的听写工具，可在 Windows、macOS、iOS 和 Android 上使用。它专注于智能文本转换——去除填充词、消除重复、自动排版列表和要点，并随时间适应你的个人写作风格。它还支持语言间实时翻译，并可在 60+ 款应用中使用，包括 Notion、Slack、Gmail、VS Code 和 Obsidian。","pros":["慷慨的免费版——每周 4,000 字（约 16,000/月）","出色的 AI 自动编辑，可适应个人风格","100+ 种语言间实时翻译","在 60+ 款应用中全系统级可用"],"cons":["无离线模式——所有转录都需要联网","所有音频都会离开你的设备进行云端处理","若非按年付费，Pro 套餐为 $30/月","非开源——专有云服务"],"pricingValue":"免费（每周 4,000 字），Pro $12/月（按年）或 $30/月","bestForValue":"想要慷慨免费额度并配强大 AI 排版的用户"},"aquaVoice":{"title":"6. Aqua Voice","intro":"Aqua Voice 是一款专为开发者和技术写作者打造的云端听写工具。其专有的 Avalon 转录模型经过训练，在处理技术术语——变量名、缩写、API 名称和领域专有行话——方面优于通用语音引擎。它支持 49 种语言，在 Windows 和 macOS 上均可全系统级使用，启动时间不到 50 毫秒。","pros":["擅长处理技术词汇（代码、缩写、行话）","上下文感知排版，可根据目标应用调整输出","启动极快——延迟低于 50 毫秒","Pro 套餐实惠，$8/月"],"cons":["无离线模式——完全基于云端","语言数量（49 种）少于竞品","相对较新的产品，用户群较小","免费版限 1,000 字"],"pricingValue":"免费（1,000 字），Pro $8/月（按年）","bestForValue":"需要准确转录代码和行话的开发者与技术写作者"},"microsoft365":{"title":"7. Microsoft 365 Dictation","intro":"Microsoft 365 Dictation 内置于 Word、Excel、Outlook、PowerPoint 和 OneNote 的桌面版与网页版中。它使用 Microsoft 基于云端的语音识别，支持自动标点，并支持约 55 种语言（15 种完全支持，约 40 种处于预览阶段）。语音排版命令（\"加粗那段\"、\"换行\"、\"删除上一句\"）让你在 Office 应用内进行基本的免手编辑。它需要保持联网连接。桌面应用需要 Microsoft 365 订阅，不过在 Office 网页版中也可免费使用听写。","pros":["如果你已有 Microsoft 365，则已包含其中","自动标点和语音排版命令","约 55 种语言（15 种完全支持，约 40 种处于预览阶段）","在 Office 的桌面版和网页版中均可用"],"cons":["仅在 Microsoft 365 应用内可用——非全系统级","需要联网——仅支持云端转录","桌面应用需要 Microsoft 365 订阅（$9.99+/月）","无 AI 文本整理或填充词去除"],"pricingValue":"网页版免费，或随 M365（$9.99+/月）在桌面版中使用","bestForValue":"只需在 Office 应用内听写的 Microsoft 365 用户"},"talon":{"title":"8. Talon Voice","intro":"Talon Voice 是一套全面的免手输入系统，远超普通听写范畴。它让你完全用语音控制整台电脑——窗口管理、编程、应用切换、文件浏览——并可通过 Tobii 设备选配眼动追踪。Talon 内置自家的 Conformer 语音引擎，同时也支持 Dragon Professional（仅 Windows）作为备选引擎。它主要以命令为基础，而非散文式听写，因此非常适合需要彻底替代键盘和鼠标的开发者和无障碍用户。","pros":["完全免手控制电脑，而不只是听写","专为语音编程打造，配各编程语言专用命令","集成眼动追踪以替代鼠标","对无障碍而言不可或缺——对 RSI 患者堪称改变人生"],"cons":["学习曲线陡峭——需要记忆命令和一套语音字母表","核心引擎为专有（社区脚本为开源）","主要支持英语——多语言支持有限","并非为长篇散文式听写设计"],"pricingValue":"免费公开版，通过 Patreon 的测试版 $25/月","bestForValue":"需要完全免手操作电脑、语音编程或无障碍驱动输入的用户"},"googleDocs":{"title":"9. Google Docs Voice Typing","intro":"Google 内置的语音输入可通过 Google Docs（工具 > 语音输入）在 Chrome、Edge 或 Safari 中使用。它使用 Google 基于云端的语音识别，支持 100+ 种语言进行转录，不过排版语音命令仅在英语下有效。除浏览器外无需任何安装，使它成为门槛最低的选择——但同时也最受限，因为它无法在 Google Docs 之外工作。","pros":["零配置——打开浏览器即可开口说话","由 Google 语音模型驱动，精度不错","支持 100+ 种语言进行转录","拥有 Google 账户即可完全免费使用"],"cons":["仅在 Google Docs（Chrome、Edge、Safari）内可用——非全系统级","所有音频都会发送到 Google 的服务器——无离线模式","无 AI 整理——你必须手动说出标点","非开源——专有的 Google 服务"],"pricingValue":"免费","bestForValue":"在 Google Docs 内进行快速听写，且不需要全系统级输入或隐私时"},"recommendations":{"title":"我们的推荐","intro":"Windows 上并不存在唯一的\"最佳\"听写工具——这取决于你看重什么。以下是我们的拆解：","mostUsersTitle":"对大多数用户而言","mostUsersStrong":"OpenWhispr","mostUsersBody":"提供了最完整的方案：精致的图形界面、全系统级按键说话、AI 整理、Whisper 级精度，且完全开源。它默认离线工作，是这里唯一一款音频绝不离开你设备的工具。","zeroInstallTitle":"追求零安装的简单方案","zeroInstallStrong":"Windows Voice Typing（Win+H）","zeroInstallBody":"已经在你的 PC 上。按下热键即可开口说话。它在精度或功能上比不上专门工具，但内置之物的便利无可匹敌。","aiEditingTitle":"追求最佳 AI 自动编辑","aiEditingStrong1":"Wispr Flow","aiEditingAnd":"和","aiEditingStrong2":"Typeless","aiEditingBody":"在 AI 后处理方面领先——去除填充词、纠正语法、调整语气。代价是所有音频都会上传到云端。Typeless 的免费版更慷慨；Wispr Flow 具备按应用调整语气的能力。","developersTitle":"面向开发者","developersStrong":"Aqua Voice","developersBody":"专为技术词汇打造。如果你要听写代码注释、API 文档或技术规范，它的 Avalon 模型在处理行话方面优于通用引擎。","accuracyTitle":"追求离线最高英语精度","accuracyStrong":"Dragon Professional","accuracyBody":"仍能提供最高的英语训练精度，但售价 $500–$669 且前景未明。仅当你需要专门的词汇配置文件，并能接受一款传统产品时，才值得考虑。","accessibilityTitle":"面向无障碍与免手操作电脑","accessibilityStrong":"Talon Voice","accessibilityBody":"完全属于另一档。如果你需要替代键盘和鼠标——而不只是用听写来补充它们——Talon 就是值得学习的工具。投入的时间相当可观，但回报是颠覆性的。"},"sources":{"title":"参考来源与延伸阅读","openwhispr":"OpenWhispr","windowsVoiceTyping":"Windows Voice Typing（Microsoft 支持）","dragon":"Dragon Professional（Nuance）","wisprFlow":"Wispr Flow","typeless":"Typeless","aquaVoice":"Aqua Voice","microsoft365":"Microsoft 365 Dictation（Microsoft 支持）","talon":"Talon Voice","googleDocs":"Google Docs Voice Typing（Google 支持）","whisperCpp":"whisper.cpp","openaiWhisper":"OpenAI Whisper"},"cta":{"title":"想试试 OpenWhispr 吗？","body":"面向 Windows、macOS 和 Linux 的开源听写工具。按键说话、whisper.cpp、AI 整理，以及 99+ 种语言——永久免费。","starButton":"在 GitHub 上为我们点 Star","footnote":"无需账户 · 可离线工作 · 永远开源"},"internalLinks":{"whisperModelSizes":"Whisper 模型大小对比","windowsSpeech":"OpenWhispr 对比 Windows Speech","dragon":"OpenWhispr 对比 Dragon","windowsUseCase":"面向 Windows 的 OpenWhispr","allComparisons":"所有对比"}},"blogLinuxTools":{"metadata":{"title":"2026 年 Linux 上最好用的听写工具 | OpenWhispr","description":"全面对比 2026 年 Linux 平台上可用的听写与语音转文字工具，涵盖 OpenWhispr、Nerd Dictation、Speech Note 等。","keywords":["Linux 最佳听写应用","Linux 语音转文字","Linux 语音识别","Linux 听写软件","Linux 语音输入","2026 Linux 语音转文字","Whisper Linux","Linux 开源听写","Linux 语音听写"],"ogTitle":"2026 年 Linux 上最好用的听写工具","ogDescription":"全面对比 2026 年 Linux 平台上可用的听写与语音转文字工具。"},"article":{"title":"2026 年 Linux 上最好用的听写工具","description":"全面对比 2026 年 Linux 平台上可用的听写与语音转文字工具。"},"breadcrumb":{"home":"首页","blog":"博客","article":"Linux 最佳听写工具"},"layout":{"title":"2026 年 Linux 上最好用的听写工具","subtitle":"多年来，Linux 用户在听写软件上一直缺乏优质选择。如今这一局面终于开始改变。以下是 2026 年所有可行的方案，我们做了诚实的对比。","date":"2026 年 2 月 14 日","tag":"博客"},"sections":[{"id":"direct-answer","title":"直接答案"},{"id":"fact-check","title":"事实核查速览"},{"id":"spectrum","title":"听写工具光谱"},{"id":"openwhispr","title":"OpenWhispr"},{"id":"nerd-dictation","title":"Nerd Dictation"},{"id":"speech-note","title":"Speech Note"},{"id":"simon","title":"Simon (KDE)"},{"id":"chrome-voice-typing","title":"Chrome 语音输入"},{"id":"whisper-cpp","title":"Whisper.cpp CLI"},{"id":"talon-voice","title":"Talon Voice"},{"id":"comparison","title":"对比表"},{"id":"recommendations","title":"推荐建议"},{"id":"sources","title":"参考来源"},{"id":"try-openwhispr","title":"试用 OpenWhispr"}],"directAnswer":{"leadStrong":"2026 年 Linux 上最好用的听写工具，取决于你的具体需求。","bodyBeforeLink":"对于大多数希望获得精致、开箱即用体验并带有 AI 文本润色的用户来说，","linkLabel":"OpenWhispr","bodyAfterLink":"提供了最完整的一揽子方案：由 whisper.cpp 驱动的按键说话听写、全系统快捷键支持、99+ 种语言，以及可选的 AI 排版——而且全部开源。对于偏好命令行、追求最少依赖的用户，Nerd Dictation 非常出色。KDE 与 Qt 用户则应关注 Speech Note。习惯在终端中处理音频管道的开发者，可以直接使用 whisper.cpp。而如果你需要超越单纯听写、实现完全免手操控电脑的能力，Talon Voice 则自成一档。","lastUpdated":"最后更新：2026 年 2 月。工具可用性与功能说明均针对每项重要论断交叉核对了至少两个来源。"},"factCheck":{"title":"事实核查速览（双来源）","items":[{"lead":"OpenWhispr 通过 whisper.cpp 在本地运行 Whisper：","body":"其功能集与本地运行时均有公开文档记载。","linkLabel1":"openwhispr.com","linkLabel2":"whisper.cpp"},{"lead":"Linux 上的选择从精致的图形界面到命令行脚本不一而足：","body":"项目仓库与发行版页面证实了其适用范围与维护状态。","linkLabel1":"Nerd Dictation","linkLabel2":"Speech Note (Flathub)"},{"lead":"以云端为先的工具以隐私换取便利：","body":"浏览器语音输入虽然方便，却会将音频经由服务商基础设施传输。","linkLabel1":"Chrome 语音输入帮助","linkLabel2":"Google STT 数据记录"},{"lead":"老牌 Linux 语音项目仍然存在，但可能已停滞：","body":"在采用前，仓库活跃度与历史发布时间线都值得关注。","linkLabel1":"Simon 仓库","linkLabel2":"KDE 停止维护归档"},{"lead":"OpenWhispr 的定位：","body":"将 OpenWhispr 纳入此处，是因为它在一款支持 Linux 的应用中集成了本地 ASR、全系统采集以及可选的后处理。","linkLabel1":"OpenWhispr","linkLabel2":"Whisper 上游"}]},"labels":{"pros":"优点","cons":"缺点","pricing":"价格：","bestFor":"最适合："},"tools":{"openwhispr":{"title":"1. OpenWhispr","intro":"OpenWhispr 是一款开源、跨平台的桌面听写应用，基于 Electron 与 whisper.cpp 构建。它提供全系统可用的按键说话语音转文字——按住快捷键、开口说话、松手，转写出的文字便会出现在光标所在之处。它还包含一个 AI 智能体模式，可借助大语言模型（OpenAI、Anthropic、Google Gemini 或本地模型）清理、排版并重组你口述的文本。","pros":["精致的图形界面，配有全系统按键说话快捷键","AI 智能体模式可自动清理并排版文本","通过 Whisper 模型（从 tiny 到 large）支持 99+ 种语言","完全开源——可从源码构建，也可使用预编译二进制文件"],"cons":["基于 Electron——比原生应用更占用资源","AI 清理功能需要 API 密钥或 Pro 订阅","较大的 Whisper 模型需要不错的硬件才能实现实时性能"],"pricingValue":"核心功能免费，可选 Pro 每月 $8","bestForValue":"希望获得完整、精致听写体验并带有 AI 文本清理的用户"},"nerdDictation":{"title":"2. Nerd Dictation","intro":"Nerd Dictation 是一个单文件 Python 脚本，借助 VOSK API 提供离线语音转文字。它专为习惯命令行的 Linux 高级用户设计。你通过 CLI 命令手动启动和停止它，它会使用 xdotool 将文字输入到当前获得焦点的窗口中。用户配置就是一个 Python 脚本，因此可以无限折腾。","pros":["极其轻量——单个 Python 文件，依赖极少","可完全离线，搭配小巧的 VOSK 语言模型（不到 50 MB）","通过 Python 配置脚本可无限定制","无后台进程——按需启动和停止"],"cons":["没有图形界面——完全靠命令行驱动，对新手不友好","VOSK 的准确度明显低于基于 Whisper 的工具","需要 xdotool（仅限 X11——对 Wayland 支持有限）"],"pricingValue":"免费且开源","bestForValue":"精通命令行、想要一款轻量、可折腾、完全离线听写脚本的 Linux 用户"},"speechNote":{"title":"3. Speech Note (Dsnote)","intro":"Speech Note 是一款基于 Qt 的 Linux 应用，可借助离线语音转文字、文字转语音和机器翻译来进行笔记、阅读和翻译。它支持多种 STT 引擎，包括 Whisper（通过 whisper.cpp 和 Faster Whisper），并以 Flatpak 形式在 Flathub 上提供。最新版本支持通过 Vulkan 为 Intel、AMD 和 NVIDIA 显卡提供 GPU 加速。","pros":["完整的图形界面，内置笔记、TTS 和翻译功能","支持包括 Whisper 模型在内的多种 STT 引擎","GPU 加速（Vulkan），转写更快","在大多数发行版上可通过 Flatpak 轻松安装"],"cons":["设计定位是笔记应用，而非全系统听写工具","虽有全局键盘快捷键，但不如按键说话那样顺畅","在基于 GNOME 的桌面上，Qt 界面可能显得不够精致"],"pricingValue":"免费且开源","bestForValue":"希望拥有一款内置语音识别与翻译的一体化笔记应用的用户"},"simon":{"title":"4. Simon (KDE)","intro":"Simon 是一套开源语音识别系统，最初作为 KDE 项目的一部分开发。它采用了独特的自己动手思路，允许用户从零开始创建自定义语言模型和声学模型，而非依赖预训练模型。然而必须直说：Simon 实际上已无人维护。最后一次重要发布是在 2013 年，尽管 2017 年曾有过短暂的复兴尝试，近年来却再无活跃开发。","pros":["可针对特定词汇训练自定义声学模型","完全开源，带有图形界面","完全离线运行"],"cons":["实际上已被弃置——自 2013 年起再无实质更新","从未迁移到 KDE Frameworks 5 或 Qt5/Qt6","识别准确度远低于现代基于 Whisper 的工具"],"pricingValue":"免费且开源","bestForValue":"仅供历史研究——不推荐在 2026 年用于实际用途","honestAssessmentLead":"诚实评价：","honestAssessmentBody":"Simon 开创了 Linux 上的开源语音识别，但 whisper.cpp 等现代工具已让它的思路过时。我们在此列出它只是为了内容完整，并不会推荐新用户使用。"},"chrome":{"title":"5. Google Chrome 语音输入","intro":"Google 内置的语音输入功能可供 Linux 用户通过 Chrome 或 Google 文档使用（工具 > 语音输入）。它使用 Google 的云端语音识别，支持多种语言。除了拥有 Chrome 之外无需任何安装——这使它成为门槛最低的选择，但也是最受限的一个。","pros":["零配置——打开 Chrome 即可开始说话","由 Google 语音模型驱动，准确度不错","支持 100+ 种语言"],"cons":["仅在 Chrome/Google 文档内可用——并非全系统","所有音频都会发送到 Google 的服务器——没有离线模式","非开源，是闭源的专有服务"],"pricingValue":"免费","bestForValue":"在不需要全系统输入或隐私时，在 Google 文档内快速听写"},"whisperCpp":{"title":"6. Whisper.cpp CLI","intro":"Whisper.cpp 是 OpenAI 的 Whisper 语音识别模型的 C/C++ 移植版本，针对 CPU 推理做了优化。严格来说它并不是一款听写应用——而是一个用于转写音频文件的命令行工具。不过，许多 Linux 用户会围绕它构建听写工作流，通过脚本将麦克风输入管道化处理。whisper-dictation 和 whispertux 等项目正是将 whisper.cpp 包装进按键说话界面，用于这一目的。","pros":["顶尖的 Whisper 准确度，本地运行","高度优化的 C++——即便在纯 CPU 机器上也很快","在所有 Whisper 模型尺寸下均支持 99+ 种语言","完全开源，维护非常活跃"],"cons":["没有图形界面，没有按键说话——纯粹是转写引擎","需要编写脚本才能搭建实时听写工作流","为文件转写而设计，而非实时流式听写"],"pricingValue":"免费且开源","bestForValue":"想要构建自定义听写流水线或转写工作流的开发者"},"talon":{"title":"7. Talon Voice","intro":"Talon Voice 是一套全面的免手输入系统，其功能远超听写。它让你完全用语音控制整台电脑——窗口管理、写代码、切换应用、浏览文件，并可通过 Tobii 设备选配眼动追踪。Talon 在 X11 上支持 Linux（Ubuntu 18.04+ 及大多数现代发行版）。不过，目前并不支持 Wayland，写作本文时也无相关计划。","pros":["完全免手操控电脑，不只是听写","专为语音编程打造，配有针对各编程语言的命令","集成眼动追踪以替代鼠标","对无障碍辅助意义非凡——对 RSI 患者来说堪称改变人生"],"cons":["学习曲线陡峭——需要记忆命令和一套语音字母表","核心引擎为专有（社区脚本是开源的）","Linux 支持仅限 X11——不支持 Wayland"],"pricingValue":"公开版免费，通过 Patreon 提供每月 $25 的测试版","bestForValue":"需要完全免手操控电脑、语音编程或以无障碍为驱动的输入方式的用户"}},"comparison":{"title":"逐项对比","headerTool":"工具","headerOpenSource":"开源","headerOffline":"离线","headerLanguages":"语言","headerAiCleanup":"AI 清理","headerSystemWide":"全系统","headerPrice":"价格","rows":[{"tool":"OpenWhispr","openSource":"full","offline":"full","languages":"99+","aiCleanup":"full","systemWide":"full","price":"免费（Pro 每月 $8）"},{"tool":"Nerd Dictation","openSource":"full","offline":"full","languages":"因 VOSK 模型而异","aiCleanup":"none","systemWide":"full","price":"免费"},{"tool":"Speech Note","openSource":"full","offline":"full","languages":"因后端/模型而异","aiCleanup":"none","systemWide":"partial","price":"免费"},{"tool":"Simon","openSource":"full","offline":"full","languages":"自定义","aiCleanup":"none","systemWide":"partial","price":"免费"},{"tool":"Chrome Voice Typing","openSource":"none","offline":"none","languages":"众多浏览器支持的语言","aiCleanup":"none","systemWide":"none","price":"免费"},{"tool":"Whisper.cpp CLI","openSource":"full","offline":"full","languages":"99+","aiCleanup":"none","systemWide":"none","price":"免费"},{"tool":"Talon Voice","openSource":"partial","offline":"full","languages":"英语","aiCleanup":"none","systemWide":"full","price":"核心免费 / Patreon 付费档位"}]},"recommendations":{"title":"我们的推荐","intro":"Linux 上并不存在唯一的“最佳”听写工具——这取决于你看重什么。以下是我们的分类建议：","mostUsers":{"title":"对大多数用户","leadBrand":"OpenWhispr","body":"提供了最完整的体验：真正的图形界面、全系统按键说话、AI 清理、Whisper 级准确度，而且开源。它最接近 macOS 和 Windows 用户习以为常的那种体验。"},"cliPurists":{"title":"对命令行纯粹主义者","leadBrand":"Nerd Dictation","body":"以其简洁而出色。单个 Python 文件、由 VOSK 驱动、可无限折腾。如果你偏好通过脚本来配置工具，又不需要 Whisper 级别的准确度，它几乎无可匹敌。"},"kdeQt":{"title":"对 KDE/Qt 用户","leadBrand":"Speech Note","body":"与基于 Qt 的桌面集成良好，除语音识别外还提供翻译和 TTS。如果你想要一款一体化的笔记工具，它是个有力的选择。"},"developers":{"title":"对开发者","leadBrand":"Whisper.cpp CLI","body":"是驱动这些工具中许多款的引擎。如果你想要最大程度的掌控，又乐于搭建自己的流水线，那就直接从源头入手。"},"accessibility":{"title":"对无障碍辅助与免手操控电脑","leadBrand":"Talon Voice","body":"完全是另一个层次。如果你需要彻底替代键盘和鼠标——而不只是用听写来辅助——那么 Talon 就是值得学习的工具。投入的时间相当可观，但回报是革命性的。"}},"sources":{"title":"参考来源与延伸阅读","links":["OpenWhispr","OpenAI Whisper","Nerd Dictation (GitHub)","Speech Note (Flathub)","Simon (KDE GitHub 镜像)","whisper.cpp","Talon Voice","Chrome 语音输入帮助","Google 语音转文字数据记录政策"]},"cta":{"title":"想试试 OpenWhispr 吗？","body":"面向 Linux、macOS 和 Windows 的开源听写工具。按键说话、whisper.cpp、AI 清理，以及 99+ 种语言——永久免费。","starButton":"在 GitHub 上给我们加星","footnote":"无需账户 · 离线可用 · 永久开源"},"internalLinks":{"whisperModelSizes":"Whisper 模型尺寸对比","talonCompare":"OpenWhispr 对比 Talon","linuxUseCase":"OpenWhispr 之于 Linux","allComparisons":"全部对比"}},"blogMeetingNotes":{"metadata":{"title":"如何做好会议记录：实用指南 | OpenWhispr","description":"学习如何记录真正有用的会议笔记：决定、行动项和待解决问题。内含免费模板和 AI 会议笔记工具。","keywords":["如何做好会议记录","如何记录会议","会议记录模板","高效会议记录","会议记录最佳实践","记笔记方法","会议中如何记笔记","会议纪要 vs 会议记录","会议行动项","康奈尔笔记法","会议记录工具","会议记录应用"]},"schema":{"title":"如何做好会议记录：实用指南","description":"一个真正会被使用的会议记录系统：记录决策、行动项和未解决问题，快速整理，并有效分享。附免费模板。"},"breadcrumb":{"home":"首页","blog":"博客","article":"如何做好会议记录"},"hero":{"title":"如何做好会议记录：实用指南","subtitle":"别再逐字转写。记录真正重要的决策、行动项和未解决问题，把其余工作交给合适的工具。","date":"2026年6月1日","tag":"指南"},"sections":[{"id":"why","title":"为什么记录重要"},{"id":"before","title":"会前五分钟准备"},{"id":"structure","title":"使用同一结构"},{"id":"capture","title":"该写什么"},{"id":"methods","title":"三种方法"},{"id":"after","title":"会后整理"},{"id":"tools","title":"让工具帮忙"},{"id":"privacy","title":"同意与隐私"},{"id":"share","title":"分享与存放"},{"id":"mistakes","title":"常见错误"},{"id":"template","title":"免费模板"},{"id":"faq","title":"FAQ"}],"intro":{"lead":"好的会议记录捕捉的是决策、行动项和未解决问题，而不是逐字稿。","paragraph1":"目标不是把每句话都写下来，而是在会后留下简短、共享的记录：决定了什么、谁要在什么时候做什么、还有什么没有解决。做得好，一通30分钟的会议就能变成整个团队可以执行的内容。做得差，下次会议就会重新讨论已经定下来的事。","paragraph2":"这份指南会带你走过一个简单系统：会前五分钟准备，每次使用同一结构，只记录重要内容，会后马上整理，并把记录分享到大家真正会看的地方。无论你手动输入，还是让会议记录工具帮你录音，这套方法都适用。","lastUpdated":"最后更新：2026年6月1日。"},"why":{"title":"为什么好的会议记录很重要","paragraph1":"会议成本很高。六个人开一小时会，就是某个人一整天的工作时间。那一小时的产出几乎从来不是谈话本身，而是决策和下一步。如果这些没有写下来，这场会议基本等于没有发生。","paragraph2":"好的记录有三个作用。它给参会者一个共享记忆，让大家下周不会重新争论已经定下来的决定。它让缺席者不用再开一场复盘会，也能快速跟上。它还把模糊的意向变成有负责人和日期的可追踪行动项。这就是“我们应该研究一下”和真正会完成的任务之间的区别。"},"prep":{"title":"会前五分钟准备","lead":"最好的记录从会议开始前就已经开始了。","intro":"五分钟准备能让你进会时已经知道该听什么，而不是从第一句话开始手忙脚乱地追。","items":[{"label":"阅读议程。","text":"它告诉你会议的目的，也就告诉你哪些时刻值得记录。"},{"label":"打开你的模板。","text":"新建一页，在顶部写上日期、参会者和一句话目的，下面放好四个空栏目（后面会讲）。"},{"label":"记下上次遗留的问题。","text":"把未解决问题带到这次会议里，能防止它们悄悄消失。"}],"outro":"如果没有议程，就自己写两行：这次会议的目标，以及你预计要做出的决定。仅仅这一步，就会让讨论和记录都更清晰。"},"structure":{"title":"每次使用同一结构","lead":"固定格式是提升会议记录质量的最大单项改进。","intro":"当页面结构已经摆好，你就不用再分心想内容放哪里，而是把注意力放在倾听上。几乎所有会议都适合这四个栏目：","items":[{"label":"决策。","text":"团队实际同意的事项。每条一行，用清楚的日常语言写。"},{"label":"行动项。","text":"下一步任务，每项都要有负责人和截止日期。没有名字的任务只是愿望，不是行动。"},{"label":"未解决问题。","text":"任何被提出但尚未解决的事项，确保之后会被继续处理而不是遗忘。"},{"label":"备注和背景。","text":"少量值得保留的背景细节，例如数字、链接、决策背后的理由。"}],"outro":"顺序可以按你的习惯调整，但栏目集合要固定。结构一致的记录更容易浏览、搜索，也更容易交给没参会的人。"},"capture":{"title":"该写什么（以及该跳过什么）","lead":"记笔记最难的部分，是决定哪些内容不写。","intro":"试图捕捉所有内容，反而什么都捕捉不好。你低头转写时，真正的决策可能已经发生。相反，留意一些表示“值得保留”的信号短语：","signals":[{"label":"决策：","text":"“就选……吧”、“我们已经决定……”、“我们会发布……”"},{"label":"负责人：","text":"“我来处理”、“Sam负责……”、“你能接这个吗……”"},{"label":"截止时间：","text":"“周五前”、“发布前”、“下个迭代”"},{"label":"风险和阻塞：","text":"“问题在于……”、“我们卡在……”、“这取决于……”"},{"label":"难以记住的细节：","text":"一小时后你不会记得的数字、姓名、日期和链接。"}],"skip":"其余内容跳过：寒暄、边想边说、跑题，以及议程或幻灯片里已经写好的内容。没必要把邀请里已经附上的图表再抄一遍。","testTitle":"一周测试","testBody":"写下某件事之前，先问自己：“下周会有人需要这个吗？”如果会，就记录。如果不会，就放过它。这个问题会安静地过滤掉大部分噪音。"},"methods":{"title":"三种有效的会议记录方法","intro":"记会议记录没有唯一“正确”的方式，只有适合当前会议的方法。值得掌握的有三种：","list":[{"label":"大纲法。","text":"嵌套项目符号：主点下面放子点。快速、灵活，也是多数会议的默认好选择。它能很好地映射到上面的四栏结构。"},{"label":"康奈尔笔记法。","text":"把页面分成窄提示栏、宽笔记栏和底部摘要区。它原本是学习技巧，但在一对一和学习密集型会议中很有用，方便之后回顾。"},{"label":"决策与行动法。","text":"完全跳过叙述，只写两张清单：已经做出的决策，以及需要完成的行动项。对于状态更新、站会和规划会这种只看结果的会议，它非常高效。"}],"outro":"按会议类型选择，并在同一类型内保持一致。规划会和客户访谈不需要同一种方法，但每场规划会都应该使用同一种方法。"},"after":{"title":"五分钟内整理干净","lead":"会议一结束，你对会议的记忆就开始衰减。","intro":"会中看得懂的碎片和缩写会很快变模糊。会后立刻花五分钟，在下一件事把你拉走之前，把粗糙记录整理成陌生人也能读懂的内容。","items":["趁你还记得含义，展开自己的缩写。","检查每个行动项是否都有负责人和截止日期。现在补齐空缺，不要等到下周。","删除混进来的噪音，修正那些只写了一半的内容。"],"outro":"现在五分钟，胜过之后一小时的来回澄清。大多数人会跳过这一步，而这一步正是可信记录和无人理会的记录之间的区别。"},"tools":{"title":"让工具承担重活","lead":"手动记会议记录有一个硬限制：你无法同时完全倾听又完全转写。","paragraph1":"会议记录工具能消除这个取舍。它录制通话、转写内容并标注谁说了什么，让你留在对话中，而不是边开会边打字；会后再编辑一个现成草稿。","paragraph2Pre":"这会把你的工作从","capturing":"捕捉","paragraph2Mid":"变成","curating":"整理","paragraph2Post":"：文字稿已经存在，你只需要提取决策和行动项。当你是会议主持人、不能低头看键盘时，这尤其有用。","paragraph3":"但在开启工具之前，有两件事值得先处理好：每个人的同意，以及你的音频最终去了哪里。下一节会讲这个。"},"privacy":{"title":"录制会议：取得同意并保护隐私","lead":"录音是捕捉会议最准确的方法，也是在不谨慎时最快制造隐私问题的方法。","intro":"按下录制前有两件事很重要：取得同意，以及知道音频和文字稿之后会存在哪里。","consentLead":"先取得同意。","consentBody":"各地录音法律不同。有些地方允许单方同意，有些地方要求通话中所有人同意，欧盟GDPR等规则还会把录制的声音视为个人数据。你不需要背下所有条文；你需要养成一个习惯：每次录音前都先询问。大声说出来，并在记录中注明“已获同意录制”，能让你更稳妥。","scriptTitle":"一句话同意话术","scriptBody":"“我可以录一下吗？这样我能专注参与讨论，会后再分享记录。”——开始前先问。如果有人反对，就改为手动记笔记。","audioLead":"知道音频会去哪里。","audioBody":"大多数云端笔记工具——Otter、Fireflies、Granola——会把你的会议上传到它们的服务器处理，录音和文字稿可能被保存、用于训练AI，或在数据泄露中暴露。对于普通内部同步，这通常是可以接受的取舍。但对于任何敏感内容——HR谈话、法律或医疗细节、客户数据、交易条款——录制前都值得认真考虑这是真实风险。","habitsIntro":"几个习惯能让你站在安全的一边：","habits":["按敏感度选择工具——机密会议不要放到第三方服务器上。","检查工具是否会用你的录音训练AI。很多工具默认会这样做，而且通常是选择退出，不是选择加入。","如果不需要音频本身，记录写完后删除录音。","当对话是私密的，优先选择在设备端处理的工具。"],"ctaPre":"这正是","ctaMid":"采用本地优先架构的原因——它是一个免费、开源的会议记录工具，可以在你自己的设备上完成录制、转写和“谁说了什么”的标注。音频和文字稿不会离开你的机器，因此“我的数据去哪了？”有一个简单答案。我们在","diarizationLink":"本地话者分离","ctaMid2":"中拆解了设备端说话人标注，在","transcriptionLink":"本地 vs 云端转写","ctaPost":"中解释了完整取舍。"},"share":{"title":"把记录分享并存放到大家会看的地方","lead":"没人看的记录就是浪费。","intro":"最后一步，会把个人涂写变成团队资产。","items":[{"label":"一小时内发送，","text":"趁会议对所有参会者还很新鲜。"},{"label":"先放决策和行动项。","text":"把大家需要执行的内容放在顶部，直接标记每个负责人，避免有人漏掉自己的任务。"},{"label":"存到团队已经工作的地方","text":"——共享文档、项目工具、频道。每个项目一个可查找的归宿，几个月后记录仍然能被搜索到。"}],"outro":"好的记录系统有一个测试：三个月后，有人能否在30秒内回答“我们当时关于X决定了什么？”如果你的记录能做到，它就在发挥作用。"},"mistakes":{"title":"要避免的常见错误","items":[{"label":"逐字转写全部内容。","text":"一堵文字墙帮不了任何人。记录结果，不要记录整场对话。"},{"label":"行动项没有负责人或日期。","text":"“应该有人研究一下”永远不会完成。给每个任务都附上名字和截止时间。"},{"label":"每次格式都不同。","text":"不一致的记录写起来慢，也无法快速浏览。复用一个模板。"},{"label":"从不分享。","text":"只存在你笔记本里的记录无法让团队对齐。每次都发出去。"},{"label":"把整理工作交给记忆。","text":"“我之后再整理”通常等于永远不会。会后五分钟内就做完。"}]},"template":{"title":"可复制的会议记录模板","intro":"下面是一页版完整系统。复制它，放在方便的位置，每次开会复用：","code":"会议：\n日期：<date> 参会者：<names>\n目的：<one line — why are we here?>\n\n决策\n-\n\n行动项\n- [ ] <task> — <owner> — <due date>\n\n未解决问题\n-\n\n备注 / 背景\n-\n\n下次会议 / 后续跟进：<date or trigger>","outro":"就是这样。会前用它准备，会中填写，会后花五分钟整理，然后发出去。这个结构会替你完成大部分工作。"},"faq":{"title":"常见问题"},"faqs":[{"question":"怎样做好会议记录？","answer":"会前准备一个简单模板，然后只捕捉重要内容：决策、有负责人和截止日期的行动项，以及未解决问题。不要试图逐字转写，留意“我们已经决定”“我来处理”“周五前”这样的信号。会议结束后几分钟内整理记录，然后分享到团队已经工作的地方。"},{"question":"会议记录应该包含什么？","answer":"好的会议记录顶部应包含日期、参会者和目的，然后是四个部分：已做决策、行动项（每项都有负责人和截止时间）、未解决问题，以及值得保留的背景信息。其他内容——寒暄、边想边说、幻灯片里已有的内容——都可以省略。"},{"question":"会议记录最好的格式是什么？","answer":"一致、可复用的格式胜过任何聪明格式。最可靠的结构是四个栏目：决策、行动项、未解决问题、备注。每次使用同一模板，意味着你把注意力放在倾听上，而不是决定页面怎么排版，也让不同会议的记录之后更容易浏览和搜索。"},{"question":"会议记录和会议纪要有什么区别？","answer":"会议纪要通常是正式且常常必须保存的记录，包括出席情况、动议和投票，常见于董事会和法律场景。会议记录更轻量、更实用，目的是帮助团队记住决策并推进行动项。大多数工作会议需要的是好的记录，而不是正式纪要。"},{"question":"怎样一边记笔记一边专心听？","answer":"停止试图把所有内容都写下来。当你只记录决策、负责人和截止时间时，大多数时候双手是空的，你可以真正倾听。能录制并转写通话的会议记录工具会彻底移除这个冲突：你在会议中保持专注，会后再编辑现成草稿。"},{"question":"应该录制会议来做记录吗？","answer":"录制能让你保持参与，并获得准确文字稿，而不是实时打字；但一定要先告知参会者，很多地方要求同意。更大的问题是音频会去哪里。多数云端笔记工具会把会议上传到服务器。对于敏感对话，OpenWhispr这样的本地优先工具会在你自己的设备上录制、转写并标注说话人，因此不会上传任何内容。"},{"question":"为了做记录而录制会议合法吗？","answer":"这取决于参会者所在地区。有些地区只需要一方同意，有些地区要求通话中所有人同意，欧盟GDPR等规则还会把录制的声音视为个人数据。这不是法律建议，但在任何地方最安全的习惯都一样：开始前告诉大家你在录制，并取得他们同意。"},{"question":"使用AI会议记录工具时，我的音频会去哪？","answer":"多数云端笔记工具会把会议音频和文字稿上传到公司的服务器进行处理和保存，有些还会用这些数据训练模型。使用OpenWhispr这样的本地优先工具时，录音、文字稿和说话人标签都留在你自己的设备上，永远不会上传。这对HR、法律或涉及客户数据的敏感会议尤其重要。"},{"question":"主持会议时怎样记笔记？","answer":"当你在主持会议时，你不可能同时打字。要么指定记录员，要么让会议记录工具捕捉文字稿，这样你能专注于现场。如果必须自己处理，就只用短片段记下决策和行动项，然后在会后五分钟内趁记忆新鲜把它们展开。"},{"question":"最好的会议记录应用是什么？","answer":"最好的会议记录应用，是能准确捕捉对话并符合你隐私需求的那个。Otter、Fireflies、Granola等云端工具很方便，但会把你的音频发送到它们的服务器。OpenWhispr是免费的开源会议记录工具，可以在本地转写并识别说话人，因此音频永远不会离开你的设备。"}],"cta":{"title":"不用打一个字也能做会议记录","body":"OpenWhispr是一款免费、开源的会议记录工具，可以录制、转写并标注谁说了什么，而且完全在你的设备上完成。你可以专注参会，会后再编辑现成草稿。支持macOS、Windows、Linux 和 iOS。","starButton":"在GitHub上给我们加星","footnote":"无需账户 · 可离线使用 · 永远开源"},"internalLinks":[{"href":"/blog/local-speaker-diarization","label":"私密会议记录工具"},{"href":"/blog/local-vs-cloud-transcription","label":"本地 vs 云端转写"},{"href":"/blog/best-dictation-tools-windows-2026","label":"Windows最佳听写工具"},{"href":"/blog","label":"所有博客文章"}]},"blogHowWhisper":{"metadata":{"title":"Whisper AI 工作原理：完整指南 | OpenWhispr","description":"详解 Whisper AI 的工作原理：Transformer 架构、68 万小时训练数据、模型大小和准确率基准。开源，并可在设备本地运行。","keywords":["Whisper AI 工作原理","Whisper 语音识别","OpenAI Whisper 解释","Whisper 模型架构","Whisper AI 精度","Whisper Transformer 模型","自动语音识别","Whisper vs 其他 ASR","whisper.cpp 解释"]},"schema":{"title":"Whisper AI 的工作原理：完整指南","description":"深入解析OpenAI的Whisper语音识别模型 — 架构、训练数据、精度基准和实际使用。"},"breadcrumb":{"home":"首页","blog":"博客","article":"Whisper AI 的工作原理"},"hero":{"title":"Whisper AI 的工作原理：完整指南","subtitle":"对OpenAI开源语音识别模型的技术深挖：从梅尔频谱图到Transformer解码。","date":"2026年2月3日","tag":"技术"},"sections":[{"id":"what-is-whisper","title":"Whisper是什么？"},{"id":"architecture","title":"架构"},{"id":"training-data","title":"训练数据"},{"id":"model-sizes","title":"模型尺寸"},{"id":"accuracy","title":"精度"},{"id":"whisper-cpp","title":"whisper.cpp"},{"id":"comparison","title":"ASR对比"},{"id":"applications","title":"应用"},{"id":"sources","title":"来源"}],"intro":{"paragraph":"Whisper是OpenAI开发的自动语音识别(ASR)模型。 它于2022年9月发布，是一个通用语音识别系统，使用从互联网收集的68万小时多语言音频数据训练而成。Whisper采用编码器-解码器Transformer架构，将音频转换为log-mel频谱图，通过神经网络编码器处理，并以自回归方式解码文本token。它支持99种语言的转写、翻译为英文、语言识别和时间戳预测，而且全部在一个统一模型中完成。与多数商业ASR服务不同，Whisper的权重在MIT许可证下<openSourceLink>完全开源</openSourceLink>，由此催生了繁荣的移植、优化和应用生态，包括<whisperCppLink>whisper.cpp</whisperCppLink>这个C/C++实现，让Whisper能够实际用于实时、设备端语音识别。","lastUpdated":"最后更新：2026年2月17日。本文中的定量主张均已与至少两个主要来源交叉核对。"},"factCheck":{"title":"事实核查快照（双来源）","items":[{"label":"Whisper发布时间、训练规模和许可证：","body":"OpenAI于2022年9月发布Whisper，使用68万小时数据训练，并采用MIT许可证。"},{"label":"任务范围和语言支持：","body":"Whisper支持99种语言的转写、翻译成英文、语言识别和时间戳token。"},{"label":"英文基准背景：","body":"约3% WER是在LibriSpeech test-clean上针对英文评估的基准数字，不是通用真实世界错误率。"},{"label":"近期模型更新：","body":"large-v3在许多语言上报告的错误率低于large-v2，turbo则通过更小的解码器计算量专注于降低延迟。"},{"label":"与OpenWhispr的关系：","body":"OpenWhispr通过whisper.cpp使用Whisper，为桌面平台提供本地、离线优先的听写。"}],"sources":["<a1>arXiv论文</a1> · <a2>OpenAI Whisper仓库</a2>","<a1>Whisper模型卡</a1> · <a2>Whisper README</a2>","<a1>large-v2模型卡</a1> · <a2>基准详情</a2>","<a1>large-v3模型卡</a1> · <a2>turbo发布讨论</a2>","<a1>whisper.cpp</a1> · <a2>OpenWhispr</a2>"],"sourceUrls":[["https://arxiv.org/abs/2212.04356","https://github.com/openai/whisper"],["https://github.com/openai/whisper/blob/main/model-card.md","https://github.com/openai/whisper"],["https://huggingface.co/openai/whisper-large-v2","https://github.com/openai/whisper/blob/main/model-card.md"],["https://huggingface.co/openai/whisper-large-v3","https://github.com/openai/whisper/discussions/2363"],["https://github.com/ggml-org/whisper.cpp","https://openwhispr.com"]]},"whatIs":{"title":"Whisper是什么？","paragraph1":"Whisper由OpenAI的Alec Radford、Jong Wook Kim、Tao Xu、Greg Brockman、Christine McLeavey和Ilya Sutskever在论文<paperLink>“Robust Speech Recognition via Large-Scale Weak Supervision”</paperLink>中提出。代码和模型权重于2022年9月以MIT许可证发布在<githubLink>GitHub</githubLink>上。","paragraph2":"在Whisper之前，最先进的语音识别系统通常使用经过整理、人工标注的数据集训练，这些数据集往往局限于特定语言、口音或领域。Whisper打破了这一做法，使用从互联网抓取的音频与文字稿配对数据，规模达到68万小时，且数据非常多样。关键洞见在于，这种“弱监督”方式——使用不完美的机器生成或用户上传文字稿，而不是人工逐条标注的数据——可以产生一个在语言、口音、背景噪声和技术词汇上都非常稳健的模型。","paragraph3":"结果是一个可以处理多种语音任务的单一模型：转写（同语种语音转文本）、翻译（任意语言语音转英文文本）、语言识别，以及带时间戳的语音活动检测。这种多任务能力通过一套特殊token系统嵌入到模型架构中，用这些token指定要执行的任务。","paragraph4":"Whisper的开源发布意义重大。它让高质量ASR变得人人可用，开发者、研究人员和公司都能使用一个可与商业API竞争的模型，而且不需要按分钟付费、不需要把音频发送到云端，也没有供应商锁定。这推动了<whisperCppLink>whisper.cpp</whisperCppLink>、<fasterWhisperLink>Faster Whisper</fasterWhisperLink>等工具，以及建立在它们之上的几十个应用的发展，其中也包括<openWhisprLink>OpenWhispr</openWhisprLink>。"},"architecture":{"title":"架构深挖","intro":"Whisper使用编码器-解码器Transformer架构，这与原始Transformer（Vaswani et al., 2017）以及许多机器翻译系统背后的基本设计相同。编码器处理音频，解码器生成文本。下面是每个阶段的工作方式。","preprocessing":{"title":"音频预处理","intro":"在任何神经网络处理之前，原始音频会被转换成一种声音的可视化表示，称为log-mel频谱图。流程如下：","steps":["音频被重采样为16,000 Hz（16 kHz单声道）。","使用25毫秒窗口和10毫秒步幅（hop length）计算短时傅里叶变换(STFT)。","频谱被投影到80个梅尔频率滤波器组（large-v3为128个）上，这些滤波器通过对数间隔近似人类听觉感知。","应用对数缩放，生成最终的log-mel频谱图。","频谱图被分成30秒块。短于30秒的音频会补零，长音频会按顺序分块处理。"],"outro":"结果是一个形状为<code>(80, 3000)</code>的二维表示——80个梅尔通道，3,000个时间步（30秒，10ms步幅）。它类似一张图像，编码器处理它的方式也很像视觉模型处理像素数据。"},"encoder":{"title":"编码器","intro":"编码器把梅尔频谱图转换成一串学习到的音频表示。它包含：","steps":["两个1D卷积层 — 第一个卷积的kernel size为3、padding为1，后接GELU激活。第二个卷积的kernel size为3、stride为2、padding为1，会把时间维度减半，把3,000个时间步下采样到1,500个位置。","正弦位置嵌入 — 与解码器不同，编码器使用固定的正弦嵌入（不是学习得到的）来编码每个时间步的位置。","Transformer块 — 一叠标准Transformer编码器层，每层包含多头自注意力和带GELU激活的前馈网络，并通过残差连接和层归一化连接。"],"outro":"输出是一串1,500个上下文化音频特征向量——输入音频每20毫秒对应一个向量——解码器在生成文本时会对这些向量进行注意力计算。"},"decoder":{"title":"解码器","intro":"解码器以自回归方式生成文本token：一次生成一个token，每个token都依赖编码后的音频和之前生成的所有token。它包含：","steps":["Token嵌入层 — 将token ID转换为稠密向量表示。","学习到的位置嵌入 — 与编码器的正弦嵌入不同，解码器使用随模型一起训练的位置嵌入。","带交叉注意力的Transformer块 — 每个解码器层有三个子层：masked self-attention（防止模型看到未来token）、对编码器输出的cross-attention（让模型“听”音频），以及前馈网络。"],"outro":"最终的解码器输出会被投影到词表上，生成token概率。解码策略包括贪心搜索、束搜索和基于温度的采样。"},"tokens":{"title":"特殊Token与多任务训练","intro":"Whisper通过一套巧妙的特殊token系统，用单个模型执行多种任务。这些token充当指令。解码器输入遵循如下结构：","startOfTranscript":"— 表示输出序列开始。","language":"— 指定语言（99个语言token之一）。可以自动检测，也可以手动设置。","taskJoin":"或","task":"— 指定是转写原语言，还是翻译成英文。","timestampsPart1":"— 控制是否生成时间戳token。启用时间戳时，模型会生成类似","timestampsJoin":"和","timestampsPart2":"这样的时间偏移token，将文本与音频对齐。","outro":"这种统一token格式意味着，一个模型无需单独的模型头或微调，就能执行转写、翻译、语言检测和带时间戳转写。任务完全由提供给解码器的token序列决定。"}},"trainingData":{"title":"训练数据","paragraph1":"原始Whisper模型（tiny到large-v2）使用从互联网收集的音频和文字稿配对数据训练，总量为680,000小时。这个数据集并未公开。它的关键特征是弱监督：文字稿标签没有经过人工标注者逐条验证，而是来自字幕、闭合字幕以及其他用户生成的音频配套文本。","compositionTitle":"数据组成","compositionIntro":"根据<modelCardLink>官方模型卡</modelCardLink>，68万小时训练集组成如下：","headers":{"segment":"片段","hours":"小时","share":"占比","description":"说明"},"rows":[{"segment":"英语 ASR","hours":"438,000","share":"65%","description":"带英文文字稿的英文音频"},{"segment":"翻译（其他语言到英语）","hours":"126,000","share":"18%","description":"带英文文字稿的非英文音频"},{"segment":"多语言 ASR","hours":"117,000","share":"17%","description":"带母语文字稿的非英文音频（覆盖98种语言）"}],"paragraph2":"训练数据中英文占比很高（65%），这解释了为什么Whisper在英文上的表现明显优于低资源语言。对于2023年11月发布的large-v3模型，OpenAI将训练集扩展到100万小时弱标注音频，并额外加入400万小时伪标注音频，这些伪标签是用Whisper large-v2在未标注数据上生成的，是一种自训练或知识蒸馏形式。","weakTitle":"为什么“弱监督”很重要","weakParagraph1":"大多数早期ASR系统使用LibriSpeech（960小时）或Common Voice（每种语言几千小时）这样的数据集训练，这些数据集经过精心整理并由人类验证。Whisper选择用标签质量换规模：68万小时对比传统系统使用的几百到几千小时。论文证明这个取舍是值得的。来自互联网的数据多样性让Whisper在真实世界条件下非常稳健：背景噪声、重叠语音、口音、领域词汇，以及会让基于录音棚级朗读语音训练的模型感到困难的声学环境。","weakParagraph2":"不过，弱监督也带来了一个已知限制：幻觉。因为训练文字稿并不完美，模型有时会生成听起来合理但实际并未被说出的文本，尤其是在静音或非常安静的片段中。这是该方法固有的取舍，也是仍在持续改进的方向。"},"modelSizes":{"title":"模型尺寸","intro":"Whisper提供多个尺寸，从3900万到15.5亿参数不等。小模型更快但精度较低；大模型更准确但需要更多计算和内存。<code>.en</code>变体是英文专用模型，通常在英文上表现更好，尤其是较小尺寸。large模型没有英文专用版本。","headers":{"model":"模型","parameters":"参数量","englishOnly":"英文专用","vram":"VRAM","speed":"相对速度"},"rows":[{"model":"tiny","parameters":"39 M","englishOnly":"tiny.en","vram":"~1 GB","speed":"~10x"},{"model":"base","parameters":"74 M","englishOnly":"base.en","vram":"~1 GB","speed":"~7x"},{"model":"small","parameters":"244 M","englishOnly":"small.en","vram":"~2 GB","speed":"~4x"},{"model":"medium","parameters":"769 M","englishOnly":"medium.en","vram":"~5 GB","speed":"~2x"},{"model":"large (v1, v2, v3)","parameters":"1,550 M","englishOnly":"--","vram":"~10 GB","speed":"1x"},{"model":"turbo (large-v3-turbo)","parameters":"809 M","englishOnly":"--","vram":"~6 GB","speed":"~8x"}],"tableNote":"速度以large模型为参照。VRAM要求指原始Python实现中通过fp16精度进行GPU推理。<whisperCppLink>whisper.cpp</whisperCppLink>使用的内存明显更少（例如large模型需要约3.9 GB RAM，而不是约10 GB VRAM）。","turbo":"turbo模型（2024年10月发布）是large-v3的蒸馏版本，解码器显著更小。它保留了large-v3的大部分精度，同时速度约快8倍，因此在速度重要时是很强的选择。注意，turbo不支持翻译任务。","large":"large模型经历了三次迭代：large-v1（2022年9月）、large-v2（2022年12月）和large-v3（2023年11月）。每个版本都提高了精度，其中large-v3引入了128个梅尔频率bin（从80个提升），并使用更大的数据集训练。关于每个尺寸以及如何选择的详细拆解，请参阅我们的<guideLink>Whisper模型尺寸指南</guideLink>。"},"accuracy":{"title":"Whisper有多准确？","intro":"Whisper的精度通常用词错误率(WER)衡量，即被错误转写的词所占比例（插入、删除和替换合并计算）。WER越低越好。","englishTitle":"英文基准","englishParagraph":"在LibriSpeech test-clean上，Whisper large-v2的WER约为3.0%<ref1>[1]</ref1><ref2>[2]</ref2>。这是最常用的英文ASR基准之一，由有声书中的干净朗读语音组成。这个结果可与商业ASR系统和专门训练的模型竞争，尽管在LibriSpeech这个特定基准上，针对该数据集微调的专用模型可以达到更低WER。Whisper的强项不在于击败狭窄基准，而在于稳健性，即在广泛真实世界条件下保持稳定表现。","largeV3Title":"Large-v3的改进","largeV3Paragraph":"根据OpenAI的评估，在Common Voice 15和Fleurs评估数据集上，对于模型错误率低于60%的语言，Whisper large-v3相较large-v2显示出10-20%的错误减少<ref2>[2]</ref2><ref3>[3]</ref3>。这些改进在非英文语言上尤其明显，因为扩大的训练数据（总计500万小时）在这些语言上带来的差异最大。","compareTitle":"Whisper的对比表现（2026年2月）","compareParagraph":"自Whisper发布以来，更新的开源ASR模型已经在干净语音基准上超过它。NVIDIA的Parakeet TDT（0.6B参数）在LibriSpeech test-clean上达到1.69% WER，Canary模型达到1.6% WER，都明显低于Whisper large-v3的约2.7%。不过，由于生态成熟、语言覆盖广，以及whisper.cpp等优化运行时可用，Whisper仍是部署最广泛的开源ASR模型。","chartNote":"注：商业云API（Deepgram Nova-3、Google Chirp、AssemblyAI Universal-2）不公布LibriSpeech WER数字，因此无法在这张图中直接比较。它们的基准使用专有真实世界测试集。OpenAI还在2025年3月发布了GPT-4o-transcribe，这是一个仅API可用的模型（不开源，也不基于Whisper），据称错误率更低，但它只能在云端使用，无法本地推理。","excelsTitle":"Whisper擅长的地方","excels":[{"title":"对口音和方言的稳健性","body":"— 使用多样化互联网音频训练，Whisper比只在朗读语音上训练的系统更能处理地区口音。"},{"title":"背景噪声耐受性","body":"— 即使有音乐、其他说话人或环境噪声，模型也能保持合理精度。"},{"title":"专业词汇","body":"— 由于训练数据覆盖广，Whisper比许多通用ASR系统更能处理特定领域术语（医疗、法律、技术）。"},{"title":"多语言能力","body":"— 单一模型支持99种语言，无需按语言单独微调。"}],"strugglesTitle":"Whisper吃力的地方","struggles":[{"title":"幻觉","body":"— 最常被提到的问题。在静音或非常安静的片段中，Whisper可能生成从未说出的文本。这是弱监督训练方式的后果。"},{"title":"低资源语言","body":"— 训练数据较少的语言（68万小时中大部分是英文）错误率明显更高。"},{"title":"重复文本生成","body":"— 自回归解码器可能会“卡住”，反复生成同一句话。带特定参数的束搜索可以缓解，但不能完全消除。"},{"title":"默认并非实时","body":"— Whisper处理30秒块，因此存在固有延迟。流式方案（如whisper.cpp的实时模式）可以绕过这一点，但会增加复杂度。"}]},"whisperCpp":{"title":"whisper.cpp与本地推理","paragraph1":"原始Whisper实现需要Python、PyTorch和支持CUDA的GPU，才能获得合理性能。这使它不适合桌面应用、移动设备和边缘部署。Georgi Gerganov创建的<whisperCppLink>whisper.cpp</whisperCppLink>解决了这个问题。","paragraph2":"whisper.cpp是一个用纯C/C++完整重新实现的Whisper推理引擎，没有依赖。它读取相同的模型权重，但运行时不需要Python、不需要PyTorch，也不需要GPU（虽然有GPU会大幅受益）。关键特性包括：","features":[{"title":"零运行时内存分配","body":"— 所有内存都预先分配，使性能可预测，并适合嵌入式系统。"},{"title":"Apple Silicon优化","body":"— 在搭载M系列芯片的Mac上，编码器可以通过Core ML运行在Apple Neural Engine上，相比仅CPU执行快3倍以上。Metal GPU加速也支持完整GPU推理。"},{"title":"量化支持","body":"— 模型可量化为4位、5位或8位整数精度，在几乎不损失精度的情况下显著降低内存使用并提高速度。"},{"title":"GPU加速","body":"— 支持NVIDIA CUDA（通过cuBLAS）、Vulkan（跨厂商）、OpenVINO（Intel）以及用于CPU加速的OpenBLAS。"},{"title":"语音活动检测(VAD)","body":"— 内置VAD可跳过静音片段，同时提升速度和文字稿质量。"}],"memoryTitle":"内存要求（whisper.cpp）","memoryHeaders":{"model":"模型","disk":"磁盘大小","ram":"所需RAM"},"memoryRows":[{"model":"tiny","disk":"75 MB","ram":"~273 MB"},{"model":"base","disk":"142 MB","ram":"~388 MB"},{"model":"small","disk":"466 MB","ram":"~852 MB"},{"model":"medium","disk":"1.5 GB","ram":"~2.1 GB"},{"model":"large","disk":"2.9 GB","ram":"~3.9 GB"}],"paragraph3":"whisper.cpp可运行在macOS（Intel和Apple Silicon）、Linux、Windows、iOS、Android、FreeBSD、Raspberry Pi，甚至可通过WebAssembly在浏览器中运行。它能在普通消费级硬件上实现快于实时的转写，包括Raspberry Pi 4和iPhone 13这样的小设备。","paragraph4":"正是这种可移植性，让本地、私密的语音识别变得可行。像<openWhisprLink>OpenWhispr</openWhisprLink>这样的应用在底层使用whisper.cpp，提供完全运行在用户设备上的实时按键说话听写，音频永远不会离开机器。"},"comparison":{"title":"Whisper vs 其他ASR系统","intro":"Whisper处在更大的语音识别系统生态中。下面是它与常见替代方案的比较。","cards":[{"title":"Google Cloud Speech-to-Text","body":"Google的云端ASR服务。精度优秀，尤其是英文，支持实时流式处理和话者分离。主要区别在于：它是专有服务，需要把音频发送到Google服务器，并按处理的音频分钟计费。Whisper在通用场景下精度相近，可本地运行且免费，但缺少内置流式处理和话者分离。","tags":["仅云端","专有","按量付费"]},{"title":"Amazon Transcribe (AWS)","body":"Amazon的云端ASR，与AWS生态紧密集成。提供医疗转写、呼叫分析和自定义词汇。取舍与Google类似：依赖云端、按分钟计费、专有。对于需要离线运行、开源许可或避免供应商锁定的应用，Whisper更合适。","tags":["仅云端","专有","AWS集成"]},{"title":"Mozilla DeepSpeech","body":"Mozilla基于Baidu Deep Speech研究的开源ASR引擎。它曾是最早的高质量开源ASR模型之一，但Mozilla在2021年停止了活跃开发。Whisper在精度、多语言支持和社区势能上已经实质性取代DeepSpeech。DeepSpeech仍可作为轻量、易理解的基线，但不再推荐用于新项目。","tags":["开源","已停止维护","偏重英文"]},{"title":"Vosk","body":"一个开源离线语音识别工具包，支持20多种语言，模型小而快。Vosk非常适合资源受限环境，它的模型只有Whisper的一小部分大小，并且能在低功耗硬件上良好运行。取舍是精度：Vosk明显不如Whisper，尤其在口音语音、噪声音频和专业词汇上。","tags":["开源","轻量","精度较低"]},{"title":"Faster Whisper (CTranslate2)","body":"使用SYSTRAN的CTranslate2推理引擎重新实现的Whisper。它运行相同的Whisper模型，精度相同，但速度最高可比原始OpenAI实现快4倍，内存使用也更低。Faster Whisper基于Python（不同于whisper.cpp的C/C++），是服务器端批量转写的强选择。它支持CUDA GPU和带INT8量化的CPU推理。","tags":["开源","Python / CTranslate2","相同Whisper精度"]}]},"applications":{"title":"实际应用","intro":"Whisper结合了精度、多语言支持和开源可用性，已经成为大量应用的基础：","cards":[{"title":"桌面听写","body":"<openWhisprLink>OpenWhispr</openWhisprLink>等应用使用whisper.cpp，在macOS、Windows和Linux上提供系统级按键说话听写。音频在本地处理，不会发送到云端。"},{"title":"字幕生成","body":"Whisper的时间戳预测非常适合生成字幕和闭合字幕。<stableTsLink>stable-ts</stableTsLink>和<autoSubtitleLink>auto-subtitle</autoSubtitleLink>等工具会自动化这一工作流。"},{"title":"播客和会议转写","body":"Whisper很擅长长音频转写，尤其是配合分块处理方式。许多播客编辑器和会议记录工具使用Whisper（或Faster Whisper）批量转写录音。"},{"title":"翻译和本地化","body":"Whisper内置的翻译能力（任意语言转英文）用于跨语言内容访问、媒体本地化和实时翻译原型。"},{"title":"无障碍","body":"面向聋人和听障用户的实时字幕、面向运动障碍用户的语音界面，以及音频描述转写，都是Whisper正在被部署的活跃领域。"},{"title":"数据标注和研究","body":"研究人员使用Whisper创建伪标注数据集来训练其他模型，这与OpenAI创建large-v3训练数据的方法相同。它也被广泛用于语言学研究和语料库构建。"}]},"sources":{"title":"来源和延伸阅读","items":[{"label":"[论文]","text":"Radford, A., Kim, J. W., Xu, T., Brockman, G., McLeavey, C., & Sutskever, I. (2022). Robust Speech Recognition via Large-Scale Weak Supervision. arXiv:2212.04356.","linkText":"arxiv.org/abs/2212.04356"},{"label":"[代码]","text":"OpenAI Whisper GitHub仓库。","linkText":"github.com/openai/whisper"},{"label":"[代码]","text":"whisper.cpp — Georgi Gerganov的C/C++移植版。","linkText":"github.com/ggml-org/whisper.cpp"},{"label":"[模型卡]","text":"Hugging Face上的Whisper large-v2（LibriSpeech test-clean WER基准：约3.0%）。","linkText":"huggingface.co/openai/whisper-large-v2"},{"label":"[模型卡]","text":"Hugging Face上的Whisper large-v3（相较v2错误减少10-20%）。","linkText":"huggingface.co/openai/whisper-large-v3"},{"label":"[模型卡]","text":"Hugging Face上的Whisper large-v3-turbo（面向低延迟的蒸馏变体）。","linkText":"huggingface.co/openai/whisper-large-v3-turbo"},{"label":"[模型卡]","text":"OpenAI Whisper模型卡（训练数据组成拆解）。","linkText":"github.com/openai/whisper/blob/main/model-card.md"},{"label":"[发布说明]","text":"OpenAI large-v3发布公告和评估说明。","linkText":"github.com/openai/whisper/discussions/1762"},{"label":"[发布说明]","text":"OpenAI turbo发布讨论和速度取舍。","linkText":"github.com/openai/whisper/discussions/2363"},{"label":"[代码]","text":"Faster Whisper — SYSTRAN基于CTranslate2的重新实现。","linkText":"github.com/SYSTRAN/faster-whisper"},{"label":"[产品]","text":"OpenWhispr通过whisper.cpp使用Whisper进行本地听写工作流。","linkText":"openwhispr.com"}],"links":["https://arxiv.org/abs/2212.04356","https://github.com/openai/whisper","https://github.com/ggml-org/whisper.cpp","https://huggingface.co/openai/whisper-large-v2","https://huggingface.co/openai/whisper-large-v3","https://huggingface.co/openai/whisper-large-v3-turbo","https://github.com/openai/whisper/blob/main/model-card.md","https://github.com/openai/whisper/discussions/1762","https://github.com/openai/whisper/discussions/2363","https://github.com/SYSTRAN/faster-whisper","https://openwhispr.com"]},"cta":{"title":"用OpenWhispr在本地试用Whisper","body":"OpenWhispr使用Whisper（通过whisper.cpp）在macOS、Windows和Linux上提供本地、私密的听写。按键说话、支持99+种语言、无需云端。免费试用。","starButton":"在GitHub上给我们加星","footnote":"无需账户 · 可离线使用 · 永远开源"},"internalLinks":[{"href":"/blog/whisper-model-sizes-explained","label":"Whisper模型尺寸对比"},{"href":"/blog/local-vs-cloud-transcription","label":"本地 vs 云端转写"},{"href":"/blog","label":"所有博客文章"},{"href":"/use-cases/developers","label":"面向开发者的OpenWhispr"},{"href":"/compare","label":"比较听写工具"}]},"blogVoicePrivacy":{"metadata":{"title":"语音数据隐私：让音频留在设备上 | OpenWhispr","description":"云端语音工具会记录你的声音。OpenWhispr 的本地 AI 转录让音频留在设备上，保持私密并完全离线；免费且开源。","keywords":["本地 AI 隐私","端侧 AI","语音数据隐私","本地语音识别","私密 AI 处理","边缘 AI 语音","本地 LLM","无需云端的 AI","语音数据安全","nvidia parakeet"]},"schema":{"title":"本地 AI 的理由：为什么你的语音数据不该离开你的设备","description":"AI 已经强大到足以在你的笔记本上运行。这里讲清楚为什么本地处理对隐私、安全和掌控权更有利。"},"breadcrumb":{"home":"首页","blog":"博客","article":"本地 AI 的理由"},"layout":{"title":"本地 AI 的理由：为什么你的语音数据不该离开你的设备","subtitle":"你口袋里的硬件，已经强大到足以运行可与云端服务匹敌的语音识别。那我们为什么还在把语音数据发往服务器？","date":"2026 年 2 月 12 日","tag":"博客"},"sections":[{"id":"opening","title":"概览"},{"id":"fact-check","title":"事实核查速览"},{"id":"local-ai-revolution","title":"本地 AI 革命"},{"id":"voice-sensitivity","title":"语音数据的敏感性"},{"id":"voice-in-the-cloud","title":"云端中的语音"},{"id":"performance-gap","title":"性能差距"},{"id":"regulatory-landscape","title":"监管格局"},{"id":"cloud-makes-sense","title":"何时该用云端"},{"id":"future-of-local-ai","title":"本地 AI 的未来"},{"id":"what-you-can-do","title":"你能做什么"},{"id":"sources","title":"参考来源"}],"opening":{"lead":"本地 AI 完全在你的设备上处理数据——无需联网，没有数据离开你的机器，也不必把你的信息托付给第三方。","body1":"借助诸如","whisperLink":"通过 whisper.cpp 高效运行的 OpenAI Whisper","body2":"以及 NVIDIA Parakeet 这样的模型，消费级笔记本如今已能以匹敌或接近云端服务的准确率转录语音。本文专门论述：为什么语音数据应当留在你的设备上——以及为什么让这一切成真的技术早已就绪。","lastUpdated":"最后更新：2026 年 2 月 17 日。文中的定量与法律性陈述均经过至少两个一手来源交叉核验。"},"factCheck":{"title":"事实核查速览（双重来源）","item1":{"lead":"在消费级硬件上运行本地 ASR 是切实可行的：","body":"Whisper + whisper.cpp 与 Parakeet 都有公开的模型/运行时文档，并附有硬件指引。"},"item2":{"lead":"云服务商的日志记录行为是可配置的，而非一刀切：","body":"数据处理方式取决于厂商默认设置以及选择加入/退出的选项。","link1":"Google 数据日志记录","link2":"AWS AI 退出政策"},"item3":{"lead":"语音数据可能触发严格的合规义务：","body":"生物识别与 AI 系统风险规则已被写入欧盟法律。","link1":"GDPR 官方文本","link2":"《欧盟 AI 法案》官方文本"},"item4":{"lead":"数据泄露的财务影响实实在在：","body":"泄露成本居高不下，这让集中存储敏感音频的代价更加沉重。","link1":"IBM 数据泄露报告","link2":"FTC 语音克隆风险警示"},"item5":{"lead":"OpenWhispr 背景：","body":"OpenWhispr 以本地优先转录为设计理念，云端使用为可选项。","link2":"whisper.cpp 后端"}},"revolution":{"title":"本地 AI 革命：发生了什么变化","p1":"五年前，在笔记本上运行一个有实用价值的 AI 模型还不切实际。模型太大，硬件太慢，软件生态几乎不存在。如今这一切已发生翻天覆地的变化。","p2a":"催化剂正是","p2b":"——Georgi Gerganov 用 C/C++ 移植的 OpenAI Whisper 语音识别模型。通过用纯 C++ 重写模型推理、且零运行时依赖，Gerganov 让 Whisper 得以在从 MacBook 到 Raspberry Pi 的各种设备上运行。该项目支持的模型从 75 MB 的「tiny」版本（约需 273 MB 内存），一直到磁盘占用 2.9 GB 的完整「large-v3」模型——并可通过整数量化进一步降低内存占用。它能在 Mac（Intel 与 Apple Silicon）、Linux、Windows、Android、iOS 上运行，甚至能在 WebAssembly 中运行。","p3a":"同样的思路在整个 AI 领域全面爆发。Gerganov 的","p3b":"之于大语言模型，正如 whisper.cpp 之于语音识别——它把 LLaMA、Mistral、Qwen、DeepSeek 以及数十个其他模型带到了消费级硬件上，量化精度从 8 位一路低至 1.5 位。诸如","p3c":"和","p3d":"等工具，则把这些能力封装进了易于上手的界面。","p4a":"硬件也跟了上来。Apple 自 2020 年的 M1 起推出的 M 系列芯片，为每台 Mac 都配备了统一内存架构和专用的 Neural Engine——让本地 AI 推理快到足以实时使用。Apple 在","p4b":"上更是大举投入：默认在设备端处理 AI 任务，仅在必要时才路由到 Apple 的 Private Cloud Compute 服务器——并以强有力的保证承诺数据绝不存储、仅用于当下这一次请求。","p5":"趋势已经很清楚：模型正在同时变得更小、更好。知识蒸馏（训练小模型来模仿大模型）和量化（在不破坏准确率的前提下降低数值精度）等技术，意味着 2023 年还需要数据中心 GPU 才能跑的东西，到 2026 年已能在笔记本上运行。例如，Whisper 的「large-v3-turbo」模型在保持相当准确率的同时，速度远快于其前代。"},"sensitivity":{"title":"为什么语音数据格外敏感","p1":"并非所有数据都同样敏感。文本可以匿名化，浏览历史可以清除。但语音不同——它是一种生物识别标识符。你的声音和指纹一样，独属于你。","p2":"一段你的录音所能揭示的，远不止你说出的那些字词。语音分析领域的研究已经证明，声音特征可以反映：","identity":{"title":"身份特征","item1":"你独一无二的身份（银行和安全系统都用声音做生物识别认证）","item2":"性别、大致年龄和地域口音","item3":"母语与社会经济背景"},"health":{"title":"健康与情绪状态","item1":"情绪状态——压力、焦虑、疲劳和抑郁都可从声音模式中检测出来","item2":"神经系统状况——已有研究将声音生物标志物用于帕金森病的早期检测","item3":"呼吸系统健康——新冠（COVID-19）、哮喘等状况会影响发声特征"},"p3":"这并非空想。已有公司正在积极开发分析语音的产品，用于健康筛查、保险风险评估和招聘决策。语音数据丰富的生物识别信息，正是它价值所在——也正是它一旦脱离你的掌控就变得危险的原因。","p4a":"与密码不同，一旦声音泄露，你无法更换自己的声音。一段录音一旦落到服务器上，你就永久失去了对这份生物识别数据的独占控制。而数据泄露并非杞人忧天——IBM 的《2025 年数据泄露成本报告》发现，全球单次泄露的平均成本","breachCost1":"488 万美元","p4b":"（2024 年）","breachCost2":"444 万美元","p4c":"（2025 年）","ibmCite":"IBM，2025 年","p4d":"问题不在于泄露是否会发生，而在于何时发生。"},"cloud":{"title":"你的语音在云端会经历什么","p1":"当你使用基于云端的语音识别服务时，你的音频会被传输到远程服务器进行处理。此后会发生什么，因服务商而异——而细节至关重要。","google":{"a":"Google 的文档指出，","bold1":"默认情况下，Cloud Speech-to-Text 不会记录客户的音频数据或转录文本","b":"。不过，Google 提供了一项自愿的数据日志记录计划，用户可选择加入，以共享音频数据换取折扣价。一旦加入，Google 会用这些数据来「提升服务质量」。值得注意的是，被记录的数据","bold2":"不会在你删除项目时一并删除","c":"——你必须另行提交删除请求。","cite":"Google Cloud 文档"},"aws":{"a":"包括 Amazon Transcribe 在内的 AWS AI 服务，","bold1":"可能会使用客户内容来开发和改进 AWS 服务","b":"，除非用户通过组织级策略明确选择退出。AWS 在 2023 年迫于公众压力更新了该策略，但在许多地区，默认设置仍允许将数据用于模型改进。","cite":"AWS Transcribe 文档"},"azure":{"a":"Azure 的 Speech Service","bold1":"实时处理音频，默认不会留存客户的音频数据","b":"。不过，选择加入自定义模型训练的用户，其数据将被存储。Microsoft 的数据处理受","cite":"《数据保护附录》","c":"约束，其条款因服务层级和地区而异。"},"p2a":"撇开服务商本身不谈，还要考虑基础设施链条。你的音频可能要经过多个网络节点跳转，在另一个国家的数据中心被处理，并有可能被分包处理方或承包商访问到。即便服务商政策再严格，存储在美国境内服务器上的数据，仍可能依据","effLink":"《国家安全信函》和《外国情报监视法》第 702 条","p2b":"被政府调取，而这些都无需通知数据主体。","callout":{"lead":"需要说明的是：","body":"这些服务商通常都有强有力的安全实践，其政策也有正当理由。重点不在于云服务居心不良——而在于把数据发给任何第三方，本质上就意味着要信任它们的政策、它们的安全和它们所处的司法管辖区。本地处理则彻底消除了这种信任的必要。"}},"performance":{"title":"性能差距正在缩小","p1":"支持云端转录的传统论据很简单：云端模型准确率高得多。如今这道鸿沟已大幅收窄。","p2":"2023 年底发布的 OpenAI Whisper large-v3，在大多数语言上的词错误率已能与商用云端 API 一较高下。NVIDIA 的 Parakeet 模型则把准确率推得更远，在标准英语基准测试中取得了一些最低的 WER。针对英语语音的独立基准测试显示，这些开源模型在干净语音上的词错误率（WER）落在 3% 至 5% 区间——与 Google 和 AWS 的商用产品相当甚至更优。更新的「large-v3-turbo」版本在保持相近准确率的同时速度大幅提升，让现代硬件上的实时本地转录变得切实可行。","stat1":"Whisper 模型尺寸（从 tiny 到 large）","stat2Value":"99+ 种语言","stat2":"Whisper 模型所支持","stat3Value":"实时","stat3":"在 Apple Silicon 和现代 x86 CPU 上","p3":"硬件加速带来了重大改变。whisper.cpp 在 Mac 上支持 ARM NEON、Apple 的 Accelerate 框架和 Metal GPU，在 NVIDIA GPU 上支持 CUDA，支持用于跨平台 GPU 推理的 Vulkan，甚至还为昇腾（Ascend）NPU 和 Intel 的 OpenVINO 提供了专门的后端。与 llama.cpp 一同开发的 GGUF 模型格式实现了高效量化——在准确率损失极小的前提下，将模型的内存占用减少 50% 至 75%。","p4":"具体到听写场景——录音通常只有 5 到 30 秒——在过去三年内生产的任何机器上，本地推理几乎都是瞬时完成的。「small」版 Whisper 模型（466 MB，约 852 MB 内存）对大多数语言都能提供出色的准确率，在 8 GB 内存的机器上也能轻松运行。你不需要高端 GPU，不需要游戏 PC，只需要一台还算现代的笔记本。","p5":"Apple 很早就看准了这一趋势。其内置于 iOS 和 macOS 的端侧语音识别，随每次系统更新都大幅改进——支持数十种语言的无网联听写。Apple 这样一家拥有庞大云端基础设施的公司都在把语音识别转向设备端，这本身就足以说明这项技术的走向。"},"regulatory":{"title":"监管格局","p1":"全球各地的隐私法规都在追赶一个现实：语音数据是敏感的。对于处理语音数据的组织而言，本地处理不仅是一种隐私偏好——它正日益成为一种合规优势。","gdpr":{"title":"GDPR（欧盟）","a":"根据《通用数据保护条例》，当语音数据被用于唯一识别某人时，它被归类为","bold1":"生物识别数据","b":"——从而被纳入个人数据的「特殊类别」，受","link":"第 9 条","c":"管辖。处理生物识别数据需要明确同意，或满足一小套法律依据之一。即便不用于识别身份，语音录音也属于个人数据，须遵循 GDPR 的数据最小化原则。本地处理可彻底规避诸多 GDPR 义务——如果数据从不离开设备，就不存在数据传输、第三方处理或跨境合规问题。"},"aiAct":{"title":"《欧盟 AI 法案》","a":"《欧盟 AI 法案》于 2024 年 8 月生效，相关条款将在 2026 年前分阶段实施。该法案把为识别目的而处理生物识别数据的 AI 系统归类为","bold1":"高风险","link":"《欧盟 AI 法案》","b":"。高风险系统面临大量要求，包括合规性评估、文档义务和持续监控。在公共场所进行实时生物识别在极少数例外情形之外被彻底禁止。虽然标准的语音转文字本身未必会触发高风险归类，但任何可能从声音识别说话人的系统——哪怕是无意为之——都将在这一框架下受到审视。"},"hipaa":{"title":"HIPAA 与职业特权","a":"在医疗场景中，包含患者信息的语音录音在美国受","b":"保护。将此类录音发往云服务需要签订《业务伙伴协议》（BAA）并遵守《安全规则》。同样，律师在口述涉及客户事务的笔记时，还须顾及","bold2":"律师-当事人特权","c":"——把受特权保护的通信发往第三方服务器会带来风险。本地处理可绕开这些问题：如果音频从不离开设备，就没有需要担心的第三方数据处理方。"},"ccpa":{"title":"CCPA / CPRA（加州）","a":"《加州消费者隐私法》（经《加州隐私权法案》修订）把语音和音频数据视为受监管的个人信息，并在数据被用于生物识别时附加额外规则","link1":"加州总检察长","link2":"CPPA 常见问题","b":"。收集语音数据的企业必须提供额外披露，并尊重消费者的删除权和退出权。美国还有多个其他州——包括伊利诺伊州（BIPA）、得克萨斯州和华盛顿州——也已制定各自的生物识别隐私法，其要求和执法机制各不相同。"},"p2":"监管方向很明确：全球范围内，语音数据正受到越来越严肃的对待。对于处理语音数据的组织——无论是在医疗、法律、金融还是任何受监管行业——本地处理在结构上提供了更简洁的合规姿态。没有需要审计的数据传输，没有需要审查的分包处理方，也没有需要论证正当性的跨境数据流动。"},"cloudSense":{"title":"何时云端仍然合理","p1":"实事求是很重要。本地 AI 并非总是正确的选择，假装不是这样反而会削弱上文那些站得住脚的论点。以下是云端处理仍是更优选择的几种情形：","case1":{"title":"超大规模实时流式处理","body":"呼叫中心需要同时处理数千路并发音频流，这离不开云端基础设施。没有任何单台设备能扛得住这样的吞吐量。"},"case2":{"title":"大规模说话人分离","body":"在多人录音中识别「谁说了什么」在计算上代价高昂。Google 和 AWS 的云端 API 在这方面，目前比大多数本地方案处理得更可靠。"},"case3":{"title":"小众语言","body":"虽然 Whisper 和 Parakeet 支持多种语言，但准确率参差不齐。对于 Whisper 处理欠佳的特定语言或方言，专业云服务商可能提供更好的模型。"},"case4":{"title":"在有限硬件上处理超长录音","body":"在低端笔记本上转录一段 4 小时的会议录音可能慢到令人抓狂。云端 API 几分钟就能处理完数小时的音频。"},"p2":"理想的做法往往是混合式：默认本地，按需云端。日常听写在本地处理，敏感内容留在设备上。而当你确实需要云端级算力或专业功能时，让它成为一个有意识的决定——而非一个看不见的默认值。"},"future":{"title":"本地 AI 的未来","p1":"硬件的发展轨迹强烈利好本地 AI。现代芯片正被专门设计来高效运行神经网络：","npu1":{"title":"Apple Neural Engine","body":"Apple 的专用神经处理单元，存在于每一颗 M 系列和 A 系列芯片中，在 M4 上可达每秒 38 万亿次运算（TOPS）。Apple Intelligence 高度依赖这一硬件来进行端侧处理。"},"npu2":{"title":"Qualcomm Hexagon NPU","body":"自 2024 年起为 Windows 笔记本提供动力的 Snapdragon X Elite 系列，配备了一颗 45 TOPS 的 NPU，专为本地 AI 工作负载而设计。Microsoft 的 Copilot+ PC 计划要求设备具备 NPU 能力。"},"npu3":{"title":"Intel AI Boost NPU","body":"Intel 的 Meteor Lake Core Ultra 处理器引入了约 11 TOPS 的专用 NPU，而更新的 Lunar Lake 一代则达到 48 TOPS。Intel 明确将这些定位为支撑「AI PC」体验的能力。"},"p2a":"在模型方面，蒸馏、剪枝和高效架构方面的研究仍在不断拓展本地可运行的边界。诸如","p2b":"之类的项目证明，专门打造的小模型能以一小部分算力成本达到大模型 99% 的准确率。","p3":"这种汇聚趋势不容错辨：每一家主要芯片厂商都在加入专用 AI 硬件，每一家主要操作系统厂商都在构建端侧 AI 功能，每一家主要模型实验室都在发布更小、更高效的模型。端侧 AI 处理不是一种替代方案——它正在成为默认选项。云端处理对于重型工作负载仍将举足轻重，但对于听写这类个人计算任务，未来属于本地。"},"actions":{"title":"你今天就能做的事","intro":"你不必坐等未来。本地 AI 当下就切实可行。以下是你可以采取的具体步骤：","step1":{"title":"用本地语音识别来听写","a":"提供由 OpenAI Whisper 和 NVIDIA Parakeet 驱动的按键说话听写，并配有完整的图形界面。如果你更喜欢命令行，","b":"也可直接使用。两者都完全在你的机器上运行。"},"step2":{"title":"在本地运行 LLM 处理敏感工作","a":"诸如","b":"和","c":"等工具，让你能在本地运行有实用价值的语言模型。对于涉及个人、财务或机密信息的查询，本地模型可彻底消除数据暴露的风险。"},"step3":{"title":"审视你现有的工具","body":"检查你所使用的每一项语音服务的隐私政策。你的听写应用会把音频发往服务器吗？你的虚拟助手会存储录音吗？你的会议转录工具会用你的内容做训练吗？答案也许会让你大吃一惊。"},"step4":{"title":"优先选择提供本地处理的应用","body":"在挑选新工具时，优先选择那些提供端侧处理的产品。这种能力已经存在——去要求它。每一位选择本地处理的用户，都在向市场发出一个信号：隐私很重要。"}},"sources":{"title":"参考来源与延伸阅读","s1":"OpenAI Whisper 的 C/C++ 移植版。涵盖模型尺寸、硬件要求和支持的平台。","s2":"原始 ASR 模型发布、训练方法和评估背景。","s3Title":"NVIDIA Parakeet 模型卡","s3":"用于本地语音识别的开放 ASR 指标和基准测试细节。","s4":"C/C++ 的 LLM 推理。在数十种模型架构上支持 1.5 位至 8 位量化。","s5Title":"Google Cloud Speech-to-Text 数据日志记录","s5":"默认不记录政策及选择加入数据计划的细节。","s6Title":"AWS Transcribe 安全文档","s6":"数据处理、退出政策和加密实践。","s7Title":"《2024 年 IBM 数据泄露成本报告》","s7":"全球泄露成本、平均暴露记录数及行业分析。","s8Title":"GDPR 第 9 条","s8":"特殊类别个人数据（含生物识别数据）的处理。","s9Title":"《欧盟 AI 法案》","s9":"将处理生物识别数据的 AI 系统归类为高风险。","s10":"针对个人及敏感数据处理的加州隐私指引。","s11Title":"Azure Speech 数据隐私","s11":"Microsoft 服务层级的语音数据处理文档。","s12":"端侧处理与 Private Cloud Compute 架构。","s13Title":"用于抑郁检测的声音生物标志物","s13":"Low、Bentley、Ghosh（2020）。利用语音对精神疾病进行自动化评估。","s14Title":"EFF 关于 FISA 与《国家安全信函》","s14":"政府对美国公司所存储数据的调取。"},"cta":{"title":"你的声音，理应属于你自己","body":"OpenWhispr 默认在本地处理一切。除非你明确选择云端处理，否则没有任何音频会离开你的设备。","subtext":"开源。由 OpenAI Whisper 和 NVIDIA Parakeet 驱动。支持 macOS、Windows、Linux 和 iOS。","starButton":"在 GitHub 上给我们点 Star","footnote":"无需账户 · 离线可用 · 永久开源"},"links":{"openSource":"开源为何重要","localVsCloud":"本地转录 vs 云端转录","allArticles":"全部文章","offline":"离线使用场景","medical":"医疗听写"}},"blogDiarization":{"metadata":{"title":"说话人分离：100% 本地且私密 | OpenWhispr","description":"了解 OpenWhispr 如何用 pyannote、CAM++ 嵌入和 sherpa-onnx 在本地进行说话人分离。隐私优先、开源且免费。","keywords":["本地说话人分离","设备端说话人分离","隐私会议转录","隐私会议记录工具","本地会议笔记","离线会议转录","开源会议助手","开源会议记录工具","开源 AI 会议笔记","音频永不离开设备","隐私 AI 会议助手","自托管会议转录","说话人识别隐私","pyannote 分割","cam++ 说话人嵌入","sherpa-onnx 分离","本地声纹","谁说了什么转录","Otter.ai 隐私替代","Otter.ai 开源替代","Granola 开源替代","Fireflies 设备端替代"]},"schema":{"title":"本地说话人分离：我们如何打造 100% 隐私的会议记录工具","description":"完整技术解析 OpenWhispr 的本地说话人分离——pyannote、CAM++、sherpa-onnx。你的音频永不离开设备。"},"breadcrumb":{"home":"首页","blog":"博客","article":"本地说话人分离"},"hero":{"title":"本地说话人分离：我们如何打造 100% 隐私的会议记录工具","subtitle":"OpenWhispr 是一款音频永不离开设备的开源会议助手。每一段声纹、每一个说话人标签、每一个嵌入向量——全部在你自己的机器上处理和存储。","date":"2026 年 4 月 15 日","tag":"技术"},"sections":[{"id":"what-is-diarization","title":"什么是说话人分离"},{"id":"pipeline","title":"四阶段流程"},{"id":"vad","title":"第一阶段：语音活动检测"},{"id":"segmentation","title":"第二阶段：分割"},{"id":"embeddings","title":"第三阶段：说话人嵌入"},{"id":"clustering","title":"第四阶段：聚类"},{"id":"live","title":"实时分离"},{"id":"profiles","title":"声纹档案"},{"id":"privacy","title":"隐私"},{"id":"limitations","title":"坦诚的局限"},{"id":"why-sherpa","title":"为什么选择 Sherpa-ONNX"},{"id":"faq","title":"常见问题"}],"opening":{"lead":"本地说话人分离是指为会议录音标注“谁在何时说话”的过程，全程在用户设备上运行，不向任何外部服务器传输音频、嵌入向量或转录文本。","body":"会议转录有一个“谁说了什么”的难题。Zoom 只给你一整面文字墙。大多数 AI 记录工具——Otter.ai、Fireflies、Read、Granola——的解决办法都是把你的会议音频上传到它们的服务器，在那里运行说话人模型，再把带标签的片段发回来。你的声音，以及你同事的声音，最终都变成了别人数据库里的向量。我们不想要这样，我们的用户也不想要。OpenWhispr 是一款建立在相反假设之上的开源会议记录工具：音频永不离开设备，说话人声纹存放在你磁盘上的本地 SQLite 文件中，除非你明确选择，否则会议的任何内容都不会被上传。本文将从技术层面解析这一承诺中的分离部分——它是什么、我们选了哪些模型、它们如何协同工作，以及每个步骤中每一字节数据究竟存放在哪里。","footnote":"最后更新：2026 年 4 月 15 日。实现细节参考的是 2026 年 4 月 14 日合并进 OpenWhispr 桌面应用的说话人分离功能。下文的每一项技术声明都来自一手参考资料，或可直接在开源仓库中查验。"},"factCheck":{"title":"事实核查快照（双重来源）","items":[{"lead":"说话人分离回答的是“谁在何时说话”。","body":"它的评估基准是 NIST Rich Transcription 等社区评测以及 pyannote.audio 参考流程。","sources":[{"href":"https://www.nist.gov/itl/iad/mig/rich-transcription-evaluation","label":"NIST RT 评测"},{"href":"https://arxiv.org/abs/1911.01255","label":"pyannote.audio 论文（Bredin 等，2020）"}]},{"lead":"我们使用 pyannote-segmentation-3.0，以及来自 3D-Speaker 项目的 CAM++。","body":"两者都是开源的，都可移植为 ONNX，都直接从 sherpa-onnx 模型发布页下载。","sources":[{"href":"https://huggingface.co/pyannote/segmentation-3.0","label":"pyannote-segmentation-3.0 模型卡"},{"href":"https://arxiv.org/abs/2303.00332","label":"CAM++ / 3D-Speaker 论文（Wang 等，2023）"}]},{"lead":"所有推理都通过 sherpa-onnx 运行——一个原生 ONNX 二进制文件，无需 Python 运行时、无需 PyTorch、无需 GPU。","body":"我们在构建时锁定该二进制文件及其依赖的版本。","sources":[{"href":"https://github.com/k2-fsa/sherpa-onnx","label":"sherpa-onnx（GitHub）"},{"href":"https://github.com/k2-fsa/sherpa-onnx/releases","label":"sherpa-onnx 发布页"}]},{"lead":"声纹是 512 维的 float32 向量，作为本地 SQLite BLOB 存储，","body":"每位说话人约 2KB。它们永不离开设备。","sources":[{"href":"https://www.sqlite.org/datatype3.html","label":"SQLite 数据类型文档"},{"href":"https://onnxruntime.ai/","label":"ONNX Runtime"}]},{"lead":"分离过程中不会向设备外传输任何音频、嵌入向量或转录文本。","body":"整个功能唯一的网络活动是首次启动时的一次性模型下载，你可以亲自在开源仓库中查验。","sources":[{"href":"https://openwhispr.com/privacy","label":"OpenWhispr 隐私政策"},{"href":"https://github.com/OpenWhispr","label":"OpenWhispr（GitHub）"}]}]},"whatIs":{"title":"说话人分离究竟是什么","lead":"说话人分离是为录音标注“谁在何时说话”的过程——而不是他们说了什么。","paragraph1":"转录回答“说了什么”；分离回答“谁在说”。两者相辅相成：一款好的会议记录工具两者都需要，而让它们协同工作比单独搞定任何一个都要难。","paragraph2":"分离之所以真正困难，源于几个顽固的原因。同一通话里两个人的声音可能重叠。背景噪声会渗进来。简短的发言（“嗯”“好的”）几乎不携带说话人信息。新的说话人会在会议进行到一半时出现。而且与转录不同，分离必须做出全局决策——它无法在不与第二段对比的情况下为第七段贴标签。正是这种全局语境，使得大多数商业方案选择在服务器端针对完整录音做分离，再把带标签的转录发回来。","paragraph3":"仅有转录是不够的。一场十人参加、四十五分钟充满交叉对话的会议，如果呈现为一面无差别的文字墙，根本无法阅读。会议笔记——人们真正想要的成果——需要说话人归属，因为待办事项、决定和承诺都是挂在具体的人身上的。“Alice 会在周五前交付迁移”是一条笔记。“会在周五前交付迁移”是噪声。","calloutTitle":"分离不等于说话人识别","calloutBody":"分离说的是“这三段是说话人 A；这两段是说话人 B”。它并不知道说话人 A 是谁。说话人识别则通过把声纹与已存档案匹配，给说话人 A 安上一个名字。OpenWhispr 两者都做——分离在每场会议上运行，而一旦你为某人标注了名字，识别就开始介入。两者都保持在本地。"},"pipeline":{"title":"四阶段本地流程","lead":"OpenWhispr 的说话人分离流程分为四个阶段：语音活动检测、分割、说话人嵌入和聚类——随后还有一个合并步骤，把说话人片段与转录文本对齐。","paragraph1":"每个阶段都是一个独立的 ONNX 模型或原生二进制文件。无需 Python 运行时、无需 PyTorch、无需 CUDA。所有模型在磁盘上的总占用约为 45MB。","paragraph2":"音频采集从第一天起就是双流的。麦克风按定义就是“你”，所以我们从不浪费算力去分离你自己的声音——每一段麦克风片段都按来源标记为“你”，而非靠声音匹配。系统音频（通话中的其他所有人）才是唯一真正需要做说话人切分的东西，这一下就把工作量减半了。","paragraph3":"音频采集完成后，四个阶段依次运行。VAD 去除静音，让昂贵的嵌入阶段不会把算力浪费在空帧上。分割把连续语音切成单说话人的片段。嵌入模型把每个片段转换为一个 512 维的声纹。凝聚聚类把相似的声纹归为一组，得出最终的一组说话人 ID。最后一道工序按时间戳重叠把这些 ID 合并回转录文本，于是每个词都被打上了一个说话人标签。","paragraph4Prefix":"整套流程在一个从应用派生出的子进程内运行——没有服务器，没有监听端口，也没有任何指向 Electron 之外的 IPC。你可以通过阅读开源仓库中的","paragraph4Suffix":"来确认这一点。"},"vad":{"title":"第一阶段：语音活动检测（Silero）","lead":"语音活动检测（VAD）是在昂贵阶段运行之前，把语音和静音区分开的廉价过滤器。","paragraph1":"对静音运行说话人嵌入模型既浪费 CPU 又会产出无用的向量。VAD 正是让流程其余部分保持快速的关键。","paragraph2Prefix":"我们使用","sileroLinkLabel":"Silero VAD","paragraph2Suffix":"：一个 2MB 的 ONNX 模型，采用 MIT 许可，实际上已成为开源行业的默认选择。Silero 会针对每一帧返回当前 32 毫秒窗口包含语音的概率。在现代 CPU 上，它的开销微不足道——每个 32 毫秒片段约 0.1 毫秒，大约相当于单核性能的千分之三。","paragraph3":"实时流程在通话期间对系统音频持续运行 Silero。我们实际发布时采用的阈值是针对真实会议音频调校的，而非干净的基准测试：","list":[{"label":"窗口大小：","value":"512 个采样点（16kHz 下为 32 毫秒）"},{"label":"语音阈值：","value":"0.15——刻意设得激进，以免漏掉音量低或距离远的说话人"},{"label":"静音阈值：","value":"0.08"},{"label":"片段结束条件：","value":"连续 16 个静音窗口（约 512 毫秒）"},{"label":"嵌入所需最小片段：","value":"0.8 秒"},{"label":"实时识别节奏：","value":"一旦累积到 ≥1.6 秒的语音，每 1 秒一次"}],"calloutTitle":"坦诚的取舍","calloutBody":"0.15 这一激进的语音阈值能捕捉到远处麦克风上的轻声说话者，但偶尔也会被响亮的键盘敲击声或房间噪声误触发。0.8 秒的最小片段过滤器会在这些误报到达嵌入阶段之前剔除其中大部分，从而保护了下游的准确率。"},"segmentation":{"title":"第二阶段：分割（pyannote 3.0）","lead":"分割负责把连续语音切成各自只包含一个说话人的片段，同时检测说话人切换点和重叠区域。","paragraph1":"它是决定说话人边界在哪里的阶段。之后的所有环节都假定这些边界是正确的。","paragraph2Prefix":"我们使用","paragraph2Mid":"，这是 pyannote.audio 团队该模型的当前版本，由","paragraph2Suffix":"项目导出为 ONNX。Pyannote 是事实上的学术基准：pyannote.audio 流程在 AMI 和 CALLHOME 等标准基准上的分离错误率处于 12–15% 区间——正是商业云服务商爱引用的那组数字。","paragraph3Prefix":"为什么偏偏要 ONNX 移植版？因为在 Electron 应用里塞进 PyTorch 根本行不通——光是运行时就约 1.5GB，而且实际上要有 GPU 才够快。ONNX 导出版是一个 6.6MB 的单文件，通过","paragraph3Suffix":"在 CPU 上运行。同样的模型，重量却只有一小部分。","paragraph4Prefix":"我们通过 sherpa-onnx 的离线分离二进制文件配上几个参数来调用它。最小时长设置取自 pyannote 论文推荐的默认值：短促的语音爆发会被并入相邻的静音，而短暂的静音不会打断一个连续的说话人片段。输出是一组带临时说话人 ID 标记的起止时间对（","paragraph4Suffix":"等等）。"},"embeddings":{"title":"第三阶段：说话人嵌入（CAM++）","lead":"说话人嵌入是一串定长的数字，它代表一段声音，就像哈希值代表一个文件。","paragraph1":"同一声音的两段录音会产生在向量空间中相近的嵌入；不同的声音则产生彼此相距很远的嵌入。“相近”用余弦相似度来衡量——一个介于 -1 和 1 之间的单一数值。这是流程中其余一切的数学核心。","paragraph2Prefix":"我们的嵌入模型是来自阿里巴巴达摩院 3D-Speaker 项目的 CAM++，在","paragraph2Mid":"说话人验证基准上训练。我们实际采用的具体文件是","paragraph2Mid2":"，一个约 28MB 的 ONNX 导出版，为每个片段产出一个 512 维的 float32 向量。论文是","camppPaperLabel":"arXiv:2303.00332（Wang 等，2023）","paragraph2Suffix":"。","paragraph3Prefix":"我们选择 CAM++ 而非那个显而易见的替代——","paragraph3Suffix":"——出于三个具体原因：参数量大约只有一半、在 VoxCeleb1-O 上的等错误率更低、CPU 推理更快。NVIDIA NeMo 的 MSDD 模型没有官方 ONNX 导出，且严重依赖 PyTorch。Picovoice Falcon 带有按席位收费的商业许可。Apple 的 CoreML Speech 框架仅限 macOS。CAM++ 是唯一满足我们全部约束的选项——跨平台、ONNX、开放许可、CPU 上速度快。","paragraph4Prefix":"为了给 CAM++ 喂数据，我们在","paragraph4Mid":"中自带了一个 log-mel 滤波器组特征提取器：80 个梅尔频带、25 毫秒窗口、10 毫秒帧移，都是这类模型的标准参数。特征通过","paragraph4Suffix":"流入 ONNX 模型，在 Electron 主进程中运行。另一端出来的是一个 512 项的 Float32Array，我们把它作为 2048 字节的 BLOB 写入 SQLite。","paragraph5":"匹配步骤本身是最简单不过的事情：两个向量之间的余弦相似度。下面就是决定两个会议片段是否为同一说话人的实际代码：","calloutTitle":"为什么是 512 维？","calloutBody":"512 是当前 VoxCeleb 说话人验证排行榜上准确率与存储成本之间的最佳平衡点。256 维在区分相近但不同的声音时开始丧失辨别力；1024 维则增加了存储和算力，回报却递减。512 个 float32 × 2KB × 一千场各有十名说话人的会议，磁盘占用约 20MB。这在一个本地 SQLite 文件里绰绰有余。"},"clustering":{"title":"第四阶段：聚类——“有多少个说话人？”的难题","lead":"对每个片段做完嵌入后，你手上有 N 个向量。聚类把它们归为 K 个说话人——但难点在于你通常事先并不知道 K 是多少。","paragraph1":"一通销售电话可能有 2 个说话人。一场全员大会可能有 8 个。固定 K 值的聚类在两端都会失灵。","paragraph2":"我们使用余弦相似度阈值为 0.5 的凝聚聚类：先把每个片段各自当作一个簇，合并相似度最高的两个簇，重复进行，直到没有任何一对的相似度超过阈值为止。决定何时停止的是阈值，而非某个目标数量。这样就能从两个说话人到十个自然扩展，无需任何配置。","paragraph3Prefix":"实际的调用位于","paragraph3Suffix":"。它派生 sherpa-onnx 二进制文件，附带几个命令行参数：","paragraph4Prefix":"输出是纯文本：每个片段一行，格式为","paragraph4Mid":"。我们把它解析成一个对象数组，然后遍历转录文本，把每个转录片段匹配到时间重叠最大的分离片段。来自麦克风的片段始终被标记为","paragraph4Suffix":"——来源永远压过声音。","paragraph5Prefix":"0.5 这个阈值不是凭空猜的。我们在真实会议录音上、用自己的评测框架（","paragraph5Suffix":"）对它进行了调校。阈值过高会欠聚类——一个说话人被拆成两个。阈值过低会过聚类——两个相似的声音被合并。0.5 是我们最终收敛到的最佳平衡点。"},"live":{"title":"实时分离：边说边标注","lead":"纯批处理分离很准确，但在通话过程中是看不见的。纯实时则即时却嘈杂。OpenWhispr 两者都跑：实时标签在会议进行中出现，会后再用一次完整的批处理来优化它们。","paragraph1Prefix":"实时路径位于","paragraph1Mid":"。它以 16kHz 接入系统音频流，把每个 32 毫秒帧送进 Silero VAD，并不断累积语音帧，直到某个片段闭合（在约 512 毫秒的静音之后）或达到实时识别节奏（一旦累积到 ≥1.6 秒，每 1 秒一次）。一旦触发，它会提取一个 CAM++ 嵌入，将其与内存中当前通话嵌入及已存说话人档案的映射进行比对，并触发一个 IPC 事件——","paragraph1Suffix":"——附带最佳匹配。React 端接收到后，原地更新转录气泡。","paragraph2Prefix":"一旦有人在界面中手动设定了某个说话人的名字，这个指派就被","lockedWord":"锁定","paragraph2Mid":"。会后的批处理不得覆盖它，哪怕它自己的分析结果有异议。这一锁定行为位于","paragraph2Suffix":"，而它很重要：最糟糕的用户体验，就是让一次自动处理悄悄覆盖掉用户的修正。","paragraph3":"录制停止后，我们会对完整的 WAV 文件运行完整的离线分离。批处理拥有完整语境——它看到的是整场会议，而不只是最后的 1.6 秒——因此能产出更干净的最终标注。批处理结果会与实时标签对齐，尊重用户的每一处锁定，并用更准确的分组替换掉临时的猜测。","calloutTitle":"为什么混合模式胜过单独任何一种","calloutBody":"纯实时很快，但在短片段上很嘈杂。纯批处理很准确，却要等到会议结束才出声。混合模式让你两者兼得——通话中即时反馈，以及一份反映完整语境分析的最终转录。代价只是多一步对齐，它在后台运行一次，实际上几乎不耗成本。"},"profiles":{"title":"声纹档案：跨会议记住说话人","lead":"一旦你在某场会议里把某位说话人标注为“Alice”，OpenWhispr 就会在之后的每一场会议中自动为她标注——你无需再做任何事，而且她的声纹永远不会离开你的设备。","paragraph1":"机制很直接：每位说话人的质心嵌入都会存入你的本地 SQLite 数据库。当未来某通话出现新的嵌入时，我们用余弦相似度把它与每一个已存档案比对。相似度超过阈值的匹配会自动指派名字。这个过程完全不涉及服务器。支撑它的三张表都是普通的 SQLite：","matchingIntro":"匹配分三档，以在置信度和误报之间取得平衡：","matchingTiers":[{"label":"余弦相似度 ≥ 0.70：","body":"自动确认。标签立即出现，无需任何提示。"},{"label":"0.55 ≤ 余弦相似度 < 0.70：","body":"建议。界面会显示“这是 Alice 吗？”并附带确认 / 忽略按钮，等待你的输入。"},{"label":"余弦相似度 < 0.55：","body":"保持匿名。该片段维持为“说话人 N”，直到你手动为其命名。"}],"paragraph2Prefix":"当你确认一个匹配时，已存的质心会通过滑动平均来更新，而非直接覆盖。公式很简单：","paragraph2Suffix":"。这能应对声音漂移——一个寒冷的早晨、一支不同的麦克风、一根坏掉的线缆——而不会破坏未来的匹配。系统见过 Alice 声音的样本越多，她的档案就越稳定。","paragraph3Prefix":"新档案还会触发","retroactiveWord":"追溯重新标注","paragraph3Suffix":"。当你第一次为某位说话人命名时，一个后台任务会遍历过去每场会议已存的嵌入，找出匹配，并更新历史转录中此前未命名的说话人。被用户锁定的映射永远不会被触碰。结果是：为 Alice 命名一次，就能追溯地在她参加过的每一通话中为她标注。","paragraph4Prefix":"如果连接了 Google Calendar 且会议有参会者，说话人选择器界面会预先填入他们的姓名和邮箱。一键即可把","speakerTwo":"说话人 2","paragraph4Suffix":"映射好并永久保存——参会者与声纹档案保持关联，因此即使换了设备，正确的名字也会出现。"},"privacy":{"title":"你的数据究竟存放在哪里","lead":"音频永不离开设备。这不是一句营销口号——它是代码的机械属性，而且由于 OpenWhispr 是开源的，你可以亲自验证。","intro":"下面就是每一类数据在每个步骤究竟会发生什么。","tableHeaderWhat":"数据","tableHeaderWhere":"存放位置","tableHeaderLeaves":"是否离开设备？","tableRows":[{"what":"原始会议音频","where":"操作系统临时目录，分离完成后删除","leaves":"否——永不","leavesEmphasis":true},{"what":"分离用 ONNX 模型","where":"~/.cache/openwhispr/diarization-models/","leaves":"首次启动时下载一次，之后再不下载"},{"what":"说话人嵌入（声纹）","where":"本地 SQLite BLOB","leaves":"否——永不","leavesEmphasis":true},{"what":"说话人到姓名的映射","where":"本地 SQLite 表","leaves":"否——永不","leavesEmphasis":true},{"what":"转录文本","where":"本地 SQLite（如你开启，可选云同步）","leaves":"仅在你主动选择时"}],"paragraph1Prefix":"会议的原始 PCM 音频会被写入你操作系统临时目录中的一个文件，交给本地派生的 sherpa-onnx 二进制文件，并在分离完成时于一个","paragraph1Mid":"块中删除。OpenWhispr 的分离流程中不存在任何打开网络套接字上传音频的代码路径。由于代码是开源的，你可以在","paragraph1Mid2":"、","paragraph1Mid3":"中 grep","paragraph1Mid4":"或","paragraph1Suffix":"，亲自确认这一点。","paragraph2Prefix":"Electron 主进程与渲染进程之间的 IPC 使用内置的","paragraph2Suffix":"——它是单个进程内的内核级管道，而非网络套接字。说话人嵌入和声纹档案作为 SQLite BLOB，与你的笔记存储在同一个数据库文件中，享有相同的操作系统级文件权限。","calloutTitle":"一个例外，开诚布公","calloutBody":"这个功能唯一的网络活动，是首次启动时的一次性模型下载——从 sherpa-onnx 的 GitHub 发布页 CDN 获取三个 ONNX 文件。我们明确说明这一点，是因为“安装时一次网络调用、之后零调用”与“始终在线的云端推理”在隐私层面是截然不同的两个故事；而作为一款开源会议助手，我们宁愿主动告诉你，也不愿让你自己去发现。"},"limitations":{"title":"坦诚的局限","lead":"本地分离很好，但并不完美。下面是它确实力不从心的几种情况，以及我们针对每一种所做的应对。","items":[{"label":"说话人重叠。","body":"两个人同时讲话，对单一标签归属来说本质上是有损的。Pyannote 3.0 能检测出重叠区域，但每个片段只有一个 speaker_id，不足以表示“两个人同时在说话”。我们会把这些片段标为临时，让你在界面中加以修正。"},{"label":"冷启动准确率。","body":"与一位新同事的第一场会议只能依靠 VoxCeleb 训练出的通用特征。一旦你为他标注过一次、档案积累了几个样本，后续会议就会明显更准。"},{"label":"简短发言。","body":"“嗯”、“好的”、一声短促的笑——任何短于约 0.8 秒的内容都无法产生可靠的嵌入。我们会把这些片段从嵌入阶段剔除，转而依据周围上下文做标签传播。"},{"label":"远场或低信噪比音频。","body":"在嘈杂房间里用免提电话的人，质量会优雅但明显地下降。自动增益控制和 VAD 调校会有所帮助，但再精巧的模型也无法完全克服糟糕的物理条件。"},{"label":"仅 CPU 推理。","body":"GPU 会更快，但我们刻意不要求它。实际上，现代 CPU 的吞吐量已经足够：每个嵌入约 100 毫秒，在 M1 Mac 上对一场 45 分钟的会议做批处理分离约需 30 秒。"}],"outroPrefix":"我们在开源仓库中公开了自己的评测框架（","outroSuffix":"），这样你就能用自己的录音亲自测量上述任何一项——而不是我们精挑细选的演示片段。这比营销材料里一个光鲜的数字要诚实得多。"},"whySherpa":{"title":"我们为什么选择 Sherpa-ONNX","lead":"我们需要一个跨平台、无需 Python、CPU 上速度快且开源的分离运行时。Sherpa-ONNX 是唯一四项全中的选项。","paragraph1":"我们的要求很具体。它必须能在 macOS（Intel 和 Apple Silicon 都要）、Windows x64 和 Linux x64 上运行，且无需分别维护不同的代码路径。它不能捆绑 PyTorch——光是运行时就约 1.5GB，而且要在交互级延迟下可用基本上得有 GPU。它不能要求 CUDA。而且许可证要足够宽松，能在一款“商用但开源”的桌面应用中发布。","evaluatedIntro":"以下是我们实际评估过的选项：","evaluated":[{"label":"pyannote.audio（Python）：","body":"排除。在 Electron 应用里捆绑 PyTorch 根本行不通。"},{"label":"NVIDIA NeMo MSDD：","body":"依赖 PyTorch，无官方 ONNX 导出，以 GPU 为中心。"},{"label":"Picovoice Falcon：","body":"按活跃用户收费的商业许可，外加首次运行时需云端激活。"},{"label":"Apple CoreML + Speech 框架：","body":"仅限 macOS，没有跨平台对等性。"},{"label":"sherpa-onnx","body":"Apache 2.0 许可，原生 ONNX 二进制文件，约 35MB 的模型，派生进程读取 stdout 的接口，由 k2-fsa 积极维护。同一套工具链也支撑着其他已发布产品中的生产级 ASR。"}],"outroPrefix":"一个坦诚的提醒：sherpa-onnx 的离线说话人分离二进制文件相对较新。我们将版本锁定在","outroSuffix":"，并在升级锁定版本之前，针对每次更新运行我们完整的评测框架。对于这样一个承重的部件，这是正确的取舍。"},"faqHeading":"常见问题","faqs":[{"question":"有没有一款不会把我的音频上传到云端的隐私会议记录工具？","answer":"有。OpenWhispr 是一款免费、开源的会议记录工具，你的音频永不离开设备。转录在本地用 OpenAI Whisper 或 NVIDIA Parakeet 运行，说话人分离在本地用 pyannote 分割和 CAM++ 嵌入运行，声纹则存储在你自己机器上的本地 SQLite 文件中——而不是我们的服务器上。"},{"question":"Otter.ai、Fireflies 或 Granola 能看到我的会议吗？","answer":"能。基于云的 AI 记录工具会把你的会议音频上传到它们的服务器，以便转录和分离。说话人标签、转录文本和声音嵌入都在它们的基础设施中计算，并存储在它们的数据库里。如果你需要一个能把每一字节音频都留在设备上的替代方案，你需要的是 OpenWhispr 这样的本地优先工具。"},{"question":"什么是说话人分离？","answer":"说话人分离是为录音标注“谁在何时说话”的过程，与回答“说了什么”的转录有所区别。一份完成分离的转录会为每段发言打上说话人标签——Alice、Bob、说话人 3——而不是一整面无差别的文字墙。"},{"question":"本地说话人分离真的管用吗，还是说它是云端的降级版？","answer":"它管用。OpenWhispr 的流程——pyannote-segmentation-3.0 加上来自 3D-Speaker 项目的 CAM++ 嵌入，通过 sherpa-onnx 运行——在标准基准上的分离错误率与商业云 API 处于同一水平。主要的坦诚取舍在于对重叠说话人的处理，以及对系统从未听过的声音的冷启动准确率。"},{"question":"本地分离模型占多少磁盘空间？","answer":"总共约 45MB：pyannote 分割（6.6MB）加上 CAM++ 说话人嵌入模型（28MB）再加 Silero VAD（2MB）。这三者会在首次启动时下载一次，并缓存到 ~/.cache/openwhispr/diarization-models/。此后它们就留在磁盘上，再也不会重新下载。"},{"question":"运行 OpenWhispr 的本地分离需要 GPU 吗？","answer":"不需要。一切都通过 ONNX Runtime 在 CPU 上运行。现代 Apple Silicon 和较新的 x86 CPU 都能轻松应对——实际上，在 M1 Mac 上对一场 45 分钟的会议做批处理分离约需 30 秒，而通话过程中实时说话人标签会在一两秒内出现。"},{"question":"OpenWhispr 真的是开源的吗？","answer":"是的。分离代码、转录流程以及桌面应用的其余部分都在 GitHub 上开源。本文中的隐私声明可直接验证：在分离辅助代码里 grep fetch、axios 或 http，你在分离路径中找不到任何网络调用。"},{"question":"通话中两个人同时讲话会怎样？","answer":"Pyannote 3.0 会明确检测并标注重叠区域，但重叠期间的单一标签归属本质上是有损的。我们会把这些片段在转录中标为临时，让你通过点击说话人标签来修正。你的手动修正会被锁定，永远不会被之后的自动处理覆盖。"}],"cta":{"title":"试试这款隐私会议记录工具","body":"OpenWhispr 是一款免费、开源的会议助手，你的音频永不离开设备。本地分离、本地转录、本地声纹档案。支持 macOS、Windows 和 Linux。","starButton":"在 GitHub 上给我们点 Star","footnote":"无需账号 · 离线可用 · 永久开源"},"internalLinks":[{"href":"/blog/local-vs-cloud-transcription","label":"本地转录与云端转录对比"},{"href":"/blog/how-whisper-ai-works","label":"Whisper AI 的工作原理"},{"href":"/blog/local-ai-voice-data-privacy","label":"为何选择本地 AI"},{"href":"/blog/why-open-source-voice-tools","label":"为何选择开源语音工具"},{"href":"/blog","label":"全部博客文章"}]},"blogLocalVsCloud":{"metadata":{"title":"本地与云端语音转文字：隐私、速度和成本","description":"从隐私、准确率、速度和成本对比本地与云端语音转文字，了解为何设备端转录能让音频保持私密且免费。","keywords":["本地 vs 云端转录","本地语音转文字","云端转录隐私","离线转录","本地语音识别","隐私语音转文字","设备端转录","云端 ASR vs 本地 ASR","nvidia parakeet","whisper vs parakeet"]},"schema":{"title":"本地 vs 云端转录：隐私、速度与准确性全面对比","description":"对本地与云端语音转文字进行客观对比——隐私、延迟、准确性、成本与离线能力。"},"breadcrumb":{"home":"首页","blog":"博客","article":"本地 vs 云端转录"},"sections":{"factCheck":"事实核查速览","howCloudWorks":"云端转录","howLocalWorks":"本地转录","privacy":"隐私","accuracy":"准确性","speed":"速度与延迟","cost":"成本","offline":"离线能力","comparison":"对比表","recommendations":"推荐建议","sources":"参考来源"},"hero":{"title":"本地 vs 云端转录：隐私、速度与准确性全面对比","subtitle":"你的语音数据很有价值。它的去向，取决于你在哪里处理它。","date":"2026 年 2 月 8 日","tag":"对比"},"intro":{"lead":"本地转录会在你的设备上完整处理音频","body1":"使用 OpenAI Whisper 和 NVIDIA Parakeet 等模型，通过 whisper.cpp 等优化运行时来运行。音频从不离开你的机器。","cloudLead":"云端转录会将你的音频发送到远程服务器","body2":"由 Google、AWS、Microsoft、Deepgram 或 AssemblyAI 等服务商运营，在大型 GPU 集群上处理后返回文本转录结果。两种方案之间的关键权衡可归结为五个因素：隐私、延迟、准确性、成本和离线能力。本文将结合真实数据和真实政策逐一剖析，帮你做出明智的决定。","lastUpdated":"最后更新：2026 年 2 月 17 日。文中所有量化数据和政策表述均已与至少两个一手来源交叉核对。"},"factCheck":{"title":"事实核查速览（双重来源）","item1":{"lead":"开源本地模型已具备生产级能力：","body":"Whisper 和 Parakeet 拥有公开的模型卡和开源实现，适用于设备端推理。","source1":"OpenAI Whisper","source2":"NVIDIA Parakeet"},"item2":{"lead":"云端 STT 按用量计费：","body":"主流服务商按音频时长和模型档位计量。","source1":"Google 定价","source2":"AWS 定价"},"item3":{"lead":"各服务商的数据处理方式不同：","body":"日志记录/数据保留设置和退出选项因服务而异，须自行配置。","source1":"Google 数据日志","source2":"AWS AI 退出政策"},"item4":{"lead":"监管背景至关重要：","body":"生物识别和敏感数据相关义务因司法辖区和使用场景而异。","source1":"GDPR 官方文本","source2":"《欧盟 AI 法案》官方文本"},"item5":{"lead":"OpenWhispr 的部署模式：","body":"OpenWhispr 默认采用本地处理，并在用户需要时支持 BYOK 云端路由。","source1":"OpenWhispr","source2":"whisper.cpp 运行时"}},"howCloudWorks":{"title":"云端转录的工作原理","paragraph1":"当你使用云端语音转文字服务时，你的音频会经历一段多步骤的旅程。首先，你的设备从麦克风采集原始音频，并压缩成适合传输的格式——通常是 Opus、FLAC 或线性 PCM。随后，这段压缩后的音频会通过互联网发送到服务商的 API 端点，通常经由 HTTPS 或用于实时流式传输的 WebSocket 连接。","paragraph2":"在服务器端，服务商会让你的音频通过托管在 GPU 集群上的大型神经网络模型进行处理。这些模型通常基于数十万小时的标注语音数据训练而成，远远超出任何个人所能收集的规模。生成的转录结果随后被发送回你的设备。","providersIntro":"主流的云端转录服务商包括：","providers":{"google":"——语言覆盖广泛，提供批量和实时流式两种选项。","aws":"——亚马逊的产品，支持自动语言识别、自定义词汇和 PII 脱敏。","azure":"——与 Microsoft 生态深度集成，支持自定义语音模型和实时转录。","deepgram":"——以速度和开发者体验著称，其 Nova-3 模型在具竞争力的价格下提供出色的准确性。","assemblyai":"——专注于准确性和开发者工具，内置说话人分离和内容审核。"},"paragraph3":"云端延迟是可变的，因为它涵盖了音频采集、上传、服务器端推理和结果返回。在网络良好的环境中，它可能感觉很快；而不稳定的网络或高延迟地区则会让体验明显变慢。批量转录可能耗时数秒到数分钟不等，具体取决于文件长度和排队情况。"},"howLocalWorks":{"title":"本地转录的工作原理","paragraph1":"本地转录直接在你的设备上运行语音识别模型。整个过程不涉及网络——音频从麦克风直接进入运行在 CPU 或 GPU 上的模型，文本从另一端输出。整个过程都在你机器的内存中完成。","paragraph2a":"让高质量本地转录变得切实可行的突破，是 OpenAI 于 2022 年 9 月发布的 Whisper——一个基于 68 万小时多语言音频训练的开源模型。最近，NVIDIA 发布了其 Parakeet 系列 ASR 模型，在英语基准测试中达到了业界领先的准确性，并以开源许可证提供。不久之后，Georgi Gerganov 打造了","paragraph2b":"，一个针对 CPU 推理优化的 C/C++ 重新实现，让 Whisper 的准确性在无需独立 GPU 的情况下也能在日常硬件上发挥。","modelTable":{"intro":"Whisper 提供多种模型规格，各自在准确性与速度之间做出取舍：","headerModel":"模型","headerParameters":"参数量","headerVram":"显存","headerSpeed":"相对速度","rows":[{"model":"Tiny","params":"39M","vram":"~1 GB","speed":"~10x"},{"model":"Base","params":"74M","vram":"~1 GB","speed":"~7x"},{"model":"Small","params":"244M","vram":"~2 GB","speed":"~4x"},{"model":"Medium","params":"769M","vram":"~5 GB","speed":"~2x"},{"model":"Large-v3","params":"1550M","vram":"~10 GB","speed":"1x（基准）"},{"model":"Turbo","params":"809M","vram":"~6 GB","speed":"~8x"}]},"paragraph3":"在 Apple Silicon（M1 至 M4）上，whisper.cpp 通过 Metal 在 GPU 上完整运行推理，即便使用 medium 模型也能达到实时甚至超实时的速度。Turbo 模型——large-v3 的优化变体，准确性损失极小——在现代 MacBook 上能以高吞吐量运行，这也是它常被用于低延迟本地听写工作流的原因。","paragraph4a":"其他本地转录引擎还包括","paragraph4b":"（轻量级，准确性较低）和","paragraph4c":"（支持 ONNX 运行时的新一代 Kaldi）。不过，通过 whisper.cpp 运行的 OpenAI Whisper 和 NVIDIA Parakeet 仍是本地转录质量的黄金标准。"},"privacy":{"title":"隐私：核心权衡","paragraph1":"隐私是本地转录与云端转录之间最大的区别所在。采用本地处理时，\"我的音频数据会怎样？\"这个问题有一个简单的答案：什么都不会发生。它留在你的设备上、内存里，转录完成后即被丢弃。没有需要信任的第三方，没有需要阅读的政策，也无需担心数据泄露。","paragraph2":"云端转录则需要你信任你的服务商。以下是主流服务商关于如何处理你音频的真实说法：","google":{"body1":"Google 的语音转文字数据日志默认是关闭的。在关闭状态下，Google 表示音频数据会在内存中处理，仅用于返回转录结果，不会存储在服务器上。但如果你选择开启数据日志（或使用某些需要它的功能），你的音频和转录内容可能会被存储并用于改进 Google 的模型。Google 的","link":"数据日志文档","body2":"详细说明了这些区别。","footnote":"数据处理位置可按地区指定，这对 GDPR 合规至关重要。"},"aws":{"body1":"默认情况下，AWS 可能会使用由 Amazon Transcribe 处理的内容来开发和改进 AWS 的 AI/ML 服务。不过，你可以通过","link":"AWS AI 服务退出政策","body2":"来选择退出。在退出后，AWS 表示你的内容不会被存储或用于服务改进。音频数据在传输中和静态存储时均经过加密。","footnote":"重要提示：退出并非默认设置。你必须主动进行配置。"},"azure":{"body1":"Azure 默认不会使用客户数据来改进其基础模型。其","link":"数据隐私文档","body2":"指出，发送到语音服务的音频会被处理并立即从其服务器删除。如果你创建自定义模型，你提供的训练数据会存储在与该资源相同的地区，直到你明确删除为止。","footnote":"Azure 提供断网容器部署，可实现完全本地化的处理。"},"regulatory":{"title":"监管考量","gdprLead":"GDPR：","gdprBody":"根据欧盟法律，语音录音被视为生物识别数据。即便采用标准合同条款（SCC）或欧盟-美国数据隐私框架，将音频发送到位于美国的云端服务商仍会引发 GDPR 下的数据跨境传输问题。本地处理通过将生物识别数据保留在数据主体自己的设备上，从根本上规避了这一问题。","hipaaLead":"HIPAA：","hipaaBody":"使用云端转录的医疗机构必须确保其服务商提供商业伙伴协议（BAA）。Google、AWS 和 Azure 均提供 BAA，但配置要求十分严格。本地转录则绕开了 HIPAA 的数据处理规则，因为受保护的健康信息从不脱离涵盖实体的控制范围。"},"localAdvantage":{"title":"本地的优势","body":"采用本地转录，没有需要研读的隐私政策，没有需要信任的数据保留期限，也没有需要寻找的退出开关。你的音频在内存中处理，从不写入磁盘（除非你选择保存录音）。这是一种根本不同的信任模式——你无需信任任何人，因为数据从不离开你的机器。"}},"accuracy":{"title":"准确性：本地能有多接近？","paragraph1":"说句实话：在某些使用场景下，顶级云端服务商在原始准确性上仍占有优势。它们基于海量专有数据集训练，提供实时模型更新，还能利用上下文相关的自定义词汇。对于难度较高的音频——浓重口音、嘈杂环境、领域专业术语——像 Google 的 Chirp 2 或 Deepgram 的 Nova-3 这类云端服务往往表现出色。","paragraph2":"话虽如此，差距已经大幅缩小。基于 68 万小时多语言音频训练的 OpenAI Whisper large-v3，在大多数常见语言上都实现了具竞争力的词错率（WER）。NVIDIA 的 Parakeet 模型更进一步，在标准英语基准测试中取得了一些最低的 WER。独立基准测试一再表明，对于英语，这些开源模型的表现与商用云端服务相差仅几个百分点，在某些长尾语种上甚至超过它们。","paragraph3":"Turbo 模型——large-v3 的蒸馏变体，参数量为 8.09 亿而非 15.5 亿——保留了大部分准确性，同时运行速度快约 8 倍。对于对着像样麦克风清晰口述的实时听写而言，对大多数用户来说，Turbo 与云端 API 之间的差异微乎其微。","cloudAdvantages":{"title":"云端在准确性上的优势","item1":"自定义词汇与领域自适应","item2":"服务器端的持续模型改进","item3":"更出色的说话人分离（谁说了什么）","item4":"针对各语言调优的自动标点与格式化"},"localAdvantages":{"title":"本地在准确性上的优势","item1":"无网络传输带来的音频压缩失真","item2":"稳定、可复现的性能——不受服务器波动影响","item3":"完整无压缩的 16kHz 音频直接喂给模型","item4":"Whisper large-v3 和 Parakeet 在清晰语音听写上可媲美云端"},"bottomLine":"结论：如果你在安静环境中用像样的麦克风口述，本地运行的 Whisper Turbo、Whisper large-v3 或 Parakeet 等模型在大多数实际用途上能给出与云端服务难以区分的结果。而在嘈杂的多说话人场景、小众语言和领域专业词汇方面，云端则更胜一筹。"},"speed":{"title":"速度与延迟","paragraph1":"延迟对实时听写最为关键——你希望话音刚落，文字就立刻出现。在这方面，本地转录与云端转录有着根本不同的延迟特征。","cloudBreakdown":{"title":"云端延迟构成","item1":"音频采集 + 编码","item2Lead":"网络上传与下载","item2Rest":"（因线路质量而异）","item3Lead":"服务器端排队 + 推理","item3Rest":"（因服务商负载而异）","summary":"用户感知到的延迟可能因网络和地区而显著不同。"},"localBreakdown":{"title":"本地延迟（Apple Silicon）","item1Lead":"Tiny/base：","item1Rest":"响应最快，质量上限较低","item2Lead":"Small：","item2Rest":"速度与质量兼顾","item3Lead":"Turbo：","item3Rest":"高质量实时听写的理想之选","item4Lead":"大型模型：","item4Rest":"质量最高，算力开销更大","summary":"无网络依赖。无论连接状况如何都保持稳定。"},"paragraph2":"对于实时流式传输，在连接良好时，云端服务商能快速返回部分识别假设。但这仍需要稳定的低延迟连接。在飞机上、拥挤的公共 Wi-Fi 中，或受限的 VPN 线路之后，感知延迟可能迅速恶化。","paragraph3":"传统意义上，使用 whisper.cpp 的本地转录并非实时流式。大多数本地实现采用\"按住说话\"的模式：你说话，音频被缓冲，然后模型处理完整的语音片段。在搭载 Apple M 系列芯片上使用 Turbo 模型时，这一处理步骤足够快，对于听写来说延迟几乎可以忽略——短句通常在一秒以内完成。","footnote":"对于预先录制文件的长篇转录，云端服务能在 GPU 集群间并行处理，往往能快速完成大批量任务。本地吞吐量受限于你的机器和模型选择，因此处理超长录音时运行时间可能更慢。"},"cost":{"title":"成本：免费 vs. 按分钟付费","paragraph1":"本地转录除了你已拥有的硬件之外，不需要任何额外花费。没有 API 密钥，没有计费面板，没有按分钟收费。你下载一个模型文件（从 Tiny 的 75MB 到 Large-v3 的 3GB 不等），然后就大功告成了。永远如此。","paragraph2":"云端服务商按用量收费（按分钟或按小时，视厂商和模型而定）。套餐和折扣经常变动，因此在做预算前请先核实当前费率：","table":{"headerProvider":"服务商","headerBatch":"批量 / 预录","headerStreaming":"流式 / 实时","headerFreeTier":"免费额度","rows":[{"provider":"Google Speech-to-Text","batch":"按用量计费（按模型分档）","streaming":"按用量计费（按模型分档）","free":"试用/免费配额因账户而异"},{"provider":"AWS Transcribe","batch":"按用量计费","streaming":"按用量计费","free":"入门免费额度（限时）"},{"provider":"Azure Speech","batch":"因地区/模型而异","streaming":"因地区/模型而异","free":"有限免费配额"},{"provider":"Deepgram","batch":"按模型档位计费","streaming":"按模型档位计费","free":"基于额度的试用"},{"provider":"AssemblyAI","batch":"按模型档位计费","streaming":"按模型档位计费","free":"基于额度的试用"},{"provider":"本地（Whisper / Parakeet）","batch":"免费","streaming":"免费","free":"无限量，永久"}]},"estimate":{"title":"如何估算你的云端开销","body1a":"使用这个简单的公式：","formula":"年度音频分钟数 x 厂商费率","body1b":"如果你的团队大量使用听写，总成本会随用量和席位数线性增长。","body2":"本地推理（Whisper/Parakeet）避免了持续的 API 费用，这正是许多团队将本地作为默认方案、只在特殊情况下才路由到云端的原因。"},"footnote":"注意：云端定价通常会对说话人分离、PII 脱敏或自定义词汇等功能收取额外费用。上述基础价格仅适用于标准转录。"},"offline":{"title":"离线能力：本地的绝对胜场","paragraph1":"这一点是非此即彼的：云端转录需要联网，本地转录则不需要。对许多人来说，这是决定性因素。","scenariosIntro":"离线能力不可或缺的场景：","travel":{"title":"出行与远程办公","body":"航班、火车、乡村地区、发展中地区——任何 Wi-Fi 时有时无或根本没有的地方。本地转录在 3.5 万英尺高空的表现，与在你办公桌前完全一样。"},"airGapped":{"title":"物理隔离环境","body":"政府机构、国防承包商、安全研究实验室和金融机构通常运行在物理隔离网络上，对外网络访问被完全屏蔽。本地转录是唯一选择。"},"fieldWork":{"title":"野外作业","body":"冲突地区的记者、偏远野外站点的研究人员、乡村诊所的医护人员——这些专业人士需要在蜂窝信号可能不稳定或不可用的环境中获得可靠的转录。"},"reliability":{"title":"可靠性","body":"即便在网络良好的办公室里，云端服务也会出现宕机。AWS、Google Cloud 和 Azure 都曾发生过波及语音 API 的数小时事故。本地转录没有任何可能宕机的外部依赖。"}},"comparisonTable":{"title":"逐项对比","headerFactor":"因素","headerLocal":"本地","headerCloud":"云端","headerNotes":"说明","rows":[{"factor":"隐私","local":"full","cloud":"none","notes":"本地：数据从不离开设备。云端：音频被发送到第三方服务器。"},{"factor":"准确性（英语）","local":"partial","cloud":"full","notes":"云端模型略胜一筹；Whisper large-v3 和 NVIDIA Parakeet 大幅缩小了差距。"},{"factor":"准确性（多语言）","local":"partial","cloud":"full","notes":"云端服务商凭借定制训练的模型拥有更广的语言覆盖。"},{"factor":"延迟","local":"full","cloud":"partial","notes":"本地避免了网络往返。云端延迟取决于连接质量和地区。"},{"factor":"成本","local":"full","cloud":"none","notes":"本地没有按分钟的 API 计费。云端服务采用按用量定价。"},{"factor":"离线使用","local":"full","cloud":"none","notes":"本地随处可用。云端需要联网。"},{"factor":"配置便利性","local":"partial","cloud":"full","notes":"云端只需一次 API 调用。本地需要下载模型并考虑硬件。"},{"factor":"说话人分离","local":"none","cloud":"full","notes":"多说话人识别是云端的一大优势。"},{"factor":"可扩展性","local":"none","cloud":"full","notes":"云端可无限扩展。本地受限于你的硬件。"}]},"recommendations":{"title":"何时选择哪种方案","intro":"两种方案都没有绝对的优劣。正确的选择取决于你看重什么。","chooseLocal":{"title":"如果满足以下情况，请选择本地……","item1":"隐私不容妥协（医疗、法律、个人日记）","item2":"你需要离线或物理隔离运行","item3":"你希望转录没有任何持续成本","item4":"你的主要用途是听写（单说话人、清晰音频）","item5":"你受 GDPR、HIPAA 或类似法规约束"},"chooseCloud":{"title":"如果满足以下情况，请选择云端……","item1":"你需要为多人会议进行说话人分离","item2":"在困难音频条件下追求最高准确性","item3":"你正在构建一个需要扩展到大量用户的应用","item4":"你需要实时流式转录","item5":"你的硬件有限，无法运行大型模型"},"hybrid":{"title":"混合方案","body":"你不必只能二选一。一些应用——包括 OpenWhispr——同时支持两种模式：默认使用 Whisper 和 Parakeet 等本地模型以兼顾隐私与零成本，并可在需要云端提供的额外准确性或功能时，选择自带云端 API 密钥（OpenAI、Deepgram 等）。这种\"BYOK\"（自带密钥）模式让你兼得两者之长，而不会被任何一种方案所锁定。"}},"sourcesSection":{"title":"参考来源与延伸阅读","whisper":"——原始模型仓库，包含模型规格、基准测试和文档。","whisperCpp":"——Whisper 的 C/C++ 移植版，针对 CPU 和 Apple Silicon GPU 推理优化。","parakeetLabel":"NVIDIA Parakeet RNNT 模型卡","parakeet":"——开源 ASR 模型指标与使用指南。","googlePricingLabel":"Google Cloud Speech-to-Text 定价","googlePricing":"——Google 语音 API 的当前定价档位。","awsPricingLabel":"AWS Transcribe 定价","awsPricing":"——亚马逊批量与流式的当前按用量定价。","azurePricingLabel":"Azure Speech Services 定价","azurePricing":"——Microsoft 实时与批量语音转文字的定价。","deepgramPricingLabel":"Deepgram 定价","deepgramPricing":"——Deepgram 各模型档位的定价详情。","assemblyaiPricingLabel":"AssemblyAI 定价","assemblyaiPricing":"——AssemblyAI 用量与模型档位的定价详情。","googleLoggingLabel":"Google Speech-to-Text 数据日志","googleLogging":"——Google 关于音频数据如何处理和存储的文档。","awsOptOutLabel":"AWS AI 服务退出政策","awsOptOut":"——如何阻止 AWS 使用你的内容改进其 AI 服务。","azurePrivacyLabel":"Azure Speech 数据隐私","azurePrivacy":"——Microsoft 关于语音服务的数据隐私与安全文档。","whisperPaperLabel":"通过大规模弱监督实现稳健语音识别","whisperPaper":"——Radford 等人（2022）的原始 Whisper 论文，详述了训练方法与 WER 基准。","gdprLabel":"GDPR（欧盟官方文本）","gdpr":"——个人/生物识别数据处理的法律基准。","aiActLabel":"《欧盟 AI 法案》（欧盟官方文本）","aiAct":"——与生物识别及高风险系统相关的 AI 风险分级义务。","openwhispr":"——在单一应用中实现本地优先 + 可选 BYOK 云端工作流的范例。"},"cta":{"title":"在一个应用中体验两种方案","body":"OpenWhispr 同时支持本地模型（OpenAI Whisper 和 NVIDIA Parakeet）以兼顾隐私与零成本，并在你需要最高准确性时支持自带云端 API 密钥。开源，永久免费。","starButton":"在 GitHub 上为我们点星","footnote":"无需账户 · 可离线使用 · 永久开源"},"internalLinks":{"localAi":"本地 AI 的理由","offline":"离线转录","allBlog":"全部博客文章","allComparisons":"全部对比"}},"blogModelSizes":{"metadata":{"title":"Whisper 模型大小：完整指南 | OpenWhispr","description":"对比全部 Whisper 模型大小的参数、速度、准确率和内存需求，选择适合本地转录的 tiny、small、large-v3 或 turbo。","keywords":["Whisper 模型尺寸","Whisper tiny 与 large 对比","Whisper 模型对比","该用哪个 Whisper 模型","Whisper large v3","Whisper 模型准确率","Whisper 模型速度","Whisper 模型参数","whisper.cpp 模型尺寸"]},"schema":{"title":"Whisper 模型尺寸详解：Tiny、Base、Small、Medium 与 Large 对比","description":"一份实用指南，帮你选择合适的 Whisper 模型。对比参数量、速度、准确率和硬件要求。"},"breadcrumb":{"home":"首页","blog":"博客","article":"Whisper 模型尺寸详解"},"layout":{"title":"Whisper 模型尺寸详解：Tiny、Base、Small、Medium 与 Large 对比","subtitle":"全面对比每个 Whisper 模型。参数量、速度、准确率，以及你究竟该用哪一个。","date":"2026 年 2 月 5 日","tag":"博客"},"sectionsToc":{"factCheck":"事实核查速览","comparisonTable":"模型对比","tiny":"Tiny（39M）","base":"Base（74M）","small":"Small（244M）","medium":"Medium（769M）","large":"Large（1.55B）","turbo":"Turbo（809M）","englishVsMultilingual":"英语专用版与多语言版","hardware":"硬件要求","recommendations":"选型建议","openwhispr":"OpenWhispr","sources":"参考来源"},"cards":{"resources":"资源占用","speed":"速度","accuracy":"准确率","bestFor":"最适合："},"intro":{"lead":"OpenAI 的 Whisper 提供 9 个模型版本，分为 5 个尺寸类别：","body1":"tiny（3900 万参数）、base（7400 万）、small（2.44 亿）、medium（7.69 亿）和 large（15.5 亿）。每个尺寸都有多语言版本和英语专用版本（large 和 turbo 除外，二者仅有多语言版）。此外还有一个 turbo 模型（8.09 亿）——它是 large-v3 的蒸馏版本，准确率几乎不相上下，但速度大幅提升。模型越大越准确，但速度越慢，占用的内存也越多。","body2":"大多数本地转录应用使用的是 whisper.cpp 量化（GGML）格式，其模型文件大小从 75 MB（tiny）到 2.9 GB（large）不等。运行时内存占用从 tiny 的约 273 MB 到 large 的 3.9 GB 不等。合适的模型取决于你的硬件、语言以及你对准确率的要求。","lastUpdated":"最近更新：2026 年 2 月 17 日。参数、内存和基准测试数据均已与至少两个权威来源交叉核对。"},"factCheck":{"title":"事实核查速览（双重来源）","item1":{"lead":"官方模型系列与参数量：","body":"tiny/base/small/medium/large 以及 turbo 均由 OpenAI 官方记录。","link1":"OpenAI Whisper README","link2":"Whisper 模型卡"},"item2":{"lead":"GGML/GGUF 磁盘与运行时内存背景：","body":"本地部署占用数据来自 whisper.cpp 的转换/运行时文档。","link1":"whisper.cpp","link2":"whisper.cpp 内存说明"},"item3":{"lead":"基准测试注意事项：","body":"WER 因数据集和语言不同而差异巨大；LibriSpeech 的数据并非通用的质量评分。","link1":"Whisper 论文","link2":"large-v2 模型卡"},"item4":{"lead":"Turbo 模型定位：","body":"turbo 是一个侧重速度的版本，其准确率上的取舍在发布说明/模型卡中均有记录。","link1":"turbo 模型卡","link2":"turbo 发布讨论"},"item5":{"lead":"OpenWhispr 使用场景：","body":"OpenWhispr 提供多个 Whisper 尺寸，让用户能在设备本地调节速度与准确率。","link1":"OpenWhispr","link2":"whisper.cpp 后端"}},"table":{"title":"完整模型对比","intro":"本表涵盖了每一个官方 Whisper 模型。「相对速度」以 large 为基准（1 倍）。VRAM 数据针对 GPU 上的 PyTorch 推理。GGML 列显示的是 whisper.cpp 的文件大小和内存占用，而大多数桌面应用使用的正是 whisper.cpp。","headers":{"model":"模型","params":"参数量","englishOnly":"英语专用","vram":"VRAM（GPU）","ggmlDisk":"GGML 磁盘","ram":"内存（whisper.cpp）","speed":"速度","englishWer":"英语 WER","multilingualWer":"多语言 WER"},"notAvailable":"不适用","notesLead":"说明：","notesBody":"WER（词错误率）数据是来自 Whisper 论文、HuggingFace 模型卡（英语用 LibriSpeech 测试集）以及 OpenAI 公开基准测试的近似平均值。确切的 WER 会因数据集、语言、音频质量和评测方法的不同而显著变化。数值越低越好。速度以 large 为基准（1 倍 = 最慢）。"},"tiny":{"title":"Tiny（3900 万参数）","para1":"tiny 模型是 Whisper 中最小、最快的版本。仅有 3900 万参数，GGML 文件大小为 75 MB，几乎可以在任何硬件上轻松运行——包括 Raspberry Pi、老旧笔记本电脑和低端移动设备。相对于 large 模型，它的转录速度约为实时速度的 10 倍，因此在延迟比完美准确率更重要时，它是最佳选择。","resources":{"l1":"GGML：磁盘占用 75 MiB","l2":"内存：运行时约 273 MB","l3":"VRAM：约 1 GB（GPU 模式）"},"speed":{"l1":"比 large 快约 10 倍","l2":"在 Apple Silicon 上近乎瞬时","l3":"在大多数 CPU 上可实时运行"},"accuracy":{"l1":"英语 WER：约 7.6%","l2":"多语言 WER：约 12%","l3":"难以应对口音/噪声"},"para2":"在 tiny.en 版本（英语专用）上，LibriSpeech test-clean 的 WER 约为 5.6%，在 test-other（嘈杂/带口音音频）上升至约 14.9%。多语言版 tiny 模型在英语上表现稍逊，但支持全部 99 种 Whisper 语言。","para3":"tiny 非常适合快速记笔记、低风险口述、实时字幕原型，以及算力有限的嵌入式或边缘设备。它是许多基于 whisper.cpp 工具的默认起步模型，因为它几秒钟即可下载完成并即刻运行。","bestFor":"追求极致转录速度、低资源硬件、Raspberry Pi、实时应用"},"base":{"title":"Base（7400 万参数）","para1":"base 模型的参数量是 tiny 的两倍（7400 万对 3900 万），但仍然非常轻量。GGML 文件 142 MiB，运行时内存约 388 MB，依然能在几乎任何现代设备上轻松运行。相比 tiny，准确率的提升十分明显——尤其是在嘈杂音频和带口音的语音上。","resources":{"l1":"GGML：磁盘占用 142 MiB","l2":"内存：运行时约 388 MB","l3":"VRAM：约 1 GB（GPU 模式）"},"speed":{"l1":"比 large 快约 7 倍","l2":"在 CPU 上依然很快","l3":"在 Apple Silicon 上低于一秒"},"accuracy":{"l1":"英语 WER：约 5.0%","l2":"多语言 WER：约 10%","l3":"在嘈杂音频上优于 tiny"},"para2":"base.en 版本在 LibriSpeech test-clean 上的 WER 约为 4.3%，在 test-other 上约为 12.8%。相比 tiny.en（分别为 5.6% 和 14.9%），这是一次有意义的提升，尤其是在高难度音频上。","para3":"对于 tiny 准确率尚不够用的低功耗设备，base 是个不错的选择。它在实时转录流水线中也很受欢迎——你想要速度与质量的平衡，又不希望内存占用超过 500 MB 时，它正合适。","bestFor":"需要比 tiny 更高准确率的低功耗设备、实时流式转录、入门级硬件"},"small":{"title":"Small（2.44 亿参数）","para1":"从 small 开始，Whisper 才真正让人感觉准确。凭借 2.44 亿参数，它相比 base 实现了 3.3 倍的飞跃，准确率显著提升——尤其是在非英语语言和嘈杂录音上。GGML 文件为 466 MiB，运行时内存占用约 852 MB。它的运行速度约为 large 的 4 倍。","resources":{"l1":"GGML：磁盘占用 466 MiB","l2":"内存：运行时约 852 MB","l3":"VRAM：约 2 GB（GPU 模式）"},"speed":{"l1":"比 large 快约 4 倍","l2":"在现代笔记本上很快","l3":"在 Apple Silicon 上可实时运行"},"accuracy":{"l1":"英语 WER：约 3.4%","l2":"多语言 WER：约 7%","l3":"能很好地处理口音"},"para2":"small.en 版本在 LibriSpeech test-clean 上的 WER 约为 3.0%——对于一个不到 500 MB 的模型来说，这是相当出色的成绩。在许多仅需英语的场景中，small.en 能以一小部分资源成本提供接近 medium 的准确率。","para3":"small 常被推荐为大多数用户的默认模型。它在任何现代笔记本上都运行良好（包括配备 8 GB 内存的 MacBook Air M1），在各种语言上都有不错的准确率，且转录速度足够支撑实时口述工作流。","recommendationLead":"我们的建议：","recommendationBody":"如果你不确定从哪个模型入手，就从 small 开始。对于绝大多数硬件和使用场景，它在速度、准确率和资源占用之间取得了最佳平衡。","bestFor":"使用现代笔记本的大多数用户、默认推荐模型、良好的多语言支持"},"medium":{"title":"Medium（7.69 亿参数）","para1":"从 medium 开始，边际收益逐渐递减——但相比 small，准确率的提升仍然有意义，尤其是在多语言转录、专业词汇和高难度音频条件下。凭借 7.69 亿参数，它需要 1.5 GiB 磁盘空间（GGML），运行时内存约 2.1 GB。","resources":{"l1":"GGML：磁盘占用 1.5 GiB","l2":"内存：运行时约 2.1 GB","l3":"VRAM：约 5 GB（GPU 模式）"},"speed":{"l1":"比 large 快约 2 倍","l2":"在 8 GB 及以上机器上从容运行","l3":"需要不错的硬件"},"accuracy":{"l1":"英语 WER：约 2.9%","l2":"多语言 WER：约 5%","l3":"在专业内容上表现强劲"},"para2":"medium.en 模型在 LibriSpeech test-clean 上的 WER 约为 3.0%，在 test-other 上约为 7.5%。然而在更广泛的基准测试中——尤其是涉及多样口音、背景噪声和特定领域词汇的测试——medium 始终优于 small。","para3":"对于看重准确率、但又没有硬件（或耐心）去跑 large 的专业转录工作流来说，medium 是个有力的选择。它在配备 8 GB 及以上内存的机器上运行良好，并能从 GPU 加速中获得显著收益。","bestFor":"专业用途、多语言转录、专业内容、配备 8 GB 及以上内存的机器"},"large":{"title":"Large（15.5 亿参数）","para1":{"part1":"large 模型是最准确的 Whisper 版本，拥有 15.5 亿参数。它有三个版本：","part2":"（最初的发布版），","part3":"（训练轮数增加 2.5 倍并加入正则化），以及","part4":"（在更多数据上训练，并采用 128 个梅尔频率分箱而非 80 个）。large-v3 是目前追求最高准确率的推荐版本。"},"resources":{"l1":"GGML：磁盘占用 2.9 GiB","l2":"内存：运行时约 3.9 GB","l3":"VRAM：约 10 GB（GPU 模式）"},"speed":{"l1":"1 倍（基准——最慢）","l2":"强烈推荐使用 GPU","l3":"CPU：可能慢于实时速度"},"accuracy":{"l1":"英语 WER：约 2.4%（v3）","l2":"多语言 WER：约 3.5%（v3）","l3":"在所有语言上都最佳"},"versionsTitle":"large-v2 与 large-v3 对比","v2":{"l1":"训练轮数比 large-v1 多 2.5 倍","l2":"加入正则化以获得更好的泛化能力","l3":"LibriSpeech test-clean WER：约 3.0%","l4":"在某些音频类型上更稳定"},"v3":{"l1":"128 个梅尔分箱（对比 80 个）——更精细的频率分辨率","l2":"在 100 万小时标注音频 + 400 万小时伪标注音频上训练","l3":"跨语言相比 large-v2 降低 10-20% 的错误率","l4":"新增粤语支持"},"whichLead":"该用哪个 large 模型：","whichBody":"对于新项目，请使用 large-v3。在各语言的平均表现上，它都严格优于 large-v2。不过，一些用户反映，在音频片段非常安静或静音的特定边缘情况下，large-v2 产生的幻觉（生成音频中并不存在的文本）更少。如果你在使用 large-v3 时遇到幻觉问题，可以尝试退回 large-v2 作为备选。","bestFor":"追求最高准确率、专业/医疗/法律转录、低资源语言、批量处理"},"turbo":{"title":"Turbo（8.09 亿参数）","para1":"turbo 模型是 large-v3 的蒸馏版本，它在保留大部分准确率的同时大幅缩短了推理时间。OpenAI 通过将解码器从 32 层精简至仅 4 层实现了这一点——参数量从 15.5 亿减少到 8.09 亿。而负责繁重工作的编码器与 large-v3 保持完全一致。","resources":{"l1":"GGML：磁盘占用约 1.6 GiB","l2":"内存：运行时约 2.3 GB","l3":"VRAM：约 6 GB（GPU 模式）"},"speed":{"l1":"比 large 快约 8 倍","l2":"准确率接近 large，速度接近 tiny","l3":"配合 torch.compile 在 GPU 上表现出色"},"accuracy":{"l1":"英语 WER：约 2.5%","l2":"多语言 WER：约 3.7%","l3":"相比 large-v3 质量略有损失"},"para2":"当你想要接近 large-v3 的准确率、却无法承受完整 large 模型的延迟时，turbo 是最佳选择。在 GPU 上配合 torch.compile 等优化，turbo 相比标准推理可实现最高 4.5 倍的速度提升，因而极为快速。","limitationLead":"重要限制：","limitationBody":"turbo 模型未针对翻译任务进行训练。如果你需要把一种语言的语音翻译成英语，请改用 medium 或 large-v3。","bestFor":"高准确率实时转录、对延迟敏感的应用、large-v3 太慢的场景"},"englishVsMultilingual":{"title":"英语专用版与多语言版模型","para1":{"part1":"对于 tiny、base、small 和 medium 尺寸，Whisper 提供两个版本：多语言模型和英语专用（","part2":"）模型。large 和 turbo 模型仅有多语言版——没有英语专用版本。"},"enTitle":"何时使用 .en（英语专用）","en":{"l1":"你只转录英语音频","l2":"在 tiny/base 尺寸级别上英语准确率更高","l3":"稍快一些——无需语言检测开销","l4":"文件大小与多语言版相同"},"multiTitle":"何时使用多语言版","multi":{"l1":"你需要转录非英语音频","l2":"你的音频包含多种混合语言","l3":"你需要语音翻译（译为英语）","l4":"使用 large 或 turbo 时必须用它（不存在 .en 版本）"},"para2":".en 版与多语言版模型之间的性能差距在较小尺寸上最为明显。对于 tiny 和 base，.en 版本在英语上明显更好。等到了 small 和 medium，差距就大幅缩小了。在 large 级别只有多语言模型——而它在英语上的表现同样出色。","ruleLead":"经验法则：","ruleBody":"如果你只用英语，并且在 tiny 或 base 中选择，那就选 .en 版本。对于 small 及以上，多语言版用于英语效果很好，还能让你灵活应对其他语言。"},"hardware":{"title":"硬件要求：哪些模型能在哪里运行","intro":"你需要的硬件在很大程度上取决于模型尺寸，以及你使用的是 whisper.cpp（CPU/Metal/CUDA）还是 PyTorch 实现（侧重 GPU）。以下是常见硬件类别上的预期表现。","m1Air":{"title":"MacBook Air M1（8 GB 内存）","r1Label":"tiny/base：","r1":"瞬时完成。快于实时速度。","r2Label":"small：","r2":"很快。借助 Metal 可达实时或更优。","r3Label":"medium：","r3":"可用。在 CPU 上可能略慢于实时速度。Metal 能带来显著帮助。","r4Label":"large：","r4":"可行但内存吃紧。预计会慢于实时速度。","r5Label":"turbo：","r5":"很合适。借助 Metal 加速可接近实时速度。"},"mbpPro":{"title":"MacBook Pro M2/M3（16-36 GB 内存）","r1Label":"tiny/base/small：","r1":"瞬时完成。全部远在硬件上限之内。","r2Label":"medium：","r2":"很快。借助 Metal 从容运行。","r3Label":"large：","r3":"良好。借助 Metal 可达实时或接近实时。","r4Label":"turbo：","r4":"出色。快速实时转录。"},"midRange":{"title":"中端 Windows/Linux 笔记本（8 GB 内存，集成显卡）","r1Label":"tiny/base：","r1":"很快。仅用 CPU 即可。","r2Label":"small：","r2":"良好。在大多数现代 CPU 上从容运行。","r3Label":"medium：","r3":"可行。在 CPU 上可能比实时速度慢 2-3 倍。","r4Label":"large：","r4":"有挑战。仅用 CPU 会很慢。","r5Label":"turbo：","r5":"可行。若有 Vulkan 可从中受益。"},"nvidia":{"title":"配备 NVIDIA 显卡的台式机（RTX 3060 及以上，8 GB 及以上 VRAM）","r1Label":"所有模型：","r1":"很快。GPU 能从容应对一切。","r2Label":"large：","r2":"需要 10 GB VRAM（RTX 3060 12GB 或更高）。","r3Label":"turbo：","r3":"配合 CUDA 是最快的选择。明显快于 large。","r4":"在 whisper.cpp 中使用 CUDA 后端可获得最佳性能。"},"backendsTitle":"whisper.cpp 加速后端","backendsIntro":"whisper.cpp——大多数桌面应用使用的 C/C++ 移植版本——支持多种硬件加速后端，可大幅提升性能：","metal":{"title":"Metal（Apple Silicon）","body":"在 M1/M2/M3/M4 Mac 上进行完整的 GPU 推理。整个模型在 GPU 上运行，零内存拷贝。这是 macOS 用户最快的后端，也是 OpenWhispr 在 Mac 上所采用的方案。"},"coreml":{"title":"Core ML（Apple）","body":"使用 Apple 的神经网络引擎进行推理。在较新芯片上，某些模型尺寸的速度可能快于 Metal，且更省电。需要先将模型转换为 Core ML 格式。"},"cuda":{"title":"CUDA（NVIDIA）","body":"为 NVIDIA 显卡提供 GPU 加速。对于拥有独立 NVIDIA 显卡的用户，这是最快的后端。需要 CUDA 工具包。是台式机上运行 large 和 turbo 模型的最佳选择。"},"vulkan":{"title":"Vulkan（跨厂商 GPU）","body":"可与 Intel、AMD 和 NVIDIA 显卡配合使用。对于非 NVIDIA 系统是不错的备选方案。在 Linux 和 Windows 上均可用。性能因 GPU 厂商和驱动而异。"},"speedTitle":"近似转录速度","speedIntro":"针对一段 60 秒音频的实时系数（RTF）。RTF < 1.0 表示快于实时速度。这些只是粗略估计——实际性能取决于音频内容、模型量化和系统负载。","speedTable":{"model":"模型","m1":"M1 MacBook Air","m3":"M3 Pro MacBook","intel":"Intel i7（CPU）","rtx":"RTX 3060（CUDA）"},"readingLead":"如何读懂此表：","readingBody":"0.1x 表示转录 60 秒音频约需 6 秒。1.0x = 实时速度。高于 1.0x 的数值表示慢于实时速度。所有数据均使用默认设置的 whisper.cpp，并在适用时采用 Metal（Mac）或 CUDA（NVIDIA）后端。"},"recommendations":{"title":"你该使用哪个模型？","intro":"以下是基于常见场景的具体建议。拿不定主意时，就从 small 开始，再根据使用体验向上或向下调整。","fastest":{"q":"「我想要尽可能快」","part1":"使用","part2":"或","part3":"。在现代硬件上以毫秒级完成转录。准确率较粗糙，但用于快速记笔记和低风险口述足够了。"},"laptop":{"q":"「我想在笔记本上获得不错的准确率」","part1":"使用","part2":"。对于现代笔记本，它在速度和准确率之间取得了最佳平衡。如果你的机器有 16 GB 及以上内存，","part3":"也是个绝佳选择。","tag":"small 或 medium"},"accuracy":{"q":"「我想要最高的准确率」","part1":"使用","part2":"。它是整体上最准确的 Whisper 模型。如果速度也重要，","part3":"能以 8 倍速度提供 95% 的准确率。","tag":"large-v3 或 turbo"},"englishOnly":{"q":"「我只用英语」","part1":"使用你所选尺寸的","part2":"版本，以获得最佳英语准确率。对于 tiny 和 base，.en 模型明显更好。对于 small 及以上，差异很小。","tag":"small.en 或 medium.en"},"multiLang":{"q":"「我使用多种语言」","part1":"使用多语言版本。","part2":"或","part3":"可获得均衡的表现。","part4":"可获得最佳的多语言准确率。","tag":"small、medium 或 large-v3"},"limited":{"q":"「我的硬件非常有限」","part1":"使用","part2":"或","part3":"。两者都能在 1 GB 内存下运行，甚至可在 Raspberry Pi 和其他单板计算机上工作。base 相比 tiny 准确率有明显提升，而额外成本极小。","tag":"tiny 或 base"}},"openwhispr":{"title":"OpenWhispr 如何处理模型选择","intro":"是一款开源的桌面口述应用，底层使用 whisper.cpp。它让你直接在设置中下载并切换任意 Whisper 模型——无需命令行。模型只需下载一次，即可本地存储。你可以随时更换模型，为你的硬件平衡速度与准确率。","c1":{"title":"下载任意模型","body":"在设置中从 tiny 到 large-v3 任选。OpenWhispr 会自动下载 GGML 版本。"},"c2":{"title":"即时切换","body":"随时更换模型。无需重启。试用不同尺寸，找到最适合你硬件的那一个。"},"c3":{"title":"针对硬件优化","body":"在 Apple Silicon 上使用 Metal，在 NVIDIA 上使用 CUDA，在其他平台上使用优化过的 CPU 推理。全程本地，全程私密。"},"suggestionLead":"我们的建议：","suggestionPart1":"从","suggestionPart2":"开始。如果觉得慢，就降到 base 或 tiny。如果你想要更高的准确率且硬件能够胜任，就升到 medium 或 large-v3。OpenWhispr 让尝试变得轻松。"},"sources":{"title":"参考来源","s1":{"link":"Radford 等人《Robust Speech Recognition via Large-Scale Weak Supervision》（2022）","desc":"——附有完整基准测试结果的 Whisper 原始论文。"},"s2":{"link":"OpenAI Whisper GitHub 仓库","desc":"——官方模型表、参数量、VRAM 要求和相对速度数据。"},"s3":{"link":"OpenAI Whisper 模型卡","desc":"——官方训练数据构成与评测注意事项。"},"s4":{"link":"whisper.cpp GitHub 仓库","desc":"——GGML 模型尺寸、内存占用和硬件加速详情。"},"s5":{"link":"Whisper large-v2 模型卡（HuggingFace）","desc":"——众多 WER 对比所采用的 LibriSpeech 基准背景。"},"s6":{"link":"Whisper large-v3 模型卡（HuggingFace）","desc":"——训练细节、相对 large-v2 的架构变化和多语言改进。"},"s7":{"link":"Whisper large-v3-turbo 模型卡（HuggingFace）","desc":"——Turbo 架构细节、解码器精简的具体做法和性能取舍。"},"s8":{"link":"OpenAI Whisper Turbo 发布公告（GitHub Discussion #2363）","desc":"——官方 turbo 发布说明和跨语言性能对比。"},"s9":{"link":"OpenWhispr","desc":"——选择模型尺寸的实用本地部署背景。"}},"cta":{"title":"在 OpenWhispr 中试用不同的 Whisper 模型","body":"开源、本地优先的口述工具。下载任意 Whisper 模型，一键切换尺寸，为你的硬件找到最完美的平衡。","starButton":"在 GitHub 上给我们加星","footnote":"无需账户 · 离线可用 · 永久开源"},"internalLinks":{"howWhisperWorks":"Whisper AI 的工作原理","linux":"Linux 上的语音口述","allPosts":"全部博客文章","developers":"面向开发者的 OpenWhispr","localVsCloud":"本地转录与云端转录对比"}},"blogOpenSource":{"metadata":{"title":"为什么开源对语音工具至关重要 | OpenWhispr","description":"语音工具处理着最为私密的数据之一——你真实的声音。这就是为什么开源是唯一能赢得真正信任的方式。","keywords":["开源语音识别","开源听写","开源语音转文字","为什么开源很重要","开源语音隐私","开源 AI 工具","透明的语音软件","开源 whisper","nvidia parakeet"]},"breadcrumb":{"home":"首页","blog":"博客","article":"为什么开源很重要"},"schema":{"title":"为什么开源对语音工具至关重要","description":"语音工具处理着最为私密的数据。这就是为什么开源是唯一能赢得真正信任的方式。"},"layout":{"title":"为什么开源对语音工具至关重要","subtitle":"你的声音是生物特征数据。处理它的软件理应可供审计。","date":"2026年2月10日","tag":"博客"},"sections":[{"id":"fact-check-snapshot","title":"事实核查速览"},{"id":"voice-data-is-different","title":"语音数据与众不同"},{"id":"the-trust-problem","title":"信任难题"},{"id":"what-open-source-gives-you","title":"开源带给你的东西"},{"id":"the-open-model-effect","title":"开放模型效应"},{"id":"open-source-polish","title":"开源同样精致"},{"id":"business-model","title":"商业模式"},{"id":"what-to-look-for","title":"该关注什么"},{"id":"sources","title":"参考来源"}],"opening":{"lead":"开源语音工具让你能够准确核实你的音频究竟经历了什么。","body":"使用专有语音软件时，你是在信任一家公司的隐私政策。而使用开源软件时，你可以阅读代码、自行构建，并确认你的语音数据从未离开过你的设备。这一区别对语音工具的意义，远超几乎任何其他类别的软件，因为语音数据独一无二地私密——它是生物特征，难以匿名化，而且一旦被录制，就无法当作没听过。","lastUpdated":"最后更新：2026年2月17日。涉及事件和政策的论断均已对照至少两个一手来源进行交叉核实。"},"factCheck":{"title":"事实核查速览（双来源）","voiceData":{"lead":"语音数据可识别身份且风险极高：","body":"法律与政策框架日益将生物特征处理视为敏感事项。","source1":"GDPR 官方文本","source2":"欧盟《AI 法案》官方文本"},"assistants":{"lead":"主流语音助手都曾陷入语音数据争议：","body":"Amazon、Google 和 Apple 的相关事件已有监管机构和/或主流新闻媒体记录在案。","source1":"FTC 诉 Amazon 案","source2":"路透社关于 Siri 和解的报道"},"cloning":{"lead":"语音克隆滥用是现实存在的消费者风险：","body":"美国监管机构与网络安全指南均提示了冒充风险。","source1":"FTC 警示","source2":"IBM 风险背景资料"},"openModel":{"lead":"开放模型生态切实存在：","body":"Whisper 及其下游运行时让可审计的本地部署成为可能。","source1":"OpenAI Whisper","source2":"whisper.cpp"},"positioning":{"lead":"OpenWhispr 的定位：","body":"OpenWhispr 在开放 ASR 之上构建，是一套开放、本地优先的语音工作流。","source1":"OpenWhispr","source2":"Whisper 上游"}},"voiceData":{"title":"语音数据与众不同","paragraph1":"文本就是文本。你可以剥离元数据、为人名匿名化，并将其汇总为统计数据。而语音不具备这些特性。","paragraph2":"你的声音是一种生物特征标识。就像指纹一样，它独属于你——银行和政府机构使用的声纹认证系统正是依赖这一事实。但同样的独特性，也使语音录音本身就具有身份识别性。一份泄露的文本转录数据库是一起隐私事件，而一份泄露的语音录音数据库则是一起生物特征数据泄露。","reveal":{"title":"语音录音可能暴露的信息","items":["身份——声纹生物特征可唯一识别说话者","健康状况——声学生物标记可提示帕金森病、抑郁症、呼吸系统疾病","情绪状态——压力、愤怒、疲惫都可从声音模式中被检测出来","人口统计特征——年龄、性别、口音、母语、地域来源"]},"anonymize":{"title":"为何语音更难匿名化","items":["你可以从转录文本中抹去文字，但你无法在不破坏录音的前提下从录音中抹去声音","语音克隆技术意味着泄露的音频可被用来生成你声音的深度伪造","欧盟《AI 法案》将生物特征识别系统——包括语音——归类为高风险 AI，须遵守严格的合规要求"]},"paragraph3a":"语音克隆已从研究奇观变成一项商品化服务。只需几秒钟的音频，现代语音合成模型就能生成以假乱真的人声复刻。这并非假想的风险——","ftcLink":"FTC 已就基于 AI 的语音克隆诈骗发出警告","paragraph3b":"自 2023 年起便是如此。当处理你声音的软件是一个黑箱时，你根本无从知晓录音是否被存储、传输，或被用于模型训练。"},"trustProblem":{"title":"专有语音工具的信任难题","intro":"支持开源的理由并非纸上谈兵。每一款主流语音助手，都曾被发现以用户意料之外的方式处理音频。","alexa":{"title":"Amazon Alexa","body1":"2019年4月，Bloomberg","reportedLink":"报道","body2":"称 Amazon 雇用了全球数千名员工，收听在客户家中和办公室采集到的 Alexa 语音录音。这些员工对音频片段进行转录、标注和审核，以改进语音识别系统。一些审核人员表示曾听到令他们不安的录音，包括疑似性侵的内容。Amazon 在回应中承认了这一做法，但强调它仅适用于“极小的样本”。该公司后来增加了退出选项，但这些录音早已根据大多数用户从未仔细阅读的隐私政策被采集。2023年，Amazon","settlementLink":"支付了 2500 万美元","body3":"以了结 FTC 的指控——该指控称其无限期保留儿童的 Alexa 语音录音，违反了儿童隐私规定。"},"google":{"title":"Google Assistant","body1":"2019年7月，比利时广播机构 VRT NWS","obtainedLink":"获取了超过 1000 段 Google Assistant 录音","body2":"录音来自一名承包商。这些录音包含卧室对话、商务通话以及与儿童的互动——其中许多是在无人说出“OK Google”时因误触发而被采集的。Google 确认人工审核人员收听了约 0.2% 的全部语音录音，并在泄露事件后于欧洲暂停了这一做法。汉堡数据保护机构随后展开的调查，促成了在整个欧盟范围内临时禁止该做法。"},"siri":{"title":"Apple Siri","body1":"2019年7月，一名举报人向 The Guardian 透露，Apple 的承包商","heardLink":"经常听到机密的医疗信息、毒品交易和性接触内容","body2":"——这些都发生在他们为 Siri 录音评级的过程中。Apple 此前从未在其隐私文档中披露过这一人工审核项目。该公司致歉，在全球范围内暂停了该项目，并将其改为需用户主动选择加入——但那些已被审核的录音，是在未经知情同意的情况下采集的。Apple 后来就 Siri 隐私侵权问题以","lawsuitLink":"9500 万美元","body3":"和解了一项集体诉讼，时间为 2025 年 1 月。"},"pattern":"这一模式始终如一：一家公司口口声声“我们高度重视你的隐私”，而实际的数据处理做法却只能通过泄露、举报或监管行动才被揭露。这些可不是偷工减料的小公司。它们是 Apple、Google 和 Amazon——这个星球上技术最尖端、资源最雄厚的三家机构。","quote":"“我们不会存储你的数据”是一个需要信任的说法。而开源是一个只需阅读理解能力的说法。"},"whatOpenSource":{"title":"开源真正带给你的东西","intro":"开源不是一个营销标签。它是一组具体的属性，会改变你与所用软件之间的信任模型。","properties":[{"title":"可审计性","body":"任何人都可以阅读源代码，准确核实音频数据是如何被处理的。这款应用会把录音发送到服务器吗？它会把音频存到磁盘上吗？它会回传遥测数据吗？你不必去信任一份隐私政策——你可以自己查。"},{"title":"可复现性","body":"你可以从源代码构建应用，并将其与发布的二进制文件进行比对。这弥合了“我们公开了代码”与“你下载的应用确实运行的就是这份代码”之间的鸿沟。可复现构建是软件信任的黄金标准。"},{"title":"社区审查","body":"安全研究人员、隐私倡导者和经验丰富的开发者都可以独立审查代码。漏洞被公开地发现并修复。这并非纸面上的好处——正是这一点让 Linux、OpenSSL 以及整个互联网基础设施得以建立在开源之上。"},{"title":"分叉权利","body":"如果维护者做出了你不认同的决定——比如增加遥测、移除功能、卖给收购方——社区可以分叉该项目并独立继续开发。这不仅是一份保险，更是一种结构性的激励，促使维护者以用户的利益为重。"},{"title":"无锁定","body":"你的工作流不会被一家公司的商业决策所绑架。如果一款专有听写工具被收购、转型或关停，你为学习和配置它而投入的心血便会付诸东流。而开源工具只要还有人愿意运行它，就会一直存在。"},{"title":"长久存续","body":"公司会消失，开源项目却会长存。Apache HTTP Server 自 1995 年起运行至今。PostgreSQL 自 1996 年。GCC 自 1987 年。Dragon NaturallySpeaking 曾数十年稳居听写产品的领先地位——然后 Nuance 被 Microsoft 收购，这款独立产品也基本停止了开发。开源不存在这种崩溃模式。"}]},"openModel":{"title":"开放模型效应：从 Whisper 到 Parakeet","paragraph1a":"2022年9月，OpenAI","releasedLink":"发布了 Whisper","paragraph1b":"——一个在 68 万小时多语言音频上训练的通用语音识别模型——并以 MIT 许可证将其开源。这是语音工具领域的一个分水岭时刻。","paragraph2":"在 Whisper 之前，高质量的自动语音识别（ASR）既昂贵又封闭。Google Cloud Speech-to-Text、Amazon Transcribe 和 Microsoft Azure Speech Services 都按音频分钟数收费，并要求将录音发送到它们的服务器。最好的离线方案——CMU Sphinx、Kaldi、VOSK——虽能用，但准确率明显逊于云端 API。","paragraph3":"OpenAI Whisper 一夜之间改变了这一格局。第一次，一个准确率比肩甚至超越商业 API 的模型，可供任何人免费下载并在本地运行，且没有任何数据离开他们的机器。NVIDIA 也走了类似的路线，推出了其 Parakeet 系列 ASR 模型——在英语识别上达到业界领先的准确率，并以开放许可证发布。趋势已经很清楚：最优秀的语音识别模型正越来越多地走向开源。","ecosystem":{"title":"随之而来的生态系统","items":[{"name":"whisper.cpp","desc":"——Georgi Gerganov 的 C/C++ 移植版，针对 CPU 推理进行了优化。让 Whisper 能在从笔记本到树莓派的各种设备上运行。GitHub 星标超过 46,000。"},{"name":"faster-whisper","desc":"——基于 CTranslate2 的重新实现，在保持相同准确率的同时，推理速度最高可达原版的 4 倍。"},{"name":"WhisperX","desc":"——为 Whisper 增加了强制对齐和说话人分离能力，实现词级时间戳和多说话人转录。"},{"name":"构建于开放模型之上的应用","desc":"——如今已有数十款桌面、移动和网页应用以 OpenAI Whisper 或 NVIDIA Parakeet 作为其语音识别引擎，包括 OpenWhispr、MacWhisper、Buzz 和 Vibe。"}]},"paragraph4":"OpenAI Whisper 的发布，是开源如何缔造生态系统最清晰的例证之一。一个开放模型催生了优化运行时、全新能力，以及数十款若模型仍困在 API 付费墙之后便永无诞生之日的产品。NVIDIA Parakeet 延续了这一模式，证明多家机构都看到了将高质量 ASR 开源的价值。如今，任何开发者都能构建一款拥有业界领先准确率的语音工具——无需按分钟付费，无需把音频发给第三方，也无需征得任何人许可。"},"polish":{"title":"开源并不意味着不够精致","intro":"有一种根深蒂固的成见，认为开源意味着粗糙，意味着你要用精致去换取自由。这在 2005 年或许还说得过去，但在 2026 年已不再成立。","apps":[{"name":"Firefox","desc":"一款被数亿人使用的主流浏览器"},{"name":"VS Code","desc":"全世界最受欢迎的代码编辑器"},{"name":"Signal","desc":"端到端加密、用户体验精致的即时通讯应用"},{"name":"Blender","desc":"斩获奥斯卡奖的 3D 动画与视觉特效工具"},{"name":"VLC","desc":"什么都能播，哪儿都能跑，没有广告"},{"name":"Linux","desc":"驱动着全世界绝大多数服务器和智能手机"}],"paragraph1":"开源软件的质量已大幅提升，因为激励结构发生了变化。如今有公司在开源内核之上建立业务（Red Hat、Elastic、GitLab）。资金充裕的团队全职维护项目。社区贡献帮助捕捉缺陷、添加小团队根本无暇顾及的功能。","paragraph2":"客观地说说权衡之处：开源语音工具有时团队规模更小，发布周期也比资金雄厚的专有竞品更慢。文档质量参差不齐。但这些都是现实层面的挑战，而非固有的局限。而且这一差距一直在迅速缩小——尤其在语音工具领域，像 OpenAI Whisper 和 NVIDIA Parakeet 这样的开放模型，让开源项目得以获得与任何商业产品相同的基础模型质量。"},"businessModel":{"title":"开源语音工具如何维持自身运转","intro":"一个合理的问题：如果软件是免费的，又有谁能拿到报酬来开发它呢？","paragraph1a":"答案是，“开源”和“免费”并非同义词。最可持续的开源项目会将核心引擎（免费、开放、可审计）与那些足以支撑付费版的便利功能区分开来。这被称为","openCoreModel":"开放核心模式","paragraph1b":"，它之所以行得通，是因为它让激励保持一致：公司靠把产品做得更好来赚取收入，而不是靠锁定用户或变现他们的数据。","openCore":{"title":"为何开放核心模式对注重隐私的工具尤为奏效","points":["核心产品可本地离线工作——免费版无需任何数据采集","云端功能（如托管的转录 API）提供了值得为之付费的实实在在的便利","社区贡献让产品对所有人都变得更好，付费客户也在其中","无力付费的用户依然能从中受益，并贡献缺陷报告、翻译和代码"]},"paragraph2":"这正是 OpenWhispr 所采用的模式：开源桌面应用使用 OpenAI Whisper 和 NVIDIA Parakeet 等模型在本地完成转录，完全免费。可选的 Pro 套餐为那些希望获得更快结果、或不想在自有硬件上运行模型的用户，增加了云端转录和 AI 文本清理功能。重要的是，选择权在你手中——本地、私密、功能完整的版本永远免费。"},"whatToLookFor":{"title":"挑选开源语音工具时该关注什么","intro":"并非所有“开源”的说法都等价。以下是一份用于评估语音工具的实用清单。","checklist":[{"question":"完整的源代码是否公开？","detail":"有些项目把一个库开源，却将应用本身保持专有。要确认你实际使用的产品——桌面应用、移动应用——其源代码已经公开发布。"},{"question":"你能从源代码构建吗？","detail":"无法编译的公开源代码，在实质意义上算不上开放。要留意是否有构建说明、CI 流水线，以及社区关于成功构建的反馈。"},{"question":"采用什么许可证？（MIT、Apache、GPL、AGPL？）","detail":"宽松许可证（MIT、Apache 2.0）提供最大的灵活性。Copyleft 许可证（GPL、AGPL）则确保衍生作品保持开放。两者都正当合理——只需清楚自己拿到的是什么。"},{"question":"它能离线且本地运行吗？","detail":"需要联网才能工作的开源代码，仍然会把你的数据发送到服务器。对语音工具而言，支持离线的本地处理是隐私的底线。"},{"question":"它是否在被积极维护？","detail":"查看提交历史、问题追踪器和发布节奏。一个被弃置的开源项目，可能存在未修补的安全漏洞。"},{"question":"是否存在社区？（Issues、PR、讨论）","detail":"一个健康的开源项目，贡献者不止一人。要寻找社区参与的迹象——来自外部贡献者的拉取请求、问题讨论，以及响应及时的维护者。"}]},"sources":{"title":"参考来源与延伸阅读","bloomberg":"Bloomberg：《有人在 Alexa 上监听你吗？一支全球团队在审核音频》（2019年4月）","vrt":"VRT NWS：《Google 员工在偷听，连佛兰德人的客厅也不放过》（2019年7月）","guardian":"The Guardian：《Apple 承包商“经常听到机密细节”——来自 Siri 录音》（2019年7月）","reuters":"路透社：《Apple 同意支付 9500 万美元，了结 Siri 隐私诉讼》（2025年1月）","ftcAmazon":"FTC：Amazon 因保留儿童 Alexa 录音被控违反儿童隐私法（2023年5月）","ftcCloning":"FTC：诈骗者利用 AI 语音克隆来强化家庭紧急情况骗局（2023年3月）","openaiWhisper":"OpenAI：Whisper 发布介绍（2022年9月）","whisperGithub":"OpenAI Whisper GitHub 仓库（模型、许可证、文档）","radford":"Radford 等人（2022）：《通过大规模弱监督实现稳健语音识别》","gdpr":"GDPR 官方文本（EU 2016/679）","aiAct":"欧盟《AI 法案》官方文本（EU 2024/1689）","whisperCpp":"whisper.cpp——OpenAI Whisper 的 C/C++ 移植版","fasterWhisper":"faster-whisper——基于 CTranslate2 的 Whisper 实现","openwhispr":"OpenWhispr——开放、本地优先的听写产品背景资料"},"cta":{"title":"OpenWhispr 是开源的","body":"我们以开源方式打造 OpenWhispr，因为我们相信语音软件理应透明。不必听我们的一面之词——自己查查代码吧。","starButton":"在 GitHub 上为我们点星","footnote":"无需账户 · 可离线使用 · MIT 许可 · 永远开源"},"internalLinks":{"localAi":"支持本地 AI 的理由","localVsCloud":"本地转录 vs 云端转录","allPosts":"全部博客文章","compare":"对比 OpenWhispr","useCases":"使用场景"}},"blogGoogleDocsVoiceTyping":{"metadata":{"title":"Google Docs Voice Typing：完整指南| OpenWhispr","description":"如何在 Google Docs 中打开语音输入、每个语音命令、它停止工作的原因，以及如何在任何应用程序（甚至离线）中听写。","keywords":["谷歌文档语音输入","语音输入谷歌文档","语音转文字谷歌文档","谷歌文档听写","打开语音输入 Google 文档","谷歌文档语音命令","谷歌文档语音输入不起作用","谷歌文档语音输入语言"]},"schema":{"title":"Google Docs Voice Typing：完整指南","description":"如何在 Google Docs 中打开语音输入、每个语音命令、它停止工作的原因，以及如何在任何应用程序（甚至离线）中听写。"},"breadcrumb":{"home":"首页","blog":"博客","article":"Google Docs Voice Typing"},"layout":{"title":"Google Docs Voice Typing：完整指南","subtitle":"如何在 Google Docs 中打开语音输入、主要语音命令、为什么它有时会停止工作以及如何在计算机上的其他任何地方进行口述。","date":"2026 年 6 月 26 日","tag":"指导"},"sections":[{"id":"quick","title":"快速解答"},{"id":"enable","title":"如何打开它"},{"id":"commands","title":"语音命令"},{"id":"languages","title":"语言"},{"id":"mobile","title":"在手机和平板电脑上"},{"id":"troubleshooting","title":"不工作？"},{"id":"limits","title":"局限性"},{"id":"alternative","title":"随时随地语音输入"},{"id":"faq","title":"常问问题"}],"intro":{"lead":"要在 Google Docs 中打开语音输入，请在 Chrome 中打开文档，转到“工具”>“语音输入”，或者在 Windows 上按 Ctrl+Shift+S 或在 Mac 上按 Cmd+Shift+S，然后单击麦克风并开始讲话。","body":"它是免费的、内置的，对于快速初稿非常有用。代价是它只能在桌面上的 Chrome 中运行，需要互联网连接，并且只能在 Google Docs 中输入。","summary":"下面您将找到设置步骤、重要的命令类别、支持的语言注释、故障排除修复、隐私限制以及需要在文档之外进行听写时的实用选项。"},"enable":{"title":"如何在Google Docs中开启语音输入","intro":"语音输入是 Google Docs 桌面版的一部分，专为 Google Chrome 设计。这是完整的过程：","steps":["在Chrome中打开docs.google.com处的文档。","在菜单栏中，单击工具 > 语音输入。快捷方式：Windows 上为 Ctrl+Shift+S，Mac 上为 Cmd+Shift+S。","页面左侧会出现一个麦克风框。单击它即可开始。","第一次，您的浏览器会请求麦克风权限。单击“允许”。","说话自然。当您说话时，您的话会出现在光标处。","当您想停止时，再次单击麦克风。"],"tipLead":"提示：","tipBody":"语音输入也适用于 Google Slides 的发言者备注字段，具有相同的工具菜单和麦克风控制。"},"commands":{"title":"语音命令：标点符号、格式设置和编辑","intro":"您可以说出标点符号、换行符和编辑命令。重要的一点是：格式化和编辑命令仅在文档语言设置为英语时才有效。","groups":[{"title":"标点","items":["\"period\"","\"comma\"","\"question mark\"","\"exclamation point\"","\"new line\" / \"new paragraph\""]},{"title":"格式化","items":["\"bold\" / \"italics\"","\"underline\"","\"apply heading 1\"","\"create bulleted list\"","\"align center\""]},{"title":"编辑和移动","items":["\"select [word]\"","\"delete last word\"","\"go to end of paragraph\"","\"next paragraph\"","\"stop listening\""]}]},"languages":{"title":"支持哪些语言？","paragraph1":"语音输入可以处理大约一百种语言和方言的听写。从麦克风盒的下拉菜单中选择您的。基本标点符号广泛适用于各种语言。","paragraph2":"语音格式化和编辑命令是例外。粗体、选择段落或应用标题等命令仅适用于英语，因此非英语听写通常意味着使用键盘或鼠标进行格式化。"},"mobile":{"title":"在手机和平板电脑上进行语音输入","intro":"Google Docs 移动应用程序不包含桌面语音输入工具。在手机和平板电脑上，您可以像在任何应用程序中一样进行口述：点击键盘上的麦克风。","items":["Android：当光标位于文档中时，点击 Gboard 上的麦克风或您喜欢的键盘。","iPhone和iPad：点击iOS键盘上的听写键。"],"outro":"这是您的设备听写，而不是 Google Docs 桌面功能，因此准确性、命令和行为将与 Chrome 版本不匹配。"},"troubleshooting":{"title":"语音输入不起作用？试试这个","items":[{"problem":"语音输入按钮缺失或呈灰色","fix":"在桌面上使用Google Chrome。移动应用程序不支持语音输入，并且不正式支持其他浏览器。"},{"problem":"它不会拾取我的声音","fix":"检查docs.google.com在Chrome中是否有麦克风权限，并确保其他应用没有持有麦克风。"},{"problem":"它总是自行停止","fix":"语音输入可能会在暂停后或单击远离选项卡时停止。再次单击麦克风并保持“文档”选项卡处于焦点状态。"},{"problem":"什么也没发生","fix":"您可能处于离线状态。语音输入基于云，需要互联网连接。重新连接并重新加载文档。"}]},"limits":{"title":"Google Docs Voice Typing不能做什么","intro":"这是一个真正有用的免费功能，但它是为一项工作而构建的：在 Google 文档中进行口述。如果您经常口述，这些限制很重要：","items":[{"title":"仅限Chrome。","body":"Safari、Firefox 或 Edge 中没有可靠的支持。"},{"title":"仅限在线。","body":"音频在 Google 云中处理，因此没有连接就意味着没有听写。"},{"title":"仅文档。","body":"它不会输入到 Gmail、Slack、浏览器栏、代码编辑器或任何其他应用程序中。"},{"title":"云隐私。","body":"您的声音会发送到 Google 进行转录，这对于敏感材料来说并不理想。"}]},"alternative":{"title":"想要每个应用程序都支持语音输入吗？使用OpenWhispr","introBefore":"如果您只能在文档中进行口述，那么语音输入即可完成这项工作。但是，当您想要对电子邮件、Slack 消息、代码注释、表单或笔记应用程序进行对话时，","introAfter":"涵盖了 Google Docs 没有涵盖的内容：","points":["在系统范围内的任何应用程序或文本字段中输入一键通热键","使用 OpenAI Whisper 和 NVIDIA Parakeet 在您的设备上 100% 运行，完全离线","您的音频永远不会离开您的计算机","适用于任何浏览器和桌面应用程序，而不仅仅是 Chrome","开源，所以任何人都可以检查它的功能","智能体模式可以清理并重新格式化您口述的内容"],"outro":"您可以将其视为整个计算机的 Google Docs 语音输入，无需依赖云。","viewOnGitHub":"在GitHub上查看"},"faqTitle":"常见问题解答","faqs":[{"question":"如何在Google Docs中开启语音输入？","answer":"在 Chrome 中打开文档，然后转到“工具”>“语音输入”，或者在 Windows 上按 Ctrl+Shift+S 或在 Mac 上按 Cmd+Shift+S。单击麦克风框，出现提示时允许使用麦克风，然后开始讲话。"},{"question":"为什么我的 Google Docs 语音输入不起作用？","answer":"通常的原因是使用 Chrome 以外的浏览器、缺少麦克风权限、另一个应用程序持有麦克风、暂停或未聚焦的“文档”选项卡或没有互联网连接。"},{"question":"Google Docs语音输入可以离线使用吗？","answer":"不需要。Google Docs 语音输入基于云，需要互联网连接。如果需要离线听写，可以使用OpenWhispr等本地工具。"},{"question":"Google Docs语音输入支持哪些语言？","answer":"它支持大约一百种语言和方言的听写，但语音格式和编辑命令仅在文档语言为英语时才有效。"},{"question":"Google Docs语音输入可以在iPhone或Android上使用吗？","answer":"不像桌面语音输入工具。在移动应用程序中，请改用手机键盘麦克风，例如 Android 上的 Gboard 或 iOS 听写键。"},{"question":"Google Docs语音输入私密吗？","answer":"音频会发送到 Google 进行转录，因此很方便，但不限于本地。对于机密笔记、法律工作、患者信息或未发布的业务内容，本地听写更适合。"},{"question":"我可以在 Google Docs 之外使用语音输入吗？","answer":"Google 的内置语音输入功能仅适用于 Google Docs 中，而演讲者注释则适用于 Google Slides 中。 OpenWhispr 提供了一个系统范围的热键，可以命令任何应用程序。"}],"links":{"items":[{"href":"/use-cases","label":"所有用例"},{"href":"/compare/apple-dictation","label":"VS Apple Dictation"},{"href":"/pricing","label":"定价"}]}},"useCaseMedical":{"breadcrumb":{"home":"Home","useCases":"Use Cases","medical":"For Clinicians"},"metadata":{"title":"Free Medical Dictation Software For Clinicians - OpenWhispr","description":"OpenWhispr is free, open-source medical dictation software. Dictate the visit, get the SOAP note drafted, and keep patient audio on your machine.","keywords":["medical dictation software","clinical dictation software","physician dictation","HIPAA compliant dictation software","dictation software for doctors","SOAP note dictation","offline medical transcription"]},"schema":{"name":"OpenWhispr","applicationCategory":"Medical dictation software","operatingSystem":"macOS, Windows, Linux","description":"Local-first, open-source medical dictation software. Patient audio is transcribed on the clinician's own device and never leaves it. Free for unlimited local dictation.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Medical dictation software for doctors & clinicians","title":"Dictate the visit once. Get the note drafted.","body":"OpenWhispr drafts the SOAP note, the referral letter or the discharge summary from what you dictate, on your own computer. Free for every clinician, and it keeps working on a locked-down hospital network.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Clinician walking through a sunlit hospital corridor"}},"trustBar":{"label":"Dictate into:","names":["Epic","Oracle Health","athenahealth","eClinicalWorks","NextGen","Meditech"]},"pain":{"label":"Cloud or local. Choose for the visit.","title":"A full clinic should not end with an hour of notes.","subtitle":"Dictate the findings, plan and follow-up while each visit is fresh, then review and sign the draft instead of rebuilding the day at home. Choose Cloud for speed or local processing when patient audio needs to stay on your machine.","cards":[{"title":"The whole encounter, uploaded.","body":"An ambient scribe can send the whole encounter to the cloud, including things the patient asked you not to put in the chart. OpenWhispr processes what you dictate on your device."},{"title":"Pajama-time charting.","body":"A full clinic, then an hour of notes at the kitchen table. OpenWhispr drafts the note from your dictation while the visit is fresh, so you review and sign instead of type."},{"title":"One more vendor for compliance to vet.","body":"Each cloud scribe adds a BAA, a risk assessment and another data flow to map. In local mode, no transcription vendor sits in the data path."},{"title":"\"Trust us\" is not a compliance posture.","body":"OpenWhispr's code is public on GitHub. Give your security team the repository, then choose the cloud or local path that fits your review."}]},"features":{"label":"What you get","title":"Faster charting. Same confidentiality.","subtitle":"Medical dictation software often trades privacy for speed or bills per provider every month. OpenWhispr keeps the speed without sending patient audio away or charging per clinician for local mode.","items":[{"image":{"label":"Clinician speaking beside a laptop in a medical practice"},"title":"Your mic, or the telehealth call, transcribed on-device","body":"Dictate the visit into your mic, or record a telehealth consult straight from your speakers on macOS 14.2 or later, Windows and most Linux desktops. It's transcribed locally and never uploaded. You decide what's kept and when it's deleted."},{"image":{"label":"Clinician reviewing a printed document at a hospital workstation"},"title":"SOAP notes and letters, drafted for you","body":"Save your own SOAP, referral and discharge prompts as one-click actions, then run whichever one the visit needs. You review every word before it's signed, then paste it into the chart."},{"image":{"label":"Doctor speaking while working on a laptop in an examination room"},"title":"Dictate into the chart field you're already in","body":"Press a hotkey and dictate wherever the cursor already is: the HPI box, a reply to a portal message, a refill note in your inbox. Clean, punctuated text, processed locally."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Patient audio sent to us","value":"None","footnote":"in local mode, transcription never leaves your device"},{"label":"Cost per clinician","value":"$$0","footnote":"unlimited local dictation, the whole practice, forever"},{"label":"Local dictation is","value":"Free","footnote":"unlimited on every plan"}]},"howItWorks":{"label":"How it works","title":"The note drafted before the next patient is roomed.","subtitle":"Most medical dictation is a per-provider cloud subscription or an outsourced transcription service. OpenWhispr lives on your machine, with no account or credit card required.","image":{"label":"Doctor reaching for a closed laptop in an examination room"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Pick a local model during setup."},{"title":"Dictate the visit.","body":"Speak your findings, plan and follow-up, and it's transcribed on your device."},{"title":"Review, sign, delete.","body":"Edit the draft note, paste it into the chart, and delete the audio. You control all of it."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons clinicians choose OpenWhispr","subtitle":"Choosing medical dictation software is a patient-confidentiality decision before it's anything else. Here are the facts, and your compliance officer can verify every one in the repo or at trust.openwhispr.com.","items":[{"title":"Audio never leaves your device","body":"Dictation, and note drafts with a local model, are processed on your machine. The open-source code shows there's no upload step."},{"title":"Open source, auditable by anyone","body":"Your IT or compliance team can read every line on GitHub. No faith required."},{"title":"Retention is yours, not a vendor's","body":"Audio and transcripts are files on your machine, not rows in someone else's retention schedule. Delete the audio the moment the note is signed."},{"title":"Optional Corti clinical transcription","body":"For specialist terminology, Corti Transcribe is available as a clinical-grade medical transcription provider. Choose a local model for note drafting that stays on-device."},{"title":"No bot joins the telehealth call","body":"Nothing appears in the participant list, so you never have to explain a third party to the patient before you can begin."},{"title":"Works fully offline","body":"Ward round, rural clinic, a locked-down hospital network. No signal required, so it never stalls mid-dictation."},{"title":"Dictate into any EHR","body":"A system-wide hotkey drops clean text wherever your cursor is: Epic, athenahealth, a referral letter, an email."},{"title":"Free for the whole practice","body":"Unlimited local dictation at $0, every clinician in the group, forever. No per-provider licence to renew, and you can test it on your own clinic list before anyone signs anything."},{"title":"SOC 2 Type 2 audit complete. HIPAA attestation. ISO 27001-aligned.","body":"Your security review can start at trust.openwhispr.com. Enterprise includes custom data processing agreements and compliance reviews."},{"title":"100+ languages","body":"Dictate the history in the language the visit actually happened in, transcribed locally."}]},"faq":{"label":"Before you download","title":"The questions clinicians ask first.","subtitle":"Straight answers on privacy, compliance and accuracy. If yours isn't here, ask us in Discord.","items":[{"question":"Does patient audio ever leave my device?","answer":"Not in local mode. Pick a local model during setup and transcription runs on your machine; the open-source code shows there is no upload step. Cloud transcription remains optional."},{"question":"Is it HIPAA compliant?","answer":"Compliance depends on your workflow, not the tool alone. OpenWhispr's SOC 2 Type 2 audit is complete, with the final report available under NDA once issued; it has a HIPAA attestation and an ISO 27001-aligned information-security programme. For PHI, local-only and bring-your-own-key modes route content around OpenWhispr Cloud. Cloud voice audio is processed in real time and not persisted after transcription. Enterprise includes custom data processing agreements and compliance reviews."},{"question":"Can it handle medical terminology?","answer":"Yes. It runs Whisper-class models trained on a wide range of accents and 100+ languages. The custom dictionary feeds the words that trip everything else up to both the transcriber and the AI cleanup: metoprolol succinate ER, cholecystectomy, the abbreviations your specialty writes and nobody else reads."},{"question":"How is this different from an ambient AI scribe?","answer":"An ambient scribe listens to the whole encounter, sends it to a vendor's cloud, and bills per provider every month. Medical dictation software is the other shape: you say what belongs in the note, a local model drafts it from your own prompt, and it stays on your machine at $0. For a telehealth consult it captures your speakers directly, so nothing joins the call and nothing is uploaded."},{"question":"Does it draft the note, or just transcribe?","answer":"Both. Run a local model (Llama, Qwen, Gemma) or connect your own AI key, then save your own SOAP, referral and discharge prompts as one-click actions and run whichever one the visit needs. A local model keeps the PHI on your machine; your own key means the model and data path are ones you already trust."},{"question":"Is it actually free?","answer":"Yes. Unlimited local dictation, 100+ languages, all three desktop platforms: $0 forever, for every clinician. Pro ($8/month) adds more cloud transcription and mobile dictation if you want them."}]},"reviews":{"label":"What users say","title":"Built by people who don't want to be recorded either.","subtitle":"From our Discord, GitHub and inboxes, the same instinct that matters in the exam room.","score":{"value":"0","label":"Patient recordings on a server","body":"In local mode, audio stays on your device and is never stored on a server."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for doctors & clinicians","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Clinicians who want free, private, offline dictation and are fine with a quick local setup"}]},{"parts":[{"text":"Clinicians who need unlimited cloud transcription, cross-device sync and mobile access"}]},{"parts":[{"text":"Practices and teams that need unlimited meeting recordings, speaker labels and Agent mode"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Free offline dictation, with patient audio staying on your device"}]},{"parts":[{"text":"Unlimited cloud transcription without managing local models or keys"}]},{"parts":[{"text":"Unlimited meeting recordings, speaker labels and workflow features for a team"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some conversations stay in the exam room. We built for those.","paragraphs":["OpenWhispr started with a simple objection: what a patient tells you in the exam room shouldn't have to pass through a stranger's server to become a note.","Our code is public on GitHub. In local mode, transcription runs on the device with no internet required.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Sign the last note before you leave the building.","body":"Download OpenWhispr. Dictate the visit, draft the note, then paste it into the chart while it is still fresh. Choose local models when processing needs to stay on your computer.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free unlimited local dictation on Mac, Windows & Linux."}}},"authGate":{"providers":{"google":"Google","microsoft":"Microsoft","apple":"Apple"},"continueWithProvider":"继续 {provider}","or":"或","fields":{"email":"电子邮件地址","password":"密码","fullName":"全名","passwordHint":"至少 8 个字符"},"actions":{"checking":"正在检查...","continueWithEmail":"继续使用电子邮件","back":"返回","signingIn":"正在登录...","signIn":"登录","createAccount":"创建帐户","tryAgain":"再试一次","sending":"正在发送...","sendResetLink":"发送重置链接","creatingAccount":"正在创建帐户...","resendVerificationEmail":"重新发送验证邮件","useDifferentEmail":"使用不同的电子邮件","useSso":"使用单点登录 (SSO)"},"status":{"completingSignIn":"正在完成登录...","redirecting":"正在重定向..."},"errors":{"generic":"出了点问题。请再试一次。","authenticationTimedOut":"身份验证超时。请再试一次。","sessionExpired":"会话已过期。请尝试重新注册。","waitingForVerification":"仍在等待验证。检查您的收件箱或重新发送电子邮件。","unableToCheckEmail":"无法检查此电子邮件。请再试一次。","network":"网络错误。请检查您的连接。","networkRetry":"网络错误。请检查您的连接并重试。","incorrectPassword":"密码不正确。请再试一次。","accountExists":"使用此电子邮件的帐户已存在。请登录。","createAccountFailed":"创建帐户失败。请再试一次。","resendFailed":"重发失败。请再试一次。"},"signin":{"title":"欢迎回来","forgotPassword":"忘记密码？","newHere":"新来的？"},"forgot":{"title":"重置您的密码","description":"输入您的电子邮件，我们将向您发送重置链接","emailSentTitle":"检查您的电子邮件","emailSentBody":"我们将密码重置链接发送至","didNotReceive":"没有收到电子邮件？检查您的垃圾邮件文件夹或重试。"},"signup":{"title":"创建您的帐户","haveAccount":"有帐户吗？"},"verifyEmail":{"title":"检查您的电子邮件","message":"我们将验证链接发送至"},"legal":{"line":"继续即表示您同意我们的<termsLink>服务条款</termsLink>和<privacyLink>隐私政策</privacyLink>。"}},"getStarted":{"title":"开始使用","subtitle":"下载 OpenWhispr，免费且开源。","processing":"开始下载...","skip":"无需帐户即可继续","verificationMessage":"请验证邮箱以开始下载。链接已发送至","metadata":{"title":"开始使用 | OpenWhispr","description":"创建 OpenWhispr 账户，开始使用私密的 AI 语音听写。"}},"checkout":{"title":"开始使用 OpenWhispr","processing":"正在重定向至结帐...","period":{"month":"/月","year":"/年"},"plan":{"proMonthly":"Pro · 每月","proAnnual":"Pro·年度","businessMonthly":"Business · 每月","businessAnnual":"Business·年度"},"badge":{"twoMonthsFree":"2个月免费"},"metadata":{"title":"结账 | OpenWhispr","description":"安全地完成 OpenWhispr 订阅购买。"}},"checkoutSuccess":{"appRedirect":{"title":"正在重定向至 OpenWhispr...","body":"您可以关闭此选项卡","fallback":"如果应用程序未打开，请点按此处"},"success":{"title":"欢迎来到Pro！","body":"您的订阅现已激活"},"download":{"autoNotice":"您的下载应该会自动开始...","fallback":"如果未启动，请单击此处","open":"下载OpenWhispr","all":"查看所有下载"},"linuxOptions":[{"label":"Linux AppImage (x86_64)","meta":"AppImage"},{"label":"Linux DEB (amd64)","meta":"Debian 软件包"},{"label":"Linux RPM (x86_64)","meta":"RPM封装"},{"label":"Linux tar.gz (x64)","meta":"便携式压缩包"}],"metadata":{"title":"结账成功 | OpenWhispr","description":"OpenWhispr 结账已完成，你的订阅现已可用。"}},"upgrade":{"expiredTitle":"链接已过期","redirecting":"正在重定向至结帐...","errors":{"missingToken":"缺少升级令牌。请检查您的电子邮件链接。","invalidLink":"该链接不再有效。您可以从应用程序内升级。","generic":"出了点问题。请重试或从应用程序内升级。"},"metadata":{"title":"升级 | OpenWhispr","description":"升级你的 OpenWhispr 套餐，解锁更多语音听写功能。"}},"authCallback":{"failedTitle":"登录失败","calendarFailedTitle":{"gcal":"无法连接 Google 日历","mcal":"无法连接 Microsoft 日历"},"redirecting":"重定向至OpenWhispr","closeTab":"您可以关闭此选项卡","desktopFallback":"如果OpenWhispr没有打开请点击这里","mobileFallback":"如果应用程序未打开，请点按此处","backToSignIn":"返回登录","desktopNextStep":"请返回 OpenWhispr 应用重新登录。","errors":{"access_denied":"未获得权限，所以您未登录。如果您拒绝了请求，请重新登录并选择允许。如果需要您所在组织的 IT 管理员先批准 OpenWhispr，请让他们允许后重新登录。","account_not_linked":"此电子邮件地址已有 OpenWhispr 帐户，且使用其他登录方式。请用您第一次登录的方式登录，例如用密码。","state_mismatch":"此登录链接已过期或已被使用，所以您未登录。请重新开始登录，并在 5 分钟内完成。","state_not_found":"此登录链接已过期或已被使用，所以您未登录。请重新开始登录，并在 5 分钟内完成。","please_restart_the_process":"此登录链接已过期或已被使用，所以您未登录。请重新开始登录，并在 5 分钟内完成。","invalid_state":"此登录链接已过期或已被使用，所以您未登录。请重新开始登录，并在 5 分钟内完成。","invalid_code":"登录已开始但未能完成，所以您未登录。请重新登录。如果再次失败，请发邮件至 support@openwhispr.com。","unable_to_get_user_info":"我们无法从您的 Google、Microsoft 或 Apple 帐户读取您的姓名和电子邮件地址，所以您未登录。请重新登录。","noSession":"无法确认您的登录，所以 OpenWhispr 应用仍未登录。请返回应用重新登录。","default":"出了点问题，您未登录。请重新登录。如果一直失败，请发邮件至 support@openwhispr.com。","sso_required":"贵公司要求使用自己的登录方式，即 SSO。请返回 OpenWhispr 应用，用工作邮箱登录。应用会使用贵公司的 SSO。","account_mismatch":"您登录的帐户与您输入的帐户不同。请用输入的帐户登录，或继续使用已登录的帐户。","consent_required":"您所在组织的 IT 管理员尚未批准 OpenWhispr。请让他们允许后再登录。现在重试无效。","admin_policy_enforced":"您所在组织的 IT 管理员尚未批准 OpenWhispr。请让他们允许后再登录。现在重试无效。","user_cancelled_authorize":"您在完成前关闭了登录窗口，所以您未登录。请重新开始登录。","server_error":"您用于登录的服务那边出了问题，登录未能完成。请等几分钟后重新登录。","slack_connect_invalid":"此 Slack 连接链接不完整或无效，因此 Slack 尚未连接。请回到 OpenWhispr 应用重新连接 Slack。","linear_connect_invalid":"此 Linear 连接链接不完整或无效，因此 Linear 尚未连接。请回到 OpenWhispr 应用重新连接 Linear。"},"errorTitles":{"cancelled":"登录已取消","permissionNotGranted":"未获得权限","approvalNeeded":"需要批准","linkExpired":"登录链接已过期","companySignIn":"需要公司登录","wrongAccount":"帐户不符","slackConnectFailed":"无法连接 Slack","linearConnectFailed":"无法连接 Linear"},"calendarErrors":{"access_denied":"访问被拒绝。请再试一次并授予日历权限。","no_email":"无法读取此日历账户的电子邮件地址。请再试一次。","token_exchange_failed":"无法完成与您的日历提供商的连接。请再试一次。","server_error":"连接时出错。请检查网络连接后再试一次。","admin_policy_enforced":"需要您所在组织的管理员批准 OpenWhispr 访问日历。","consent_required":"需要您所在组织的管理员批准 OpenWhispr 访问日历。","default":"无法连接您的日历。请再试一次。"},"metadata":{"desktopCallback":{"title":"正在重定向… | OpenWhispr","description":"正在完成 OpenWhispr 桌面版登录并将你重定向到应用。"},"mobileCallback":{"title":"正在重定向… | OpenWhispr","description":"正在完成 OpenWhispr 移动版登录并将你重定向到应用。"},"error":{"title":"登录错误 | OpenWhispr","description":"无法完成 OpenWhispr 登录。请查看错误并重试。","connectTitle":"{heading} | OpenWhispr"}},"mismatch":{"switchAccount":"使用其他帐户登录","continueAnyway":"继续使用已登录的帐户"}},"notFound":{"title":"哎呀！找不到页面","subtitle":"看起来此页在转录过程中丢失了！","quip":"别担心，我们思考的速度比打字的速度快......","description":"您要查找的页面可能已被移动、删除或暂时不可用。","home":"带我回家","back":"返回","autoRedirectLabel":"自动重定向：","autoRedirectBody":"{seconds} 秒带你回家..."},"verifyEmail":{"loading":"正在验证您的电子邮件...","errors":{"noToken":"未找到验证令牌。请检查您的电子邮件链接。","failed":"验证失败。该链接可能已过期。","network":"网络错误。请检查您的连接并重试。"},"resend":{"signInFirst":"请先登录，然后请求新链接。","failed":"无法重新发送。请再试一次。","sent":"验证邮件已发送！检查您的收件箱。","network":"网络错误。请再试一次。"},"success":{"title":"邮箱已验证！","desktop":"如果您没有自动重定向，请点击下面。","browser":"您可以关闭此选项卡并返回到您的注册以继续。","openApp":"打开OpenWhispr"},"actions":{"sending":"正在发送...","requestNewLink":"请求新链接"},"metadata":{"title":"验证邮箱 | OpenWhispr","description":"验证你的电子邮箱地址，以完成 OpenWhispr 账户设置。"}},"resetPassword":{"loading":"正在加载...","errors":{"passwordMismatch":"密码不匹配。","passwordTooShort":"密码必须至少为 8 个字符。","network":"网络错误。请检查您的连接并重试。"},"invalid":{"title":"重置链接无效","description":"此密码重置链接无效或已过期。","cta":"转到开始使用"},"success":{"title":"密码已更新","description":"您现在可以使用新密码登录。","cta":"继续登录"},"form":{"title":"创建新密码","description":"在下面输入您的新密码","newPassword":"新密码","confirmPassword":"确认新密码","passwordHint":"至少 8 个字符","resetting":"正在重置...","submit":"重置密码"},"metadata":{"title":"重置密码 | OpenWhispr","description":"为你的 OpenWhispr 账户设置新密码。"}},"consent":{"banner":{"ariaLabel":"Cookie 同意","title":"我们重视您的隐私","eu":"我们使用 cookie 来衡量使用情况和我们的营销。在您允许之前，分析和营销 cookie 会保持关闭状态。","us":"我们和我们的合作伙伴使用 cookie 进行分析和广告。您可以随时选择退出。","row":"我们使用 cookie 来了解使用情况并改进 OpenWhispr。","gpc":"我们检测到 Global Privacy Control 信号并已选择让您退出广告 cookie。","policyLine":"请查看我们的<cookieLink>Cookie 政策</cookieLink>和<privacyLink>隐私政策</privacyLink>。","rejectAll":"全部拒绝","customize":"定制","acceptAll":"全部接受","privacyChoices":"您的隐私选择","cookieSettings":"Cookie 设置","gotIt":"明白了","updated":"自您上次访问以来，我们更新了 Cookie 政策。您当前的选择将持续有效，直到您进行更改。"},"settings":{"title":"Cookie 偏好设置","description":"选择 OpenWhispr 可以使用哪些 cookie。您可以随时从页脚更改此设置。","necessary":{"title":"绝对必要","description":"登录和安全所需。始终活跃。","ariaLabel":"绝对必要的cookie（始终开启）"},"analytics":{"title":"分析","description":"帮助我们理解用法（Google Analytics 4）。","ariaLabel":"分析 cookies"},"marketing":{"title":"营销","description":"衡量我们的广告和营销活动（Google Ads、Meta）以及创作者推荐（Dub）的效果。","gpcDescription":"关闭 — 您的浏览器的 Global Privacy Control 信号会选择您退出。","ariaLabel":"营销cookie"},"rejectAll":"全部拒绝","acceptAll":"全部接受","save":"保存首选项"},"trigger":{"privacyChoices":"您的隐私选择","cookieSettings":"Cookie 设置"}},"referral":{"someone":"有人","loading":"正在加载...","invitedBy":"{name} 邀请","invited":"你被邀请了","title":"解锁 1 个月免费 OpenWhispr Pro","claimPrompt":"注册即可领取 Pro 的免费一个月","continueWithGoogle":"继续Google","or":"或","processing":"正在设置您的帐户...","alreadyHaveAccount":"已经有帐户？","downloadOpenWhispr":"下载OpenWhispr","fields":{"email":"电子邮件地址","password":"密码","fullName":"全名","passwordMin":"密码（最少 8 个字符）"},"actions":{"checking":"正在检查...","continueWithEmail":"继续使用电子邮件","signingIn":"正在登录...","signInClaim":"登录并领取优惠","useDifferentEmail":"使用不同的电子邮件","creatingAccount":"正在创建帐户...","createClaim":"创建帐户并领取优惠","useSso":"使用单点登录 (SSO)"},"signin":{"as":"登录身份"},"signup":{"for":"创建帐户"},"legal":{"line":"继续即表示您同意我们的<termsLink>条款</termsLink>和<privacyLink>隐私政策</privacyLink>。"},"errors":{"network":"网络错误。请再试一次。","incorrectPassword":"密码不正确。请再试一次。","accountExists":"帐户已存在。请登录。","createAccountFailed":"创建帐户失败。","signInTimedOut":"登录超时，请重试。","signInFailed":"登录失败。","ssoEmailRequired":"请在上方输入你的工作邮箱以使用单点登录。","ssoFailed":"单点登录失败。"},"spectrogram":{"friend":"一个朋友","signalId":"信号ID","scanAriaLabel":"扫描音频签名","scan":"扫描","stop":"停止","replay":"重播","unlocked":"解锁","freeMonth":"1 个月免费 Pro","signature":"{name}的音频签名","scanning":"正在扫描..."},"continueWithProvider":"使用 {provider} 继续","providers":{"google":"Google","microsoft":"Microsoft","apple":"Apple"},"metadata":{"title":"推荐邀请 | OpenWhispr","description":"通过推荐链接加入 OpenWhispr，开始使用私密的 AI 语音听写。"}},"demo":{"ariaLabel":"产品演示","tabpanelFallback":"演示","tabs":[{"label":"语音转文本"},{"label":"会议记录"},{"label":"AI 对话"},{"label":"音频上传"}],"panelLabels":["语音转文本演示","会议记录演示","人工智能聊天演示","音频上传演示","自定义词典演示","集成演示","历史演示"],"sidebar":{"search":"搜索","settings":"设置","userName":"Alex Chen","userEmail":"alex@company.com","nav":{"home":"首页","notes":"注释","upload":"上传","dictionary":"词典","integrations":"集成"}},"speech":{"dictationText":"嘿，团队，我刚刚审查完 Q2 路线图，我认为我们应该在新集成之前优先考虑语音管道。","withoutOpenWhispr":"没有 OpenWhispr","slowWpm":"~40 WPM","messageApp":"Slack — #product","placeholder":"输入消息...","withOpenWhispr":"使用 OpenWhispr","fastWpm":"~150 WPM","faster":"速度提高 3 倍"},"history":{"transcriptions":{"title":"转录","items":[{"time":"上午 9:15","text":"季度指标看起来很有希望。收入环比增长 12%，用户保留率显着提高。"},{"time":"下午 4:30","text":"让我们安排下周的设计评审。我希望整个团队在发货前看到新组件。"}]},"upcomingMeetings":{"title":"即将召开的会议","items":[{"time":"上午 11:00 - 上午 11:30","title":"设计评审"},{"time":"下午 2:00 - 下午 2:45","title":"冲刺计划"},{"time":"下午 4:00 - 下午 4:30","title":"1:1 与 Sarah"}]}},"notes":{"rawNotes":"- 3月份发布语音管道\n- 与设计团队讨论新用户界面\n- 预算问题：询问财务","meetingTitle":"冲刺计划","title":"产品同步","recording":"录音","you":"你","date":"2026 年 3 月 11 日","foldersLabel":"文件夹","notesLabel":"笔记","viewTabs":{"transcript":"转录稿","notes":"笔记","enhanced":"已增强"},"sections":[{"heading":"决定","items":["语音流程是最高优先级：在其他工作流之前，于 3 月底前发布","UI 重新设计推迟到语音流程落地之后，避免分散注意力","批准的词典功能：医学、法律和技术术语的自定义单词"]},{"heading":"行动项目","items":["Sarah：在冲刺结束时前端迁移到新的组件库","Alex：API 重构 + 延迟基准（目标 sub-200ms，当前 ~280ms）","Gabriel：跟进 Q2 预算的财务情况，需要在周五之前批准人数","新编辑器 UI 的设计评审 - 下周二下午 2 点，整个团队"]},{"heading":"开放性问题","items":["笔记上的实时协作 — CRDT 与 OT 方法？还没有时间表","公开开源发布之前需要第三方依赖审核","如何处理移动设备的离线同步——本地优先还是基于队列？"]}],"done":"完成","primaryAction":"套用模板","enhanceNotes":"增强笔记","actions":[{"name":"同步模板","description":"使用团队模板格式化"},{"name":"会议纪要","description":"总结决策和行动"},{"name":"电子邮件草稿","description":"后续电子邮件草稿"}],"folders":[{"name":"个人","count":5,"isActive":true},{"name":"会议","count":3,"isActive":false},{"name":"工作","count":2,"isActive":false}],"noteList":[{"title":"产品同步","preview":"每周同步的行动项目...","time":"上午 10:30"},{"title":"语音管道设计","preview":"v2 的架构说明...","time":"上午 9:15"},{"title":"每周站立会","preview":"冲刺进度和阻碍...","time":"昨天"}]},"aiChat":{"title":"人工智能聊天","listening":"听...","askAnything":"询问有关您笔记的任何问题","speakQuestion":"说出你的问题","voicePoweredQa":"语音驱动的数据问答","userMessage":"总结一下我们今天会议讨论的内容","response":"根据今天的讨论：\n\n• 语音管道 — 3 月底发货\n• UI 重新设计推迟到后期流程\n• 跟进Q2 预算的财务情况","releaseToFinish":"释放 {hotkey} 完成","holdToSpeak":"按住{hotkey}说话"},"upload":{"noteTitle":"设计系统迁移规划","transcript":"该团队同意继续推进新的设计系统。 Sarah 将领导前端迁移，Alex 将在冲刺结束时处理 API 重构。","now":"现在","preview":"团队一致致力于迁移到新的设计系统...","dateMeta":"2026 年 3 月 12 日 · 847 字","structuredIntro":"团队一致同意迁移到新的基于组件的设计系统。前端迁移以 Sarah 领导组件审核开始本次冲刺。 Alex 将并行重构 API 层。","sections":[{"heading":"关键决策","items":["在所有产品表面采用新的设计标记","前端迁移逐个组件处理","API 重构范围为身份验证和数据获取层"]},{"heading":"行动项目","items":["Sarah — 组件审核和迁移计划（周五截止）","Alex — 范围 API 重构；星期三之前起草 RFC","Jamie — 更新 CI 管道以获得新的令牌格式","团队 — 在下一次站立之前回顾 Figma 库 v2"]},{"heading":"时间轴","items":["Sprint 14：审核 + 迁移计划 + API RFC","Sprint 15：组件迁移、设计令牌","Sprint 16：集成测试，弃用旧版本"]}],"dropPrompt":"拖放音频文件，或点击浏览","formats":"MP3、WAV、M4A、WebM、OGG、FLAC、AAC","fileName":"meeting-recording.mp3","fileMeta":"12:04 持续时间 · 18.3 MB","folder":"个人","transcribe":"转录","transcribing":"本地转录...","complete":"转录完成","openNote":"打开笔记","newUpload":"新建上传"},"dictionary":{"title":"自定义词典","clearAll":"全部清除","placeholder":"添加单词或短语...","hint":"单词作为提示发送，以提高转录准确性。用逗号分隔多个单词。"},"integrations":{"title":"集成","subtitle":"连接您最喜欢的工具","calendarDescription":"自动加入会议并实时转录","connected":"已连接","comingSoon":"更多集成即将推出"}},"blogComponents":{"author":{"defaultRole":"工程","updated":"更新了 {date}"},"layout":{"defaultTag":"博客","tableOfContents":"目录","onThisPage":"在此页面上"},"diarizationPipeline":{"title":"本地说话人分离管道","steps":[{"label":"音频捕捉","sub":"麦克风+系统，16kHz"},{"label":"VAD","sub":"Silero, 2MB"},{"label":"细分","sub":"pyannote 3.0"},{"label":"嵌入","sub":"CAM++, 512-dim"},{"label":"聚类","sub":"Agglomerative, 0.5"},{"label":"标签","sub":"匹配 SQLite 中的配置文件"}],"whyLocal":"为什么它保持本地化：","bullets":["每个阶段都通过单个 sherpa-onnx 二进制文件在您的设备上运行 - 没有 Python，没有云。","ONNX 模型中的 ~45MB 下载一次，然后缓存在 ~/.cache/openwhispr/diarization-models/. 中","嵌入是 512-dim float32 向量，在本地磁盘上存储为 SQLite BLOBs 。"]},"linuxSpectrum":{"title":"Linux 听写景观 (2026)","left":"精致的用户体验","right":"DIY控制"},"windowsSpectrum":{"title":"Windows 听写景观 (2026)","left":"人工智能驱动","right":"专业化"},"localVsCloud":{"title":"本地与云：关键权衡","local":{"title":"本地（OpenWhispr)","points":[{"label":"隐私","detail":"音频保留在设备上"},{"label":"延迟","detail":"无网络往返"},{"label":"成本","detail":"没有按分钟 API 计费"},{"label":"离线","detail":"无需互联网即可工作"},{"label":"控制","detail":"确定性局部行为"}]},"cloud":{"title":"云 APIs","points":[{"label":"隐私","detail":"策略+配置依赖"},{"label":"延迟","detail":"网络和区域敏感"},{"label":"成本","detail":"基于使用情况的计费"},{"label":"离线","detail":"不可用"},{"label":"规模","detail":"轻松应对大批量工作负载"}]},"recommendation":"最佳实用模式：本地优先，仅在需要时将边缘情况路由到云。"},"openSourceTrust":{"title":"为什么开源可以提高语音数据信任","layers":[{"title":"可审计性","detail":"直接在代码中检查数据流"},{"title":"再现性","detail":"从源代码构建并验证二进制文件"},{"title":"分叉权利","detail":"如果维护者改变方向，用户仍能保持控制权"},{"title":"无硬锁定","detail":"迁移堆栈而不丢失工作流程"}]},"voicePrivacyFlow":{"title":"语音数据路径：本地与云风险面","microphone":"麦克风","localInference":"局部推理","localText":"本地文本","zeroExposure":"零暴露","cloudApi":"云API","remoteText":"远程文本","networkExposure":"网络曝光","defaultPath":"OpenWhispr 默认：本地路径优先，仅在显式配置时可选云。"},"whisperArchitecture":{"title":"Whisper 如何处理音频","steps":["音频输入","Log-Mel 特性","编码器","解码器","文字"],"whyItMatters":"为什么它对 OpenWhispr: 很重要","bullets":["通过 whisper.cpp 在本地运行 — 原始音频保留在设备上。","模型大小控制速度/准确性的权衡。","相同的管道支持 macOS、Windows 和 Linux 之间的听写。"]},"werChart":{"title":"词错误率：Whisper 与开源竞争对手","subtitle":"LibriSpeech test-clean（英语朗读演讲）——越低越好","werLabel":"WER","turboNote":"速度提高 8 倍","otherOpenSource":"其他开源","sources":"来源：Hugging Face 模型卡、NVIDIA 模型卡、Open ASR Leaderboard (Feb 2026)。仅限纯英语朗读语音 - 现实世界的准确性因音频质量、口音和领域而异。商业 API 被省略，因为它们不发布 LibriSpeech 基准。"},"modelTradeoff":{"title":"Whisper 模型大小：速度与准确性","subtitle":"气泡大小反映了相对参数数量","ariaLabel":"Whisper 模型速度与精度权衡的散点图","fasterAxis":"更快 →","accuracyAxis":"更准确 →","recommendation":"建议：从小处开始，根据硬件进行调整。"},"openAsrWerChart":{"averageWer":"平均 WER","title":"英语准确率：平均词错误率","caption":"Hugging Face Open ASR Leaderboard 基准测试中多个数据集的平均值（越低越好）。Nemotron 以流式配置展示。","sources":"来源：Hugging Face 上的 NVIDIA 模型卡和 Open ASR Leaderboard，2026 年 7 月。WER = 转写错误的词所占百分比。"},"streamingAsrFlow":{"title":"OpenWhispr 中的实时流式处理路径"}},"compareFathom":{"metadata":{"title":"Fathom 替代方案——本地、私密、开源","description":"在寻找 Fathom 替代方案？OpenWhispr 在本地处理语音转文字和会议笔记，无需机器人、云端或账户。免费且开源。","keywords":["Fathom 替代方案","Fathom AI 替代方案","私密 Fathom 替代方案","Fathom 开源替代方案","本地会议笔记","无机器人会议笔记"]},"breadcrumb":{"home":"首页","comparisons":"对比","fathom":"对比 Fathom"},"hero":{"title":"私密、开源的 Fathom 替代方案","description":"OpenWhispr 是为希望在自己的设备上 100% 运行<voice>语音转文字</voice>和<meetings>会议笔记</meetings>的用户打造的 Fathom 替代方案。没有机器人加入通话，音频绝不离开设备，整个应用均为开源。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"速览","fathomIntro":"是一款精致的云端会议笔记工具，以慷慨的免费方案闻名。它的机器人会加入 Zoom、Teams 和 Meet 通话，录音并转录，几乎无需设置即可生成清晰的摘要、重点和待办事项。它是最容易上手的自动会议回顾工具之一。","openWhisprIntro":"专注于 Fathom 不提供的能力：让音频留在自己的设备上。这款语音转文字与设备端会议笔记工具基于 OpenAI Whisper 和 NVIDIA Parakeet。没有机器人加入，不需要账户，也不会上传任何内容。按下快捷键即可在任意应用中语音输入，或在本地录制并转录会议。","chooseFathom":"如果你想要最轻松的自动会议摘要、强大的免费方案，并且不介意云端处理，请选择 Fathom。","chooseOpenWhispr":"如果你想要私密、本地的语音转文字和会议笔记，并且绝不上传音频，请选择 OpenWhispr。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"隐私与数据控制","openWhispr":{"status":"full","details":"100% 本地——会议音频留在设备上"},"competitor":{"status":"none","details":"云端处理——录音存储在服务器上"}},{"feature":"开源","openWhispr":{"status":"full","details":"MIT 许可，可在 GitHub 审查"},"competitor":{"status":"none","details":"专有闭源"}},{"feature":"离线运行","openWhispr":{"status":"full","details":"本地模型无需互联网"},"competitor":{"status":"none","details":"需要互联网连接"}},{"feature":"会议捕获","openWhispr":{"status":"partial","details":"本地录制系统音频，无机器人"},"competitor":{"status":"full","details":"机器人自动加入 Zoom、Teams 和 Meet"}},{"feature":"通用听写","openWhispr":{"status":"full","details":"在任意应用中进行系统级语音输入"},"competitor":{"status":"none","details":"仅限会议，并非听写工具"}},{"feature":"转录与语言","openWhispr":{"status":"full","details":"Whisper + Parakeet，支持 100+ 种语言"},"competitor":{"status":"full","details":"云端转录与 AI 摘要"}},{"feature":"免费方案","openWhispr":{"status":"full","details":"本地听写无限使用，永久免费"},"competitor":{"status":"full","details":"面向个人的慷慨免费方案"}},{"feature":"团队与 CRM 功能","openWhispr":{"status":"partial","details":"API/MCP 访问；Business 基础团队功能"},"competitor":{"status":"full","details":"付费方案提供 CRM 同步与团队功能"}}],"whyOpenWhispr":{"title":"为什么选 OpenWhispr？","items":[{"title":"默认本地运行：","body":"转录在你的设备上运行。参与者列表中没有机器人，不上传音频，也不存入第三方服务器。"},{"title":"随处听写：","body":"系统级快捷键可将语音输入任何应用——邮件、文档、编辑器或 AI 聊天。Fathom 仅用于会议。"},{"title":"开源：","body":"采用 MIT 许可且可审查，你可以准确验证音频的处理方式。"}]},"whyFathom":{"title":"为什么选 Fathom？","items":[{"title":"轻松生成会议回顾：","body":"机器人自动加入并总结通话，会议结束时重点和待办事项已准备就绪。"},{"title":"慷慨的免费方案：","body":"个人用户云端会议笔记工具中功能最强的免费方案之一。"},{"title":"精致且配置简单：","body":"清晰的摘要和 CRM 同步，多数团队几乎无需配置即可采用。"}]},"faq":{"title":"常见问题"},"faqs":[{"question":"最佳 Fathom 替代方案是什么？","answer":"这取决于你的需求。如果你需要私密、开源的日常语音转文字和设备端会议笔记，OpenWhispr 是可靠选择：它使用 OpenAI Whisper 和 NVIDIA Parakeet 在本地 100% 处理语音，通过按键说话快捷键在整个系统中工作，绝不将音频发送到云端。如果你主要想要最轻松的自动会议摘要和慷慨的免费方案，Fathom 本身非常出色。"},{"question":"有免费的 Fathom 替代方案吗？","answer":"有。OpenWhispr 免费且开源。在 macOS、Windows 和 Linux 上，可离线免费使用 100+ 种语言的无限本地听写。可选 Pro（$8/月）增加无限云端转录、更多会议录制时长、移动端和设备同步，但核心本地语音转文字永远不需要它。"},{"question":"有本地或私密的 Fathom 替代方案吗？","answer":"OpenWhispr 正是为此而建。核心转录在你的设备上进行，音频绝不离开设备，并可完全离线工作。应用开源，因此可准确审查数据处理方式。通话中没有机器人，也不会上传内容，这对人力资源、法律或涉及客户数据的敏感会议尤为重要。"},{"question":"OpenWhispr 会像 Fathom 一样让机器人加入会议吗？","answer":"不会，这正是关键。Fathom 会让机器人出现在通话中，并将音频上传到云端。OpenWhispr 直接在设备上捕获会议的系统音频，因此参与者列表中没有机器人，也不会上传任何内容。你仍应告知他人正在录音，但对话完全留在自己的设备上。"},{"question":"OpenWhispr 与 Fathom 有何不同？","answer":"Fathom 是云端会议笔记工具，专注于轻松的自动摘要和强大的免费方案。OpenWhispr 是隐私优先的语音转文字与设备端会议笔记工具：按下快捷键即可在任意应用中说话，或在本地录制和转录会议。两者解决不同问题；很多人用云端工具处理日常通话，用 OpenWhispr 进行私密听写和敏感会议。"}],"cta":{"title":"绝不上传音频的 Fathom 替代方案","description":"私密、开源的语音转文字和设备端会议笔记。在 macOS、Windows 和 Linux 上免费使用。","viewOnGitHub":"在 GitHub 上查看"}},"featureDictationSoftware":{"metadata":{"title":"面向专业人士的开源语音输入软件 - OpenWhispr","description":"OpenWhispr 是免费的开源语音输入软件，可在任何应用中将语音转换为文字。支持离线使用和本地运行，让语音数据留在你的设备上。","keywords":["听写软件","语音听写软件","免费听写软件","语音转文字软件","语音输入","听写应用","离线听写","本地听写软件"]},"breadcrumb":{"home":"首页","dictationSoftware":"听写软件"},"structuredData":{"description":"开源听写软件。在设备上私密运行，或使用 OpenWhispr Cloud 获得最高准确度，在任何应用中将语音转成文字，支持 100+ 种语言。","offerDescription":"免费开始。本地听写不限量，可选 Pro 和 Business 方案。","featureList":["在任何应用中使用全系统按键听写","本地离线语音转文字，以及高准确度的 OpenWhispr Cloud","100+ 种语言，内置翻译","OpenAI Whisper 和 NVIDIA Parakeet 模型","用于姓名、术语和缩写的自定义词典","内置 AI 写作助手，可整理和润色听写内容","在 macOS、Windows 和 Linux 上开源"]},"hero":{"badge":"免费开始 · 开源","title":"适用于所有应用的免费听写软件","description":"按下快捷键并说话，带标点的清晰文字就会出现在邮件、文档、Slack、编辑器或 AI 聊天中。可在设备上私密运行，也可启用 OpenWhispr Cloud 获得最高准确度。无论哪种方式，音频都不会被存储。","github":"在 GitHub 上查看","platforms":"在 macOS、Windows 和 Linux 上免费开始 · 无需银行卡或账户"},"trustBar":{"title":"在你常用的每个应用中听写"},"painPoints":{"eyebrow":"用户为何切换","title":"大多数听写软件都会让你有所取舍。","description":"用速度换隐私、用隐私换准确度，或让整个工作流受制于一家公司的云。OpenWhispr 让你无需选择。","items":[{"title":"你的想法比手指更快。","body":"说话速度远快于打字。OpenWhispr 以思考的速度把语音变成清晰文字，让邮件、提示词和草稿不再受键盘限制。"},{"title":"你看不到多数语音工具如何处理你的声音。","body":"多数听写软件是闭源且仅限云端。OpenWhispr 开源；无论本地还是 OpenWhispr Cloud，音频只会被转写，不会被存储、出售或用于训练，代码可查。"},{"title":"你的配置不该被某个应用或操作系统锁定。","body":"封闭工具会把工作流锁在其平台。OpenWhispr 可在 Mac、Windows 和 Linux 的所有应用中运行，并支持你自己的 AI 密钥。"},{"title":"系统自带听写还不够。","body":"Apple Dictation 和 Windows Speech 准确度不足、依赖网络且绑定单一系统。OpenWhispr 在三个平台提供 Whisper 级识别和自定义词典。"}]},"features":{"eyebrow":"你将获得","title":"一个快捷键，连接你的声音与所有应用。","description":"光标在哪里，就在哪里听写：写邮件、向 AI 发出提示词、记录想法，无需复制粘贴。","items":[{"title":"在任何应用中听写","body":"按下快捷键、说话、松开。带标点的文字会出现在邮件、文档、Slack 或代码编辑器的光标处。"},{"title":"与 AI 对话，不必打字","body":"在 ChatGPT、Claude 或 Cursor 中口述包含丰富上下文的长提示词，以最快方式提供 AI 所需细节。"},{"title":"准确识别你的用词","body":"通过自定义词典教会 OpenWhispr 识别姓名、术语和缩写，然后选择本地引擎或 OpenWhispr Cloud。"}]},"stats":{"items":[{"value":"100+","label":"听写语言","foot":"并内置翻译"},{"value":"0","label":"存储的录音","foot":"本地音频不离开设备；云端音频转写后即丢弃"},{"value":"5分钟","label":"设置时间","foot":"从下载到说出第一句话"}]},"howItWorks":{"eyebrow":"使用方法","title":"三步即可与电脑对话。","description":"无需云账户，无需注册。软件安装在你的电脑上，约五分钟即可运行。","items":[{"title":"免费下载。","body":"支持 Mac、Windows 或 Linux。选择本地模型或 OpenWhispr Cloud，无需账户或银行卡。"},{"title":"设置快捷键。","body":"一个快捷键，可按住录音或按键切换，并在全系统所有应用中生效。"},{"title":"按下。说话。完成。","body":"支持 100+ 种语言，清晰且带标点的文字会出现在光标处。"}]},"reasons":{"eyebrow":"事实无需说服","title":"选择 OpenWhispr 听写的 10 个理由","description":"切换工具需要信任。这里的每项说明都可在开源仓库中验证。","items":[{"title":"适用于所有应用","body":"全系统快捷键。只要能放置光标，就能听写。"},{"title":"隐私优先且可审计","body":"开源且不保留数据：本地音频留在设备，云端音频转写后丢弃。"},{"title":"本地或云端，由你决定","body":"在设备上运行以获得完整隐私，或用 OpenWhispr Cloud 获得准确度和速度。"},{"title":"Whisper 和 Parakeet 级准确度","body":"领先的开源语音转文字模型，并提供适合普通硬件的云端选项。"},{"title":"属于你的自定义词典","body":"添加姓名、术语和缩写，让转写第一次就正确。"},{"title":"100+ 种语言和翻译","body":"用一种语言说话，以另一种语言输出，可指定或自动检测。"},{"title":"使用自己的 AI 密钥","body":"使用 OpenAI、Anthropic、Gemini 或 Groq 密钥进行无加价的云端转写。"},{"title":"边说边整理","body":"内置 AI 写作助手可在本地或云端整理和润色听写内容。"},{"title":"Mac、Windows 和 Linux","body":"少数可在三个平台原生运行且功能一致的听写应用之一。"},{"title":"减轻双手负担","body":"如果打字缓慢、疲劳或疼痛 — RSI、疲劳、长时间工作 — 就让声音代劳。"}]},"faq":{"eyebrow":"下载之前","title":"关于 OpenWhispr 听写软件的常见问题","items":[{"question":"OpenWhispr 能在所有应用中使用吗？","answer":"可以。快捷键在全系统生效，包括邮件、文档、Slack、浏览器、代码编辑器和 AI 聊天。能打字的地方就能听写。"},{"question":"语音转文字有多准确，包括不同口音吗？","answer":"它使用 Whisper 级模型和 NVIDIA Parakeet，针对多种口音和 100+ 种语言训练。可选择更大的本地模型、添加术语或启用 OpenWhispr Cloud。"},{"question":"本地听写与云端听写有何不同？","answer":"本地模式在设备上运行，完全私密、离线且不限量。OpenWhispr Cloud 无需下载模型，提供高准确度和速度，音频转写后不存储。免费版每周 2,000 个云端单词，Pro 不限量。"},{"question":"它免费吗？哪些功能收费？","answer":"应用免费且开源：100+ 种语言的本地听写不限量，含词典、每周 2,000 个 OpenWhispr Cloud 单词和 5 小时会议录音。Pro ($8/月) 增加无限云端转写、20 小时、移动端和同步。"},{"question":"我的声音会离开设备吗？","answer":"本地模式绝不会，开源代码可以证明。使用 OpenWhispr Cloud 或 API 密钥时，音频仅为转写而发送，随后丢弃。我们不会存储、出售数据或用其训练。"},{"question":"能在 Mac、Windows 和 PC 上使用吗？","answer":"可以。OpenWhispr 是少数原生支持 macOS、Windows 和 Linux 的开源听写应用之一，三个平台使用相同引擎和功能。"},{"question":"它能学习我的词汇和术语吗？","answer":"可以。自定义词典让 OpenWhispr 学会姓名、技术术语和缩写，首次即可正确转写；所有本地或云端方案均包含。"},{"question":"这是 AI 听写吗？能格式化和清理文字吗？","answer":"可以。内置 AI 写作助手可用本地或云端模型添加标点、重新格式化和改写；智能体模式还能自动处理转写内容。"},{"question":"它与 Dragon 或系统自带听写有何区别？","answer":"Dragon 功能强但昂贵且以 Windows 为主；Apple 和 Windows 听写依赖网络和单一系统。OpenWhispr 支持三个平台，开源，并可选择本地或云端。"},{"question":"运行需要什么？支持离线吗？","answer":"一台较新的 Mac、Windows 或 Linux 电脑即可。小型本地模型可在普通硬件上运行，安装后完全离线。OpenWhispr Cloud 无需下载模型。"}]},"testimonials":{"eyebrow":"用户评价","title":"告别打字的人们。"},"pricing":{"eyebrow":"价格","title":"免费开始。升级即可获得云端速度和更多功能。","description":"本地听写永久免费且不限量。Pro 和 Business 增加无限 OpenWhispr Cloud、会议时长、移动端和团队功能。"},"openSource":{"eyebrow":"公开构建","title":"以你希望的方式处理你的声音。","description":"OpenWhispr 源于一个简单质疑：多数听写软件在闭源代码背后收听一切。因此我们公开构建，每个版本和每行代码都在 GitHub 上，已有 5,000+ 名工程师点赞、复刻和审查。无论本地还是云端，我们都不存储、出售或用你的数据训练，代码可以证明。","github":"在 GitHub 上点赞"},"finalCta":{"title":"你的下一句话不需要键盘。","description":"下载 OpenWhispr，在电脑上的任何应用中听写 — 支持 100+ 种语言，隐私优先，五分钟即可开始。","footer":"在 Mac、Windows 和 Linux 上免费开始 · 无需银行卡或账户"},"demo":{"dictationText":"团队好，我刚刚审阅完 Q2 路线图，我认为应该在新集成之前优先处理语音管线。","without":"不使用 OpenWhispr","typingSpeed":"~40 WPM","windowTitle":"Slack — #product","messagePlaceholder":"输入消息…","with":"使用 OpenWhispr","dictationSpeed":"~150 WPM","faster":"快 3x","transcribing":"正在转写…","listening":"正在聆听…","hotkeyPrompt":"按下快捷键并说话…","indicator":"按住快捷键，说话后松开 — 文字会出现在当前应用中。"}},"compareFireflies":{"metadata":{"title":"Fireflies.ai 替代方案 — 私密、本地、开源","description":"私密、开源的 Fireflies.ai 替代方案。语音转文字和会议笔记 100% 在本地运行，无机器人加入通话，也不上传音频。","keywords":["fireflies 替代方案","fireflies.ai 替代方案","私密 fireflies 替代方案","开源 fireflies 替代方案","本地会议笔记","无机器人会议笔记"]},"breadcrumb":{"home":"首页","comparisons":"对比","fireflies":"对比 Fireflies.ai"},"hero":{"title":"私密、开源的 Fireflies.ai 替代方案","description":"OpenWhispr 是为希望在自己的设备上 100% 运行<voice>语音转文字</voice>和<meetings>会议笔记</meetings>的用户打造的 Fireflies.ai 替代方案。没有机器人加入通话，音频绝不离开你的电脑，整个应用均为开源。","viewOnGitHub":"在 GitHub 上查看"},"quickTake":{"title":"快速了解","firefliesIntro":"是一款云端会议助手。它的机器人会加入 Zoom、Teams 和 Meet 通话，录音并转写内容，生成摘要和待办事项，并为销售和营收团队提供对话智能与 CRM 集成。它确实很擅长这项工作。","openWhisprIntro":"围绕另一项优先需求打造：隐私和所有权。它是一款语音转文字和设备端会议笔记工具，使用 OpenAI Whisper 和 NVIDIA Parakeet，完全在你的电脑上运行。没有机器人加入，无需账户，也不会上传任何内容。按下快捷键，即可在任何应用中输入你说的话；也可以在本地录制会议音频并在设备上转写。","chooseFireflies":"如果你需要可自动加入通话并提供团队分析和 CRM 同步的机器人，请选择 Fireflies.ai。","chooseOpenWhispr":"如果你需要私密、本地且绝不上传音频的语音转文字和会议笔记，请选择 OpenWhispr。"},"comparisonTable":{"title":"逐项功能对比","headerFeature":"功能"},"comparisonData":[{"feature":"隐私与数据控制","openWhispr":{"status":"full","details":"100% 本地 — 会议音频保留在你的设备上"},"competitor":{"status":"none","details":"基于云端 — 音频和转写稿存储在服务器上"}},{"feature":"开源","openWhispr":{"status":"full","details":"采用 MIT 许可证，可在 GitHub 上审查"},"competitor":{"status":"none","details":"专有闭源软件"}},{"feature":"离线工作","openWhispr":{"status":"full","details":"本地模型无需联网即可运行"},"competitor":{"status":"none","details":"需要互联网连接"}},{"feature":"会议采集","openWhispr":{"status":"partial","details":"在本地录制系统音频 — 无机器人加入"},"competitor":{"status":"full","details":"机器人自动加入 Zoom、Teams 和 Meet 通话"}},{"feature":"通用听写","openWhispr":{"status":"full","details":"在任意应用中进行系统级语音输入"},"competitor":{"status":"none","details":"专注于会议，并非听写工具"}},{"feature":"转写与语言","openWhispr":{"status":"full","details":"Whisper + Parakeet，100+ 种语言"},"competitor":{"status":"full","details":"支持 AI 搜索的云端转写"}},{"feature":"CRM 与团队智能","openWhispr":{"status":"partial","details":"API/MCP 访问；Business 提供基础团队功能"},"competitor":{"status":"full","details":"Salesforce/HubSpot 同步与对话分析"}},{"feature":"免费方案","openWhispr":{"status":"full","details":"无限本地听写，永久免费"},"competitor":{"status":"partial","details":"免费方案有限制 — 请查看其网站"}}],"whyOpenWhispr":{"title":"为什么选择 OpenWhispr？","items":[{"title":"默认本地运行：","body":"转写在你的电脑上运行。参与者列表中没有机器人，不上传音频，也不会在第三方服务器上存储任何内容。"},{"title":"随处听写：","body":"一个系统级快捷键即可将语音输入任何应用，包括电子邮件、文档、编辑器或 AI 聊天。Fireflies 仅用于会议。"},{"title":"开源：","body":"采用 MIT 许可证且可供审查，你可以准确验证音频的处理方式。"}]},"whyFireflies":{"title":"为什么选择 Fireflies.ai？","items":[{"title":"自动加入通话：","body":"机器人会自行加入预定通话并自动采集，无需你动手操作。"},{"title":"CRM 与营收智能：","body":"为销售团队提供深入的 Salesforce/HubSpot 同步和对话分析。"},{"title":"共享团队工作区：","body":"开箱即用，在整个组织中提供可搜索、可协作的转写稿。"}]},"faqs":[{"question":"最佳 Fireflies.ai 替代方案是什么？","answer":"这取决于你的需求。如果你想要一款私密、开源的工具，用于日常语音转文字和设备端会议笔记，OpenWhispr 是很好的选择：它使用 OpenAI Whisper 和 NVIDIA Parakeet 在本地 100% 运行语音识别，通过一键说话快捷键在整个系统中工作，并且绝不会将音频发送到云端。如果你特别需要能自动加入通话并提供团队分析和 CRM 集成的机器人，Fireflies.ai 本身仍然非常出色。"},{"question":"有免费的 Fireflies.ai 替代方案吗？","answer":"有。OpenWhispr 免费且开源。在 macOS、Windows 和 Linux 上，你可以免费离线使用支持 100+ 种语言的无限本地听写。可选的 Pro 方案（$8/月）增加无限云端转写、更多会议录制时长、移动端和设备同步，但使用核心的本地语音转文字功能完全不需要它。"},{"question":"有本地或私密的 Fireflies.ai 替代方案吗？","answer":"OpenWhispr 正是为此打造。核心转写在你自己的电脑上进行，音频绝不离开设备，并且完全支持离线工作。由于应用开源，你可以准确审查数据的处理方式。通话中没有机器人，也不会上传任何内容，这对于人力资源、法律或涉及客户数据的敏感会议尤其重要。"},{"question":"OpenWhispr 会像 Fireflies 一样通过机器人加入会议吗？","answer":"不会，这正是关键所在。Fireflies 会派出一个出现在通话中的机器人，并将音频上传至其云端。OpenWhispr 则直接在你的设备上采集会议的系统音频，因此参与者列表中没有机器人，也不会上传任何内容。你仍需告知他人正在录音，但对话会完全保留在你的电脑上。"},{"question":"OpenWhispr 与 Fireflies.ai 有何不同？","answer":"Fireflies.ai 是一款云端会议助手，专注于为销售和营收团队自动加入通话、提供对话智能和 CRM 同步。OpenWhispr 则是一款隐私优先的语音转文字工具，并提供设备端会议笔记：按下快捷键即可在任意应用中说话，或在本地录制并转写会议。两者解决不同的问题：许多人会保留机器人工具用于协作式销售通话，同时使用 OpenWhispr 处理私密的日常听写和敏感会议。"}],"faq":{"title":"常见问题"},"cta":{"title":"绝不上传音频的 Fireflies.ai 替代方案","description":"私密、开源的语音转文字和设备端会议笔记。可在 macOS、Windows 和 Linux 上免费使用。","viewOnGitHub":"在 GitHub 上查看"}},"unsubscribe":{"metadata":{"title":"取消订阅 — OpenWhispr","description":"管理您的 OpenWhispr 营销邮件订阅。","keywords":["取消订阅 OpenWhispr","电子邮件偏好设置","营销邮件"]},"breadcrumb":{"home":"首页","unsubscribe":"取消订阅"},"invalid":{"title":"此链接无效","description":"取消订阅链接缺失或已损坏。请尝试使用较新邮件中的链接，或直接联系我们将您从邮件列表中移除。","contact":"请发送邮件至 <email>support@openwhispr.com</email>，我们会为您处理。"},"success":{"title":"您已取消订阅","description":"<email>{emailAddress}</email> 将不再收到 OpenWhispr 的营销邮件。此更改已立即生效。","serviceMessages":"您仍会收到收据、安全警报和密码重置等服务邮件，因为这些邮件与您的账户相关。"},"confirm":{"title":"取消订阅营销邮件？","description":"<email>{emailAddress}</email> 将不再收到 OpenWhispr 的产品动态和优惠。与您账户相关的服务邮件不受影响。","button":"取消订阅"}},"legalDpa":{"metadata":{"title":"数据处理附录 | OpenWhispr","description":"OpenWhispr数据处理附录（Gizmo Labs Inc.作为数据处理者），无需签署即生效，包含不使用数据训练AI的承诺、SCC及UK Addendum传输条款和产品次级处理者附件。"},"breadcrumb":{"home":"首页","dpa":"数据处理附录"},"authoritativeNotice":"本译文仅为方便阅读而提供。如与英文版本存在任何不一致，应以英文版本为准。","header":{"title":"数据处理附录","subtitle":"Gizmo Labs Inc.（OpenWhispr）如何代表其客户处理个人数据。客户内容绝不会用于训练AI模型（第3条）。","effective":"生效日期：2026年7月17日 · 版本1.2（2026年8月21日更新）"},"preamble":{"title":"关于本附录","body":"本数据处理附录（\"DPA\"）构成OpenWhispr服务条款的一部分并通过引用纳入其中；如客户（\"客户\"、\"数据控制者\"）与Gizmo Labs Inc.之间另有主服务协议或企业协议（\"协议\"），则构成该协议的一部分并通过引用纳入其中。凡Gizmo Labs Inc.（一家在Delaware注册成立、注册地址为221 W 9th Street, Wilmington, Delaware 19801的公司，称\"OpenWhispr\"、\"数据处理者\"）在提供OpenWhispr服务（\"服务\"）过程中代表客户处理个人数据，本DPA即适用。本DPA无需签署即生效；如客户采购流程要求已签署文本，可使用所附签署栏。如本DPA与协议在数据保护事项上冲突，以本DPA为准；如本DPA与SCC（或UK Addendum）就受限传输发生冲突，在冲突范围内以SCC（或UK Addendum）为准。本DPA项下通知应发送至双方的账户联系人，发送给OpenWhispr的通知还应发送至support@openwhispr.com。"},"definitions":{"title":"1. 定义","body":"本文件未定义的首字母大写术语具有数据保护法律赋予的含义。\"数据保护法律\"指适用于协议项下个人数据处理的所有法律，包括欧盟GDPR（条例(EU) 2016/679）、英国GDPR及Data Protection Act 2018、瑞士FADP和适用的美国州隐私法（包括CCPA/CPRA）。\"SCC\"指欧盟委员会实施决定(EU) 2021/914中的Standard Contractual Clauses。\"UK Addendum\"指ICO的International Data Transfer Addendum to the EU SCCs。\"受限传输\"指向未获充分性决定的第三国传输个人数据。\"次级处理者\"指OpenWhispr为代表数据控制者处理个人数据而聘用的第三方。"},"scope":{"title":"2. 范围和角色","body":"OpenWhispr仅作为数据处理者代表客户处理个人数据。处理的事项、期限、性质和目的、个人数据类型及数据主体类别载于附件1。为免疑义：当最终用户仅在本地运行OpenWhispr或使用自己的模型API密钥（\"bring your own key\"）时，音频和转录/推理内容从用户设备直接发送至用户选择的提供商，或在设备上处理；这些内容流由用户指示，所涉第三方提供商依用户指示行事，而非作为OpenWhispr的次级处理者。无论采用何种模式，账户数据、云同步（如启用）、使用和账单数据仍由OpenWhispr及其次级处理者按照本DPA处理。OpenWhispr还可为运营、保护和改进服务而处理经汇总及去标识化、已不再构成个人数据的数据（例如服务性能指标），且不会尝试重新识别。"},"noAiTraining":{"title":"3. 不使用客户数据训练AI","body":"OpenWhispr不会使用客户个人数据（包括语音音频、转录文本、笔记、提示词及与AI代理的对话）训练、重新训练或改进任何人工智能或机器学习模型，无论该模型属于OpenWhispr还是第三方，并且不允许其次级处理者这样做。附件3所列模型提供商仅为提供服务而处理此类内容，OpenWhispr账户的训练功能已禁用或选择退出。本限制在协议终止后继续有效。"},"processorObligations":{"title":"4. 数据处理者义务","intro":"OpenWhispr将：","items":["仅依客户的书面指示（协议、本DPA及客户对服务的配置构成此类指示）处理个人数据，包括进行传输，但OpenWhispr所受法律要求处理的除外；在该情况下，除法律禁止外，OpenWhispr将在处理前将该法律要求告知客户；","确保获准处理个人数据的人员受保密义务约束；","实施附件2所列技术和组织安全措施，并定期审查；","考虑处理性质，就数据主体请求、安全、泄露通知、数据保护影响评估和事先咨询向客户提供协助（第6–9条）；","在服务终止时根据客户选择删除或返还全部个人数据，并在30日内删除现有副本，但OpenWhispr所受法律要求保留的除外；","提供合理必要的信息以证明合规，并按照第10条允许并协助审计；以及","如OpenWhispr认为某项指示违反数据保护法律，立即通知客户。"]},"subprocessing":{"title":"5. 次级处理","body":"客户一般性书面授权OpenWhispr聘用次级处理者。现行名单载于附件3，并发布于OpenWhispr信任中心<trustLink>trust.openwhispr.com</trustLink>（Subprocessors部分）。OpenWhispr向每位次级处理者施加与本DPA同等的数据保护义务（包括传输义务），并继续对其履行负责。OpenWhispr至少提前30日向客户书面通知拟新增或替换的次级处理者（向客户账户所有者发送电子邮件并更新公开名单）；客户可在此期间基于合理且有书面依据的数据保护理由提出异议。如双方不能解决异议，客户可终止受影响的服务，并获得已预付费用中与所终止服务相应的按比例退款。"},"security":{"title":"6. 安全","body":"OpenWhispr维持附件2中的措施，并在考虑技术发展水平和个人数据性质后确保其与风险相适应。通过云端转录路径处理的语音音频会被实时处理，转录后不会由OpenWhispr保存。"},"breach":{"title":"7. 个人数据泄露","body":"OpenWhispr在获悉影响客户个人数据的个人数据泄露后，将不无故拖延地、并在任何情况下于72小时内通知客户，提供客户为履行自身通知义务而合理需要的信息，并配合补救。"},"dataSubjectRequests":{"title":"8. 数据主体请求","body":"考虑处理性质，OpenWhispr将通过适当的技术和组织措施协助客户回应数据主体行使权利（访问、更正、删除、限制、可携带、反对）的请求。最终用户可在产品内自行导出和删除；OpenWhispr直接收到且能识别客户组织的请求将不无故拖延地转交客户。"},"dpias":{"title":"9. 数据保护影响评估和事先咨询","body":"在OpenWhispr掌握所需信息的范围内，OpenWhispr就数据保护影响评估及与监管机构的事先咨询提供合理协助。"},"audit":{"title":"10. 审计","body":"应请求，OpenWhispr将在保密前提下通过信任中心提供现行合规文件（SOC 2 Type 2报告，涵盖2026年5月1日至2026年7月31日期间，可依NDA提供；通过其合规平台出具的GDPR和HIPAA证明，2026；ISO/IEC 27001:2022认证，证书编号KCISM202607012）。如数据保护法律赋予客户强制性审计权且上述报告不能满足该权利，OpenWhispr允许客户或独立审计员在任何12个月内进行不超过一次审计，并在发生经确认、影响客户个人数据的泄露后额外进行审计；每次均须至少提前30日通知，并受保密义务和OpenWhispr安全政策约束。客户自行承担审计费用。对于受限传输，本条不影响SCC项下的审计权。"},"customerObligations":{"title":"11. 客户义务和保证","body":"客户保证，其指示及向OpenWhispr提供个人数据符合数据保护法律，包括具有合法依据、向数据主体提供必要通知，并且在最终用户可能口述或处理特殊类别个人数据（例如健康信息）时，确保具备适当法律依据并签订任何必要协议（HIPAA见第14条）。"},"transfers":{"title":"12. 国际传输","intro":"本条适用于从EEA、英国或瑞士向这些地区以外国家传输个人数据，包括任何受限传输。OpenWhispr及其次级处理者主要在美国处理个人数据（各次级处理者地点见附件3）。每项传输应采用以下第一个适用机制：","mechanisms":["充分性。 如目的地享有充分性决定，传输以该决定为依据。","Standard Contractual Clauses。 对于来自EEA的任何受限传输，双方签订SCC模块二（数据控制者 → 数据处理者），客户为数据输出方，OpenWhispr为数据输入方；该SCC通过引用纳入本DPA，并按以下方式填写：不采用第7条（加入）；第9条选项2（一般授权，提前30日通知）；不采用第11条可选救济机构；第13条/附件I.C的主管监管机构：Irish Data Protection Commission；第17条准据法：爱尔兰法律；第18条法院：爱尔兰法院；本DPA附件1–3构成SCC附件I–III。向次级处理者的后续传输适用模块二SCC第8.8及9(b)条（传递相同义务）。","英国和瑞士传输。 对于英国个人数据，双方签订并以附件信息填写UK Addendum to the EU SCCs；对于瑞士个人数据，SCC按FADP修订后适用。"],"futureFramework":"如OpenWhispr将来依据EU–US Data Privacy Framework完成自我认证，传输还可依赖该认证。","supplementaryMeasures":"如传输影响评估确定需要补充措施，双方将予以实施，OpenWhispr将支持客户的评估。OpenWhispr同意SCC规定的义务，包括在无法遵守时通知客户。"},"governmentRequests":{"title":"13. 政府访问请求","body":"如OpenWhispr收到公共机关就客户个人数据提出的具有法律约束力的请求，将在法律允许范围内：及时通知客户；审查请求合法性并在有合理依据时提出质疑；仅披露法律要求的最少数据。OpenWhispr保存此类请求记录供客户审查。"},"hipaa":{"title":"14. HIPAA","body":"本DPA并非Business Associate Agreement。如客户属于HIPAA规定的Covered Entity或Business Associate，除非双方另行签订Business Associate Agreement，否则最终用户不得通过云服务处理受保护的健康信息（protected health information，PHI）。OpenWhispr可应发送至support@openwhispr.com的请求，向Business及以上套餐的客户提供BAA。签订后，该BAA在冲突范围内取代本DPA并规管PHI。"},"usStatePrivacy":{"title":"15. 美国州隐私法","body":"在CCPA/CPRA或其他美国州隐私法适用的范围内，OpenWhispr作为\"service provider\"/\"processor\"行事：不出售或共享个人数据；除提供服务或法律允许外，不保留、使用或披露个人数据；并确认其理解且将遵守这些限制。"},"liability":{"title":"16. 责任和期限","body":"数据保护事项的责任按协议规定并受其中任何上限约束。只要OpenWhispr仍代表客户处理个人数据，本DPA即持续有效。"},"signatures":{"title":"签署（可选——供需要已签署文本的客户使用）","nameRole":"姓名和职务","signature":"签名","date":"日期","customer":"代表客户（数据控制者）","processor":"代表Gizmo Labs Inc.（数据处理者）","executive":"Gabriel Stein, CEO"},"annex1":{"title":"附件1 — 处理详情","parties":"A. 双方（SCC附件I.A）。 数据输出方：客户（数据控制者）——名称、地址和联系人以协议或订单为准。数据输入方：Gizmo Labs Inc.（数据处理者），221 W 9th Street, Wilmington, Delaware 19801, USA，联系人：support@openwhispr.com；活动：提供下述服务。","descriptionTitle":"B. 处理说明（SCC附件I.B）。","field":"字段","detail":"详情","rows":[{"field":"事项","detail":"提供OpenWhispr语音听写和AI笔记服务（桌面应用程序及API）"},{"field":"期限","detail":"协议期限，加上第4(e)条规定的终止后删除/返还期"},{"field":"性质和目的","detail":"实时语音转文字；AI辅助文本生成、格式化和推理；笔记存储、搜索和共享；账户管理；账单；交易邮件"},{"field":"个人数据类别","detail":"账户和身份数据（姓名、电子邮件、认证标识符）；语音音频（在云路径中实时处理，OpenWhispr不保存）；转录文本和笔记（可能含最终用户口述的任何个人数据）；AI功能对话历史；使用和设备元数据；账单数据（由支付次级处理者处理）"},{"field":"特殊类别数据","detail":"可能偶然包含：口述内容是用户主导的自由文本，可能含健康或其他特殊类别数据。保障措施：传输中及静态加密、不保存音频、仅本地及BYOK模式、PHI须有BAA（第14条）"},{"field":"数据主体类别","detail":"客户的最终用户（雇员、承包商）；口述内容或笔记中提及的人员"},{"field":"频率","detail":"在服务期限内持续进行"}]},"annex2":{"title":"附件2 — 技术和组织措施","intro":"OpenWhispr安全计划概要（当前详情：OpenWhispr信任中心 — <trustLink>trust.openwhispr.com</trustLink>，Security Controls部分）：","bullets":["个人数据在传输中（TLS 1.2+）及静态时加密。","云端语音音频实时处理，转录后不由OpenWhispr保存（转录/推理次级处理者依其协议临时处理音频）；仅本地及bring-your-own-key模式完全绕过OpenWhispr云。","按最小权限原则进行访问控制，生产系统启用MFA及日志记录/监控；每月审查会话和登录监控。","漏洞和补丁管理；安全开发实践。","具备韧性和备份并记录恢复目标（主要数据存储RTO ≤ 4h / RPO ≤ 1h），每年测试应急计划。","人员保密义务和安全培训。","次级处理者治理：合同义务传递（第5条）、公开名单，与主要内容存储次级处理者（数据库；云端转录/LLM）签订BAA；其余HIPAA路径次级处理者的覆盖正在推进。","独立审查：SOC 2 Type 2报告已就2026年5月1日至2026年7月31日期间出具，涵盖安全性、保密性与可用性，可依NDA提供。合规证明：GDPR和HIPAA（2026，通过OpenWhispr合规平台）。信息安全管理体系已通过ISO/IEC 27001:2022认证（证书编号KCISM202607012，由KVQA Certification Services颁发，经加拿大标准委员会认可；初次注册2026年7月27日，有效期至2027年7月27日）。"]},"annex3":{"title":"附件3 — 次级处理者","intro":"截至2026-07-17获授权的次级处理者（实时名单：OpenWhispr信任中心，<trustLink>trust.openwhispr.com</trustLink>，Subprocessors部分）。法人实体名称及处理地点已于2026-07-17根据各供应商条款/DPA核实。\"US (primary)\"指供应商条款将美国列为主要处理地点，同时允许其自身次级处理者在其他国家处理。\"第9(b)条义务传递\"指在受限传输适用时，依据模块二SCC第8.8及9(b)条对次级处理者施加相同义务的合同。","headers":{"subprocessor":"次级处理者","service":"服务","dataCategories":"数据类别","country":"国家","safeguard":"后续传输保障"},"rows":[{"subprocessor":"Neon, LLC (an affiliate of Databricks, Inc.)","service":"主要数据库和认证存储","dataCategories":"账户数据、笔记、转录文本、对话历史","country":"US（AWS US区域，项目创建时固定）","safeguard":"第9(b)条义务传递"},{"subprocessor":"Vercel Inc.","service":"应用托管和交付","dataCategories":"通过服务传输的所有数据","country":"US（主要）","safeguard":"第9(b)条义务传递"},{"subprocessor":"OpenAI OpCo, LLC","service":"云端语音转文字、语言模型、嵌入（已退出训练）","dataCategories":"语音音频（临时）、提示词、文本","country":"US（主要）","safeguard":"第9(b)条义务传递"},{"subprocessor":"Parasail, Inc.","service":"云端AI推理（默认推理；零数据保留推理）","dataCategories":"提示词、文本（传输中）","country":"US（主要）","safeguard":"第9(b)条义务传递"},{"subprocessor":"Groq, Inc.","service":"云端AI推理和转录故障转移","dataCategories":"语音音频（临时）、提示词、文本（传输中）","country":"US（主要）","safeguard":"第9(b)条义务传递"},{"subprocessor":"Embedchain, Inc. d/b/a Mem0","service":"代理功能的AI记忆","dataCategories":"与用户ID关联的对话内容","country":"US","safeguard":"第9(b)条义务传递"},{"subprocessor":"Exa Labs, Inc.","service":"代理功能的网页搜索","dataCategories":"搜索查询文本","country":"US (AWS)","safeguard":"第9(b)条义务传递"},{"subprocessor":"Stripe, LLC","service":"支付处理","dataCategories":"账单和支付数据","country":"US（主要）","safeguard":"第9(b)条义务传递"},{"subprocessor":"Plus Five Five, Inc. (d/b/a Resend)","service":"交易邮件","dataCategories":"电子邮件地址、消息内容（包括共享笔记标题）","country":"US（AWS us-east-1）","safeguard":"第9(b)条义务传递"},{"subprocessor":"Slack Technologies, LLC (Salesforce)","service":"内部运营和支持通信","dataCategories":"支持相关数据（政策：不含PHI）","country":"US（默认数据驻留地）","safeguard":"第9(b)条义务传递"}],"footnote":"本附件列出在产品服务中处理客户个人数据的次级处理者。Google（OAuth登录及可选Calendar集成）不是次级处理者：登录由Google作为独立数据控制者提供，日历数据仅缓存在用户设备上。OpenWhispr更广泛的供应商透明度名单（包括RB2B、lemlist和GitHub等营销网站及运营供应商；对于这些供应商，Gizmo Labs作为数据控制者，或其不接收客户个人数据）发布于OpenWhispr信任中心。"}},"blogBestNoteTakers":{"metadata":{"title":"2026 年最佳 AI 会议笔记工具（免费且注重隐私）| OpenWhispr","description":"对比 2026 年最佳 AI 会议笔记工具：Fathom、Otter、Granola、Fireflies，以及免费、私密、开源并在本地运行的 OpenWhispr。","keywords":["最佳 ai 会议笔记工具","ai 会议笔记工具推荐","最佳免费 ai 会议笔记","开源 ai 会议笔记工具","注重隐私的 ai 会议笔记","ai 会议笔记工具对比","私密 ai 会议笔记工具","最佳 ai 会议记录"]},"breadcrumb":{"home":"首页","blog":"博客","article":"最佳 AI 会议笔记工具"},"schema":{"title":"2026 年最佳 AI 会议笔记工具（含免费和开源选择）","description":"对比 2026 年最佳 AI 会议笔记工具：Fathom、Otter、Granola、Fireflies，以及免费、私密、开源并在本地运行的 OpenWhispr。"},"faqs":[{"question":"哪款 AI 会议笔记工具最好？","answer":"没有一款工具适合所有人，关键取决于你的需求。想以最少设置自动生成会议摘要，Fathom 和 Otter 都很出色。需要 CRM 同步和对话分析的销售团队可以选择 Fireflies。如果最看重隐私，希望音频留在自己的设备上，OpenWhispr 更为突出：它在本地转写、代码开源，而且不会向通话中派出机器人。"},{"question":"哪款免费 AI 会议笔记工具最好？","answer":"Fathom 为个人用户提供了云端产品中相当慷慨的免费方案。如果还希望免费工具能保护数据隐私，OpenWhispr 免费且开源，并提供不限量的本地转写：没有每月上限、无需银行卡，也不会上传任何内容。"},{"question":"有开源的 AI 会议笔记工具吗？","answer":"有。OpenWhispr 采用 MIT 许可证，完全开源。语音转文字通过 OpenAI Whisper 和 NVIDIA Parakeet 在你的设备上运行，因此你可以准确审查音频的处理方式。Meetily 也是一款值得了解的开源、本地优先产品。"},{"question":"哪款 AI 会议笔记工具最能保护隐私？","answer":"最私密的方案是在设备上转写，而且绝不上传音频。OpenWhispr 默认就是如此：没有机器人加入通话，录音留在你的电脑上。因此，它很适合人力资源、法律事务或涉及客户数据等敏感会议。相比之下，大多数云端笔记工具会把会议上传并存储在自己的服务器上。"},{"question":"AI 会议笔记工具必须派机器人加入会议吗？","answer":"大多数需要。Otter、Fireflies 和 Fathom 会派一个机器人加入，并显示在参会者列表中。也有少数例外：Granola 无需机器人即可采集电脑音频；OpenWhispr 也是如此，并且整个转写过程都留在设备上。"},{"question":"哪款 AI 会议笔记工具可以离线使用？","answer":"Otter、Fireflies 和 Fathom 等云端工具需要联网，因为转写在它们的服务器上完成。OpenWhispr 可以完全离线运行：本地模型无需网络即可在你的电脑上转写，适合乘飞机、进入安全设施，或任何不希望音频离开设备的场景。"}],"sections":[{"id":"how-to-choose","title":"如何选择"},{"id":"comparison","title":"快速对比"},{"id":"picks","title":"推荐工具"},{"id":"faq","title":"常见问题"}],"hero":{"title":"2026 年最佳 AI 会议笔记工具（含免费和开源选择）","subtitle":"自动生成云端会议摘要，还是在设备上保留私密笔记？这份公平、坦诚的对比会说明各款主流 AI 会议笔记工具真正擅长什么，又有哪些不足。","date":"2026 年 7 月 6 日","tag":"对比"},"opening":{"lead":"没有唯一一款最好的 AI 会议笔记工具；正确选择取决于你更重视自动化、价格还是隐私。","body":"想全自动获得云端摘要，Fathom 和 Otter 很出色；销售团队可以考虑 Fireflies；希望音频留在自己设备上，则可选择 OpenWhispr。下面会坦诚分析每款工具，包括其他产品表现更好的地方。","updated":"最后更新：2026 年 7 月 6 日。OpenWhispr 也在这份榜单中；我们尽力如实说明其他工具胜出的方面。"},"howToChoose":{"title":"如何选择 AI 会议笔记工具","intro":"几乎所有 AI 会议笔记工具都会录制通话、转写内容并生成摘要。真正的差异可以归结为三个问题：","criteria":[{"lead":"音频会被发送到哪里？","body":"云端工具会把会议上传并存储在自己的服务器上，本地工具则把数据留在设备内。对于人力资源、法律、医疗或客户对话，这往往是决定性因素。"},{"lead":"机器人会加入通话吗？","body":"大多数工具会在参会者列表中加入一个可见的机器人。少数工具无需机器人即可采集音频，干扰更小，也不会引发大家追问陌生参会者是谁的尴尬。"},{"lead":"免费方案到底包含什么？","body":"“免费”可能只是每月几场会议，也可能真正不限量。决定使用前，请先查看上限。"}]},"comparison":{"title":"快速对比","headers":{"tool":"工具","bestFor":"最适合","runs":"运行位置","bot":"机器人？","freeTier":"免费方案"}},"picks":{"title":"逐一了解最佳 AI 会议笔记工具","keepInMind":"需要注意：","openWhisprLink":"了解 OpenWhispr","compareLink":"OpenWhispr 与 {name} 对比","tools":[{"name":"OpenWhispr","bestFor":"隐私与开源","runs":"本地","bot":"无机器人","freeTier":"本地不限量，免费","blurb":"先说明：这是我们的产品，所以下面只讲真实情况。OpenWhispr 使用 OpenAI Whisper 和 NVIDIA Parakeet 在你的设备上转写，采用开源 MIT 许可证，并且无需向通话中派机器人即可采集会议音频。它也能作为全系统听写工具，让你在任何应用中用语音输入。","pros":["音频绝不离开设备；不会上传任何内容","在 macOS、Windows 和 Linux 上开源且可审查","本地转写不限量，永久免费"],"caveat":"它不会派机器人自动加入日程中的通话，你需要自己开始录音。要获得专用云端笔记工具那样精致的摘要，也需要多做一些设置。","href":"/ai-meeting-note-taker"},{"name":"Fathom","bestFor":"最佳免费云端工具","runs":"云端","bot":"机器人","freeTier":"慷慨的免费方案","blurb":"Fathom 是获取自动会议回顾最省事的方式之一，在云端笔记工具中也拥有相当慷慨的免费方案。它的机器人会加入并录制通话，几乎无需设置就能生成清晰的摘要、重点和待办事项。","pros":["面向个人的优秀免费方案","轻松生成摘要和重点","良好支持 Zoom、Teams 和 Meet"],"caveat":"它在云端运行：机器人会加入通话，会议也会上传至 Fathom 的服务器。","href":"/compare/fathom"},{"name":"Otter.ai","bestFor":"实时协作","runs":"云端","bot":"机器人","freeTier":"300 分钟/月（以往）","blurb":"Otter 是成熟的会议转写平台，具备强大的实时转写、共享和集成功能。它适合希望共同保存、搜索通话记录并协作标记重点的团队。","pros":["实时转写","团队共享和工作区功能","广泛的集成"],"caveat":"它通过机器人在云端运行，免费方案有上限，重度用户很快就会用完。","href":"/compare/otter"},{"name":"Fireflies.ai","bestFor":"销售与 CRM","runs":"云端","bot":"机器人","freeTier":"有上限的免费方案","blurb":"Fireflies 面向营收团队：机器人加入并转写通话，再提供对话分析，并与 Salesforce 和 HubSpot 深度同步。如果会议笔记需要进入销售管道，它是很有力的选择。","pros":["对话智能与分析","深度 CRM 集成","可搜索的团队转写资料库"],"caveat":"它通过机器人在云端运行；最实用的功能位于按席位付费的方案中。","href":"/compare/fireflies"},{"name":"Granola","bestFor":"无机器人的云端记事本","runs":"云端","bot":"无机器人","freeTier":"免费方案","blurb":"Granola 的做法比大多数产品更自然：无需在参会者列表中加入机器人，就能采集电脑音频，并配有简洁的记事本和日历同步。它很受希望获得整洁 AI 会议笔记的 Mac 用户欢迎。","pros":["没有机器人加入通话","简洁、快速的笔记体验","日历同步"],"caveat":"处理仍在云端进行，因此即使看不到机器人，会议音频也会离开你的设备。","href":"/compare/granola"},{"name":"Notta","bestFor":"多语言会议","runs":"云端","bot":"机器人","freeTier":"有上限的免费方案","blurb":"Notta 是云端转写和会议工具，支持的语言尤其广泛，还能上传文件和录音。会议涉及多种语言时，它是不错的选择。","pros":["广泛的语言支持","转写实时通话或上传的文件","导出简单"],"caveat":"它在云端运行；更长的录音和更高的准确率需要付费方案。","href":null},{"name":"tl;dv","bestFor":"录制与片段分享","runs":"云端","bot":"机器人","freeTier":"免费录制方案","blurb":"tl;dv 专注于录制和分享会议片段，包括带时间戳的短片、重点和交接内容，并提供实用的免费方案。需要向团队分享会议片段时很方便。","pros":["剪辑并分享关键时刻","免费录制方案","支持 Zoom、Meet 和 Teams"],"caveat":"它通过机器人在云端运行；录制内容保存在 tl;dv 的服务器上。","href":null}]},"faq":{"title":"常见问题"},"cta":{"title":"想让 AI 会议笔记永不离开设备？","body":"OpenWhispr 是免费、开源的 AI 会议笔记工具，直接在你的电脑上录音和转写：无需机器人，也不上传云端。macOS、Windows 和 Linux 均可免费使用。","github":"在 GitHub 上查看"}},"blogZoomRecording":{"metadata":{"title":"如何录制 Zoom 会议（无论是否为主持人）| OpenWhispr","description":"了解主持人或参会者如何在本地、云端或无机器人的情况下录制 Zoom 会议，并在自己的设备上获得私密转写。","keywords":["如何录制 zoom 会议","录制 zoom 会议","参会者如何录制 zoom","未经许可录制 zoom","录制带声音的 zoom","zoom 录制文件保存位置","本地录制 zoom","获取 zoom 会议转写","zoom 会议转写","录制 zoom 是否合法","mac 录制 zoom","windows 录制 zoom"]},"breadcrumb":{"home":"首页","blog":"博客","article":"如何录制 Zoom 会议"},"schema":{"title":"如何录制 Zoom 会议（无论是否为主持人）","description":"了解主持人或参会者如何在本地、云端或无机器人的情况下录制 Zoom 会议，并在自己的设备上获得私密转写。"},"faqs":[{"question":"不是主持人也能录制 Zoom 会议吗？","answer":"可以。主持人可在通话中授予权限（参会者 → 悬停你的姓名 → 更多 → 允许录制），从而启用 Zoom 内置录制。否则，可用电脑上的独立工具录制屏幕和系统音频，无需 Zoom 权限即可捕捉看到和听到的全部内容。无论哪种方式，都要先告知其他参会者。"},{"question":"Zoom 录制文件保存在哪里？","answer":"本地录制默认保存在电脑“文档”下的 Zoom 文件夹中，包含 MP4 视频和独立 M4A 音频。云录制（仅付费计划）保存在 Zoom 服务器，提供分享链接，部分计划还提供自动转写。重视隐私时应选择本地录制。"},{"question":"录制 Zoom 会议合法吗？","answer":"取决于参会者所在地。有些地区只需一人同意，有些要求所有人同意；欧盟 GDPR 也将录下的声音视为个人数据。这不是法律建议。安全做法是提前说明并取得明确同意。"},{"question":"如何录制带声音的 Zoom 会议？","answer":"Zoom 内置功能会自动录制视频和会议音频。使用独立录屏工具时，请启用系统音频，而不只是麦克风，否则视频会没有声音。OpenWhispr 等本地工具可直接捕捉完整的系统音频。"},{"question":"能在主持人不知情时录制吗？","answer":"Zoom 内置录制会向所有人显示“正在录制”。外部工具不会显示，但秘密录制可能违法，也会破坏信任。请始终告知大家；录制应当形成准确、共享的记录，而不是暗中留存。"},{"question":"如何获得 Zoom 会议转写？","answer":"付费计划可自动转写云录制。本地录制可在任何计划下自行转写。OpenWhispr 在设备上生成支持搜索的准确转写，支持 100+ 种语言，并可用 AI 汇总笔记和行动项。"},{"question":"可以在手机上录制 Zoom 吗？","answer":"付费计划的主持人可在 Zoom 移动应用中录制到云端。参会者可使用 iPhone 控制中心或 Android 的屏幕录制，并开启麦克风或内部音频。之后可转写该文件。"},{"question":"怎样最私密地录制 Zoom？","answer":"选择本地录制，让文件不离开设备，并避免把会议上传给第三方的云端笔记机器人。OpenWhispr 在电脑上完成系统音频录制和转写：没有机器人入会，也不上传数据，适合人力、法律和客户数据等敏感会议。"}],"sections":[{"id":"host","title":"主持人录制"},{"id":"participant","title":"不是主持人时"},{"id":"local","title":"本地录制，无机器人"},{"id":"mobile","title":"手机录制"},{"id":"transcript","title":"获取转写"},{"id":"legal","title":"同意与合法性"},{"id":"storage","title":"保存位置"},{"id":"troubleshoot","title":"故障排除"},{"id":"faq","title":"常见问题"}],"hero":{"title":"如何录制 Zoom 会议（无论是否为主持人）","subtitle":"以主持人或参会者身份，在云端或自己的设备上录制带声音的 Zoom 通话，再将其变成可搜索的转写，并了解同意规则。","date":"2026年7月6日","tag":"指南"},"opening":{"lead":"如果你是主持人或已获授权，请点击工具栏中的“录制”来录制 Zoom 会议。","body":"没有权限的参会者可请主持人启用录制，或用电脑上的独立工具录制屏幕和系统音频；后者无需 Zoom 权限，可捕捉完整通话。无论采用何种方式，都要先告知其他人。","scope":"本文涵盖主持人的本地与云录制、参会者录制、无机器人设备端录制、手机录制、转写、同意规则，以及录制按钮变灰时的处理方法。","updated":"最后更新：2026年7月6日。"},"host":{"title":"主持人录制（Zoom 内置功能）","intro":"由你发起会议时，只需点击工具栏中的“录制”。付费计划会要求选择两个位置之一：","options":[{"title":"在此计算机上录制","body":"会议结束时在电脑上保存 MP4 视频和 M4A 音频。所有计划均可使用，包括免费计划。"},{"title":"录制到云端","body":"上传到 Zoom 服务器并提供分享链接；部分计划还有自动转写。仅限付费计划。"}],"outroPrefix":"开始后，Zoom 会向所有人显示“正在录制”。会议结束后，本地文件会转换并默认放入“文档”中的","outroSuffix":"文件夹。"},"participant":{"title":"不是主持人时","intro":"参会者必须得到主持人允许才能使用 Zoom 的录制功能。可选两种方式：","options":[{"title":"请求权限。","body":"主持人打开“参会者”，悬停你的姓名，点击“更多”并选择“允许录制”，录制按钮便会出现。"},{"title":"录制屏幕和系统音频。","body":"独立工具可在没有主持人权限时捕捉屏幕和扬声器声音，是主持人无法或不愿启用录制时的可靠方案。"}],"warningPrefix":"普通录屏常常只录到你的麦克风，漏掉其他人的声音。请启用","systemAudio":"系统音频","warningSuffix":"，或使用专门捕捉会议声音的工具。"},"local":{"title":"本地录制，不用机器人和云端","cloudPrefix":"Otter、Fireflies、Fathom 等云端笔记服务会让","bot":"机器人","cloudSuffix":"加入通话并把音频传到服务器。即使没有机器人，有些服务仍会把会议发到设备外处理，不适合敏感会议。","approachPrefix":"采用相反方式，直接在电脑上捕捉系统音频：","noBot":"没有机器人加入通话","approachSuffix":"，音频也不离开设备。它可作为参会者在任意 Zoom 计划及 macOS、Windows、Linux 上使用。你仍需告知录制，但没有上传或第三方访问。","bullets":["录制整场通话，而不只是麦克风。","参会列表无机器人，也不上传云端。","在设备上以 100+ 种语言转写和总结。"],"github":"在 GitHub 上查看"},"mobile":{"title":"在手机上录制 Zoom 会议","paragraph1":"付费计划主持人可在 Zoom 应用中点“更多”，再选“录制到云端”。参会者可用 iPhone 或 Android 自带录屏，并开启麦克风或内部音频，以免只得到无声视频。","paragraph2":"手机录制就是视频文件，会议结束后可像其他录制一样转写。"},"transcript":{"title":"把录制变成转写和笔记","paragraph1":"只有可阅读、可搜索时，录制才真正有用。付费计划可自动转写云录制；本地录制或免费计划需自行转写保存的文件。","paragraph2Prefix":"将 MP4 或 M4A 导入","paragraph2Suffix":"，即可在设备上得到带时间戳的准确转写，并让 AI 总结决定、行动项和后续事项。由于完全在本地运行，机密录制也不会上传。"},"legal":{"title":"同意与合法性","paragraph1":"是否合法取决于参会者所在地。有些地方一人同意即可，有些要求所有人同意；欧盟 GDPR 将录下的声音视为个人数据。这不是法律建议。请在开始前说明并获得明确同意。","paragraph2":"Zoom 内置录制会自动在屏幕上通知。外部工具不会，因此告知责任在你。秘密录制存在法律和信任风险；目的应是创建所有人都能依赖的共享记录。"},"storage":{"title":"录制文件最终保存在哪里","localLabel":"本地录制","localBody":"以 MP4 和 M4A 留在电脑上，默认位于“文档”的 Zoom 文件夹。除非你分享，否则不会离开设备。","cloudLabel":"云录制","cloudBody":"保存在 Zoom 服务器上，受计划容量和组织保留策略限制。","botLabel":"机器人笔记服务","botBody":"会将整场会议上传给第三方，部分服务商可能用其训练模型。人力、法律、医疗或客户数据应通过本地录制与转写保持控制。"},"troubleshoot":{"title":"故障排除：录制按钮缺失或变灰","items":[{"title":"你不是主持人且没有权限。","body":"请主持人允许，或使用本地屏幕与音频录制。"},{"title":"账户层面禁用了录制。","body":"受管账户可能需要管理员在 Zoom 网页设置中启用。"},{"title":"没有云录制。","body":"云录制需要付费计划；免费计划仅支持本地录制。"},{"title":"录制没有声音。","body":"录屏只捕捉了麦克风。请开启系统音频或使用专用工具。"}]},"faq":{"title":"常见问题"},"closing":{"title":"私密地录制并转写会议","body":"OpenWhispr 在设备上把 Zoom 音频变成笔记，无机器人、无上传。macOS、Windows 和 Linux 免费使用。","github":"在 GitHub 上查看"}},"blogStreamingStt":{"metadata":{"title":"CPU 流式语音转文字：本地实时运行 | OpenWhispr","description":"OpenWhispr 如何在 CPU 上运行 NVIDIA Nemotron 流式 ASR：通过本地 WebSocket 进行缓存感知流式处理，实时返回部分结果，无需 GPU 和云端。","keywords":["流式语音转文字","本地流式asr","nemotron asr","缓存感知流式asr","sherpa onnx","本地实时转录","实时听写预览","流式asr electron","端侧语音识别"]},"breadcrumb":{"home":"首页","blog":"博客","article":"在 CPU 上运行流式语音转文字"},"schema":{"title":"在 CPU 上运行流式语音转文字：将 NVIDIA Nemotron 集成到桌面应用","description":"OpenWhispr 如何在 CPU 上运行 NVIDIA Nemotron 流式 ASR：通过本地 WebSocket 进行缓存感知流式处理，实时返回部分结果，无需 GPU 和云端。"},"faqs":[{"question":"什么是流式语音转文字？","answer":"流式（online）语音转文字会在说话尚未结束时处理音频，并在固定延迟范围内输出部分结果，通常为80毫秒到约1秒。Whisper 等批处理（offline）模型则要等录音结束后一次性处理全部内容。实时字幕、语音智能体和边说边显示的听写预览都依赖流式处理。"},{"question":"没有 GPU，能否在本地运行流式语音转文字？","answer":"可以。NVIDIA Nemotron 流式模型有600M参数，以约650MB的INT8量化ONNX文件提供，通过无需Python或PyTorch的原生二进制运行时sherpa-onnx，可在现代CPU上实时运行。OpenWhispr为macOS、Windows和Linux提供了这套配置。"},{"question":"什么是缓存感知流式 ASR？","answer":"这是 NVIDIA cache-aware FastConformer 架构：模型会在音频分块之间保留编码器内部状态。每个新分块只处理一次，并使用此前语音的缓存上下文，而不是重新编码相互重叠的音频窗口。因此，低延迟流式识别可以轻量到在笔记本CPU上运行。"},{"question":"实时预览文字为何显示后还会变化？","answer":"流式模型会随着更多音频到达而修正假设。单词边界或同音词有时只能靠后文确定，因此部分结果会原地替换。这是有意设计。停止听写时，OpenWhispr会刷新模型尾部，并将稳定后的流式文字写入转录；若刷新中断或结果被截断，应用会自动重新解码完整录音，绝不会把不稳定的流式结果粘贴出去。"},{"question":"流式转录是否不如批处理准确？","answer":"略低，因为流式模型无法看到未来音频。Nemotron 英语流式模型在 Open ASR Leaderboard 数据集上使用1.12秒分块时，平均词错误率为6.93%。这已非常接近批处理模型，但仍落后于 NVIDIA 最佳英语批处理模型的5.91%。选择流式模型，就是以这点差距换取实时文字和停止即得的转录；若最高准确率更重要，可选择offline Parakeet模型。"},{"question":"OpenWhispr 的流式模式支持哪些语言？","answer":"目前有两个 Nemotron 模型：仅英语模型，以及支持自动语言检测的 Nemotron 3.5。后者有15种可直接转录的语言（英语、西班牙语、法语、意大利语、葡萄牙语、荷兰语、德语、土耳其语、俄语、阿拉伯语、印地语、日语、韩语、越南语、乌克兰语）。Parakeet 和 Whisper 等 offline 模型覆盖更多语言，但预览使用缓冲分块而不是实时流。"}],"sections":[{"id":"the-problem","title":"问题：分块式预览"},{"id":"cache-aware","title":"缓存感知流式处理是什么"},{"id":"architecture","title":"系统架构"},{"id":"gotchas","title":"实现过程中踩过的坑"},{"id":"single-pass","title":"直接提交流式结果"},{"id":"footprint","title":"性能与资源占用"},{"id":"limitations","title":"坦诚的局限"},{"id":"model-choice","title":"如何选择模型"},{"id":"faq","title":"常见问题"}],"hero":{"title":"在 CPU 上运行流式语音转文字：将 NVIDIA Nemotron 集成到桌面应用","subtitle":"OpenWhispr 如何通过本地 WebSocket 上的缓存感知流式模型，在不用 GPU 和云端的情况下边说边显示文字，以及让它可靠运行所需的工程工作。","date":"2026年7月18日","tag":"技术"},"opening":{"lead":"流式语音转文字不必等待录音结束，而是在你说话时同步转录。从版本1.7.6开始，OpenWhispr完全在设备端完成这项工作：NVIDIA Nemotron 流式模型（600M参数、INT8量化）通过本地sherpa-onnx WebSocket服务器在CPU上运行；部分结果到达时听写预览会实时更新，停止时则立即将流式文字提交为正式转录。","body":"无需 GPU，无需 Python 环境，音频也不会离开设备。","guidePrefix":"本文记录具体工程实现：旧版实时预览为何迟缓，缓存感知流式处理究竟是什么，整个管线如何装进Electron应用，版本固定错误为何导致全面误解码，以及我们如何从每次重新解码整段录音改为直接提交流。模型之间的比较请参阅","comparisonLink":"Parakeet vs Whisper vs Nemotron","guideSuffix":"。","updated":"最后更新于2026年7月18日。实现细节对应 OpenWhispr 1.7.6 发布的流式转录功能；文中内容均可在开源代码库中核查。"},"factCheck":{"title":"事实速查（第一手来源）","items":[{"lead":"Nemotron 的 cache-aware FastConformer 只处理新增音频，并复用缓存的编码器上下文，","body":"运行时可将分块大小配置为80ms到1.12s。","sources":[{"href":"https://huggingface.co/nvidia/nemotron-speech-streaming-en-0.6b","label":"Nemotron Streaming EN 模型卡"},{"href":"https://huggingface.co/blog/nvidia/nemotron-speech-asr-scaling-voice-agents","label":"NVIDIA 对缓存感知流式处理的说明"}]},{"lead":"英语模型在1.12s分块下流式运行，平均WER为6.93%；","body":"多语言3.5模型支持40种语言与地区组合，并能自动检测。","sources":[{"href":"https://huggingface.co/nvidia/nemotron-3.5-asr-streaming-0.6b","label":"Nemotron 3.5 ASR 模型卡"}]},{"lead":"全部推理通过 sherpa-onnx 完成；它是无需 Python 的原生 ONNX 运行时，","body":"我们使用其online（流式）WebSocket服务器，并固定和打包版本1.13.4。","sources":[{"href":"https://github.com/k2-fsa/sherpa-onnx","label":"GitHub 上的 sherpa-onnx"}]},{"lead":"INT8 ONNX模型在磁盘上分别占632MB（英语）和650MB（多语言），","body":"只需下载一次，之后保存在本地缓存中。","sources":[{"href":"https://github.com/OpenWhispr/openwhispr","label":"OpenWhispr 模型注册表（开源）"},{"href":"https://openwhispr.com/models","label":"OpenWhispr 模型页面"}]}]},"problem":{"title":"问题：在批处理模型上搭建实时预览","paragraphs":[{"body":"OpenWhispr一直有转录预览：听写时用一个小窗口显示模型听到的内容。1.7.6之前，我们采用了让批处理模型看起来实时的唯一办法：缓存约1.5秒麦克风音频，完整转录这个分块，追加文字，然后重复。"},{"body":"这种方式有三个结构性问题。第一是延迟：缓冲区填满前文字无法出现，因此预览总会落后1.5秒加推理时间。第二是边界：跨越两个分块的单词会被切断，而分块独立转录，双方都无法修正。第三是上下文：第五块不知道第四块说了什么，模型每1.5秒都要重新判断同一类歧义。"},{"body":"缩短分块会显得更快，却因上下文减少、边界增多而大幅降低准确率。重叠分块能修复边界，却要重复转录同一音频。真正需要的是为转录音频流而训练的模型，这就是Nemotron。"}]},"cacheAware":{"title":"缓存感知流式处理究竟是什么","paragraphs":[{"lead":"缓存感知流式模型会在分块之间保留内部状态。","body":"下一段音频到达时，编码器不会重读此前全部语音，只处理新帧，并查询已计算上下文的缓存。它和让聊天机器人更快的KV缓存原理相同，只是用在语音编码器上。"},{"body":"这种设计带来两个结果。计算量不再随录音长度增长：无论录到10秒还是10分钟，每块成本都一样，因此纯CPU流式处理成为可能。上下文也会延续：模型利用此前全部语音解决歧义，后续音频给出新证据时会修正较早的词。文字在你说完后稍微“稳定”一下，不是故障，而是模型用更充分的依据改正判断。"},{"body":"延迟预算可以调节。Nemotron提供80ms（最快，每步上下文最少）到1.12s（准确率最高，在leaderboard数据集上平均6.93% WER，与最佳批处理模型相差约一个百分点）的分块。训练采用NVIDIA cache-aware FastConformer和transducer解码器；其"}],"engineeringPostLink":"工程文章","engineeringPostSuffix":"深入解释了相关原理。"},"architecture":{"title":"OpenWhispr 内部架构","paragraphs":[{"body":"管线刻意保持简单。AudioWorklet以16kHz采集麦克风PCM，将帧转换为float32传输格式，再通过WebSocket推送到sherpa-onnx online服务器。OpenWhispr在本地启动这个原生二进制，并通过127.0.0.1通信，每次录音保持一条长连接。服务器以INT8权重运行Nemotron并返回JSON部分结果；预览窗口会原地替换全文，而不是不断追加。"},{"body":"为何采用独立服务器进程，而不在进程内推理？为了隔离。原生推理运行时可能崩溃；ONNX内核一次错误的内存分配不应拖垮整个桌面应用。辅助进程崩溃只需重连，不会丢失听写。它也让原生代码完全离开Electron主进程：应用负责协议，辅助进程负责计算。"},{"body":"同一模式也用于我们的"}],"diarizationLink":"本地说话人分离","diarizationSuffix":"——一套sherpa-onnx工具链，多种任务，全都在设备端完成。"},"gotchas":{"title":"实现过程中踩过的坑","items":[{"label":"运行时版本比预想的更重要。","body":"Nemotron需要sherpa-onnx 1.13.4或更新版本才能准确解码。在为Parakeet打包的旧运行时上，它不会明显报错，而是直接转录错误。修复很普通：升级并重新固定各平台二进制。普遍教训是，对模型运行时而言，“能运行”和“结果正确”是两项测试，只有后者算数。"},{"label":"流会失败，所以旧路径必须保留。","body":"端口可能被占用，模型文件可能缺失，服务器也可能在录音中退出。若长连接无法启动，预览会自动回退到缓冲分块路径——更慢，但不会空白。offline模型（Parakeet、Whisper）本来就保留分块预览；注册表中的每个模型声明运行时为online或offline，应用据此选择。"},{"label":"部分结果需要不同的渲染方式。","body":"分块路径追加文字，流式假设却会自我修正，因此UI在每个部分结果到达时替换整段预览。如果把会变的假设不断追加，就会产生断断续续的重复文字——这是我们很早修复的真实UX问题。"}]},"singlePass":{"title":"直接提交流：从两次解码到一次","paragraphs":[{"body":"第一版只把流式结果当作预览：停止听写后，应用会从头重新解码完整录音，并粘贴那份结果。这样很稳妥，因为最终处理能看到完整上下文；但每次听写都付出两次解码的成本，而且你已经看着文字稳定下来，却仍要等待几乎总是相同的第二意见。"},{"body":"于是我们在同一版本中把流设为事实来源。无论预览窗口是否打开，听写都会走长连接；停止时，应用刷新模型尾部，直接提交流式文字，不再进行第二次解码。结束延迟只剩一次尾部刷新，每次听写的CPU开销约减半。"},{"body":"安全网没有消失，而是成为回退机制。停止刷新会感知截断，只要结果仍在到达便延长期限。若连接中断或结果被截断，应用会丢弃流，改用经典的“先录音、后转录”方式处理完整录音。只有既快又正确时才走快速路径。"}]},"footprint":{"title":"性能与资源占用","items":[{"label":"模型：","body":"Nemotron Speech Streaming EN 0.6B（632MB）和Nemotron 3.5 ASR Streaming 0.6B（650MB、15种可转录语言、自动检测）——INT8 ONNX，只下载一次并在本地缓存。"},{"label":"运行时：","body":"sherpa-onnx 1.13.4，自包含的原生二进制。无需Python、PyTorch或CUDA。"},{"label":"硬件：","body":"现代笔记本CPU可实时处理；Apple Silicon和较新的x86都能轻松跟上说话速度。"},{"label":"体感：","body":"部分文字落后语音远不到1秒；停止时一次尾部刷新即可提交转录，无需再次解码。"}],"privacyPrefix":"这条路径完全不接触互联网。应用通过127.0.0.1与服务器通信（Windows上还用限定范围的防火墙规则阻止网络访问该端口），模型文件位于本地缓存，音频从不离开设备。这与","privacyLink":"我们的本地转录体系","privacySuffix":"其余部分采用相同的隐私标准。"},"limitations":{"title":"坦诚的局限","items":[{"label":"流式处理以少量准确率换取即时性。","body":"在相同基准上，流式约为~6.9%，NVIDIA最佳英语批处理模型为5.91%。现在正式转录来自流，因此你确实接受了这项取舍。若只在意最终结果，offline Parakeet仍是准确率更高的选择。"},{"label":"语言覆盖范围小于Whisper。","body":"多语言模型有15种可直接转录语言，Whisper则有99种。普通话和泰语流式识别目前尚不可用。"},{"label":"部分文字会变化。","body":"假设会随上下文修正。我们认为看着它逐渐稳定是一项优点；若觉得分心，可关闭预览，听写仍和以前完全一样。"},{"label":"磁盘会再占用650MB","body":"如果已经装有offline模型，这就是为不同任务优化第二套引擎的代价。"}]},"modelChoice":{"title":"在 OpenWhispr 中选择模型","wantHeader":"你的需求","pickHeader":"选择","rows":[{"want":"听写时实时预览英语文字","pick":"Nemotron Speech Streaming EN 0.6B"},{"want":"实时预览西班牙语、日语、阿拉伯语、印地语等","pick":"Nemotron 3.5 ASR Streaming 0.6B"},{"want":"英语准确率最高，不需要实时文字","pick":"Parakeet Unified EN 0.6B"},{"want":"使用NVIDIA范围外的语言","pick":"Whisper (turbo 或 large)"}],"catalogPrefix":"这些模型都可在“设置”的下拉菜单中选择；完整的容量和取舍信息见","catalogLink":"模型页面","catalogSuffix":"。OpenWhispr免费、开源，并支持macOS、Windows和Linux。","githubCta":"在 GitHub 上加星"},"faq":{"title":"常见问题"}},"blogParakeetNemotron":{"metadata":{"title":"Parakeet、Whisper 与 Nemotron：2026 年最佳本地 STT | OpenWhispr","description":"客观比较主流本地语音转文字模型：准确率（WER）、速度、语言覆盖和流式识别。数据均来自一手资料。","keywords":["parakeet 与 whisper 对比","nemotron asr","最佳本地语音转文字","whisper 替代方案","nvidia parakeet","本地语音转文字模型","开源语音转文字","2026 最佳语音转文字模型","nemotron 与 whisper 对比","离线语音识别"]},"breadcrumb":{"home":"首页","blog":"博客","article":"Parakeet、Whisper 与 Nemotron 对比"},"schema":{"title":"Parakeet、Whisper 与 Nemotron：2026 年最佳本地语音转文字方案","description":"客观比较主流本地语音转文字模型：准确率（WER）、速度、语言覆盖和流式识别，并引用一手资料中的数据。"},"faqs":[{"question":"2026 年最佳本地语音转文字模型是什么？","answer":"对于英语和主要欧洲语言，NVIDIA Parakeet TDT 0.6B v3 的综合表现最佳：它在 Open ASR Leaderboard 上的平均词错误率为 6.34%，优于 Whisper large-v3 约 7.4% 的成绩，体积仅为后者的四分之一，CPU 吞吐量也高得多。如果需要边说边显示文字，NVIDIA Nemotron 流式模型是最佳选择。如果所需语言不在 NVIDIA 模型的覆盖范围内，请使用 Whisper。"},{"question":"Parakeet 比 Whisper 更准确吗？","answer":"在 Open ASR Leaderboard 的英语基准中，是的。Parakeet TDT 0.6B v3 的平均 WER 为 6.34%，Whisper large-v3 约为 7.4%，仅支持英语的 Parakeet Unified 更达到 5.91%。但需要说明的是，日常听写中约 ~1 个 WER 百分点的差距很难察觉。Parakeet 更明显的优势在于速度、效率，以及不会在静音时生成幻觉文本；后者是 Whisper 的已知问题。"},{"question":"在本地运行这些模型需要 GPU 吗？","answer":"不需要。三者都能在现代 CPU 上运行。Parakeet 和 Nemotron 使用针对 CPU 推理设计的 INT8 量化 ONNX 模型（磁盘占用约 630–680MB）。Whisper 通过 whisper.cpp 在 CPU 上运行，不过其 3GB 大模型能从 GPU 加速中受益。OpenWhispr 支持 Apple Silicon 上的 Metal、NVIDIA 上的 CUDA 以及 AMD/Intel GPU 上的 Vulkan，并可自动回退至 CPU。"},{"question":"Nemotron ASR 支持哪些语言？","answer":"NVIDIA Nemotron 3.5 ASR 在单个 600M 参数模型中覆盖 40 个语言区域，并支持自动语言检测。OpenWhispr 已启用其中 15 种可直接转录的语言：英语、西班牙语、法语、意大利语、葡萄牙语、荷兰语、德语、土耳其语、俄语、阿拉伯语、印地语、日语、韩语、越南语和乌克兰语。另有仅支持英语的 Nemotron 流式模型。"},{"question":"Whisper 已经过时了吗？","answer":"没有。Whisper 仍支持 99 种语言，远多于 Parakeet 的 25 种或 Nemotron 的 40 个语言区域。对于日语、韩语和越南语之外的亚洲语言、低资源语言以及频繁切换语言的音频，它仍是最稳妥的选择。它也是经过最多实际检验、生态最成熟的模型。变化只在于 Whisper 已不再占据英语准确率和速度的首位。"},{"question":"OpenWhispr 支持其中哪些模型？","answer":"全部支持。OpenWhispr 通过 whisper.cpp 提供六种大小的 Whisper，通过 sherpa-onnx 提供两款 Parakeet 和两款 Nemotron 流式模型，并可在设置中切换。所有处理均在设备上完成，音频绝不会离开您的电脑；应用免费且开源。"}],"sections":[{"id":"contenders","title":"三个候选模型"},{"id":"accuracy","title":"准确率"},{"id":"speed","title":"速度与效率"},{"id":"languages","title":"语言覆盖"},{"id":"streaming","title":"流式识别：真正的分水岭"},{"id":"hardware","title":"硬件要求"},{"id":"which-to-choose","title":"应该选哪一个？"},{"id":"try-them","title":"三个都试一遍"},{"id":"faq","title":"常见问题"}],"hero":{"title":"Parakeet、Whisper 与 Nemotron：2026 年最佳本地语音转文字方案","subtitle":"如今，三款开放模型都能完全在您的电脑上将语音转成文字。本文依据一手资料中的可靠数据，比较它们在准确率、速度、语言覆盖和流式识别方面的真实表现。","date":"2026 年 7 月 18 日","tag":"对比"},"opening":{"lead":"对于英语和主要欧洲语言，NVIDIA Parakeet TDT 0.6B v3 是 2026 年最佳本地语音转文字模型：其实测准确率优于 Whisper large-v3，体积仅为后者四分之一，在普通 CPU 上也快得多。如果希望文字在说话时实时出现，NVIDIA 新推出的 Nemotron 流式模型是首批专门为此设计的本地模型。如果您使用的是 NVIDIA 模型尚未覆盖的众多语言之一，那么支持 99 种语言的 OpenAI Whisper 仍是正确选择。","paragraph2":"2026 年最出人意料的变化是，Whisper 失去了保持三年的开放模型准确率桂冠。不过，“哪款模型最好”已经真正变成三方竞争，因为它们各有所长：Parakeet 优化单位功耗下的准确率，Nemotron 优化延迟，Whisper 则优化语言广度。本文使用模型卡和公开 Leaderboard 的数据进行比较，不凭感觉，也不夸大差异。","disclosurePrefix":"最后更新于 2026 年 7 月 18 日。我们开发","disclosureSuffix":"，这是一款同时提供三大模型系列的开源听写应用，因此我们会在生产环境中测试并支持每一种模型。这也说明了我们的立场：我们不销售其中任何模型，下面所述的实际取舍与我们向用户解释的完全相同。"},"factCheck":{"title":"关键事实速览（一手资料）","items":[{"lead":"Parakeet TDT 0.6B v3 在 Open ASR Leaderboard 的八项基准中平均词错误率为 6.34%，吞吐速度达到实时的 3,332×。","body":"600M 参数、25 种语言、CC-BY-4.0 许可证。","sources":[{"href":"https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3","label":"Parakeet TDT v3 模型卡"},{"href":"https://huggingface.co/spaces/hf-audio/open_asr_leaderboard","label":"Open ASR Leaderboard"}]},{"lead":"Whisper large-v3 在同一榜单上约为 7.4%，并支持 99 种语言","body":"拥有 1.55B 参数，采用 MIT 许可证。","sources":[{"href":"https://github.com/openai/whisper","label":"OpenAI Whisper 代码仓库"},{"href":"https://huggingface.co/openai/whisper-large-v3","label":"Whisper large-v3 模型卡"}]},{"lead":"Nemotron Speech Streaming EN 0.6B 在流式识别时的平均 WER 为 6.93%","body":"（1.12 秒音频块），在 LibriSpeech test-clean 上达到 2.32%，训练数据包含 530,000 小时音频。","sources":[{"href":"https://huggingface.co/nvidia/nemotron-speech-streaming-en-0.6b","label":"Nemotron Streaming EN 模型卡"}]},{"lead":"Nemotron 3.5 ASR 以单个 600M 流式模型覆盖 40 个语言区域","body":"支持自动语言检测，于 2026 年 6 月 4 日以 OpenMDW 许可证发布。","sources":[{"href":"https://huggingface.co/nvidia/nemotron-3.5-asr-streaming-0.6b","label":"Nemotron 3.5 ASR 模型卡"}]},{"lead":"下文提到的所有磁盘大小和应用内行为均可核实","body":"可在 OpenWhispr 的开源模型注册表中查证。","sources":[{"href":"https://github.com/OpenWhispr/openwhispr","label":"GitHub 上的 OpenWhispr"},{"href":"https://openwhispr.com/models","label":"OpenWhispr 模型页面"}]}]},"contenders":{"title":"三个候选模型","intro":"三者都是可免费下载、离线运行的开放权重模型。架构上的不同，几乎解释了实际使用中的所有差异。","items":[{"title":"OpenAI Whisper (2022)","body":"让开放语音识别进入主流的模型。它是一个编码器—解码器 Transformer，使用 680,000 小时音频训练，提供从 75MB（tiny）到 3GB（large-v3）的六种大小，覆盖 99 种语言。它像语言模型一样逐个 token 生成文字，因此也是三者中最慢的。MIT 许可证。"},{"title":"NVIDIA Parakeet (2024–2025)","body":"为批量转录设计的 600M 参数转导器（TDT）模型。它不采用自回归方式生成 token，而是在一次读取音频的过程中输出文字，速度快得多，也不会在静音时编造文字。两款重要版本是多语言 v3（25 种语言，INT8 ONNX 为 680MB）和仅支持英语的 Unified（631MB）；后者目前拥有领先的英语准确率。CC-BY-4.0 许可证。"},{"title":"NVIDIA Nemotron ASR (2026)","body":"面向流式识别的最新系列：具有缓存感知能力的 FastConformer 转导器会随音频到达而转录，无需等待录音结束。英语模型于 2026 年 1 月推出，覆盖 40 个语言区域的多语言 3.5 模型于 6 月跟进。两者均为 600M 参数（INT8 ONNX 为 632–650MB），采用开放的 OpenMDW 许可证。"}]},"accuracy":{"title":"准确率：Parakeet 胜出，但差距没有看起来那么大","definitionPrefix":"语音模型通常使用","definitionTerm":"词错误率（WER）","definitionSuffix":"衡量，即模型识别错误的单词比例，越低越好。标准参考是 Hugging Face Open ASR Leaderboard，它对八个不同的英语数据集（会议、财报电话会、TED Talks、有声书等）的 WER 取平均值，避免单个简单数据集人为抬高模型成绩。","analysis":"应当客观看待图表：这里的每款模型都已达到生产级别，总差距仅约 1.5 个百分点。在日常听写的清晰音频上，6.3% 与 7.4% 的体验接近；差别会在口音、噪声和专业术语等困难音频中显现。NVIDIA 数据值得关注的是取得成绩所需的成本：Parakeet 仅用 600M 参数便达到该准确率，而 Whisper 需要 1.55B；Nemotron 更是在完全看不到后续音频的流式条件下取得 6.93%。","calloutTitle":"Whisper 的已知弱点：在静音中产生幻觉","calloutBody":"Whisper 的解码器像语言模型一样工作，因此在静音或噪声中可能生成流畅但完全虚构的句子；整个生态都记录过这种故障。Parakeet 和 Nemotron 等转导器在结构上更能抵抗它：没有音频依据，就没有 token。对于开头和结尾常有静音的听写来说，这比基准中的一个百分点更重要。"},"speed":{"title":"速度与效率：完全不是一个级别","paragraph1":"在 Leaderboard 使用的硬件上，Parakeet TDT v3 的转录速度为实时的 3,332×，约一秒即可处理一小时音频。Whisper large-v3 的速度远不到它的十分之一。原因在于架构：Whisper 逐个 token 写出文本，每一步都要等待上一步；转导器只读取一次音频，并随进度输出文字。没有循环，也没有逐词往返处理。","paragraph2Prefix":"在笔记本电脑上，绝对数字会下降，但比例依然成立：Whisper large 需要数秒处理的一段听写，Parakeet 用 CPU、无需 GPU 就能近乎瞬间完成。独立基准还显示，在质量相当时，NVIDIA 转导器每小时音频大约少用","energyLink":"8–10× 的能耗","paragraph2Suffix":"；如果经常用电池长时间听写，这一点很实际。","paragraph3Prefix":"Whisper 的应对方案是 turbo（1.6GB），它是 large-v3 的蒸馏版本，以少量准确率换取约 8× Whisper-large 的速度。它是适合听写的 Whisper，但只能缩小差距，无法追平。完整的型号大小对比请参阅","guideLink":"Whisper 模型大小指南","paragraph3Suffix":"。"},"languages":{"title":"语言覆盖：Whisper 的护城河","headers":{"model":"模型","languages":"语言","notes":"说明"},"rows":[{"model":"Whisper large-v3","languages":"99","notes":"开放模型中覆盖最广，包括低资源语言"},{"model":"Nemotron 3.5 ASR","languages":"40 个语言区域","notes":"自动语言检测；15 种语言可直接转录，包括日语、韩语、阿拉伯语和印地语"},{"model":"Parakeet TDT v3","languages":"25","notes":"欧洲语言以及英语和俄语；自动检测"},{"model":"Parakeet Unified / Nemotron EN","languages":"1","notes":"仅支持英语，以换取同类最佳准确率"}],"rule":"经验法则很简单。英语或主要欧洲语言：三者都可用，按速度选择。日语、韩语、越南语、阿拉伯语和印地语：Nemotron 3.5 如今可提供流式识别，这是本地模型真正的首次突破。普通话、泰语、印度尼西亚语、斯瓦希里语或大量语言切换：Whisper 仍是唯一严肃的选择，短期内不太可能改变。"},"streaming":{"title":"流式识别：真正的分水岭","lead":"在流式识别方面，Nemotron 不只是小幅领先，而是在完成另外两者受架构限制无法完成的事。","paragraph1":"Whisper 以 30 秒窗口处理音频，并重新编码看到的全部内容。因此所谓“实时”Whisper，其实是把同一个批处理模型反复运行在相互重叠的音频上，成本高、延迟大，接缝处也不稳定。Nemotron 的缓存感知架构专为流式场景训练：它只处理最新音频块，延续内部状态，并在可配置的 80 毫秒至 1.12 秒延迟范围内输出部分文字。","paragraph2Prefix":"在大规模部署中，效率差异非常显著：NVIDIA 报告同一 GPU 上可实现比上一代流式模型","sourceLink":"多 17× 的并发流","paragraph2Middle":"。而在笔记本电脑上，您感受到的优势更直观：模型在本机 CPU 上运行，文字随说话出现。我们在","deepDiveLink":"流式 ASR 技术深入解析","paragraph2Suffix":"中详细介绍了将其用于桌面应用所需的工程工作。","paragraph3":"如果不在乎实时文字，只是说完后获取最终转录，那么流式识别没有额外价值，Parakeet 批处理模型更合适。流式识别适用于实时字幕、语音智能体以及边说边显示的听写预览。"},"hardware":{"title":"硬件：三者都能在普通笔记本上运行","paragraph1":"这些模型都不需要 GPU。NVIDIA 模型以 INT8 量化 ONNX 文件（631–680MB）提供，通过无需 Python 和 PyTorch 的原生二进制运行时 sherpa-onnx，可在任何较新的 CPU（Apple Silicon 或 x86）上流畅运行。Whisper 通过 whisper.cpp 可在各平台运行；small 和 turbo 适合 CPU，large-v3（3GB、~10GB RAM）则确实能从加速中获益。","paragraph2":"如果有 GPU，建议用于 Whisper：Apple Silicon 内置 Metal，OpenWhispr 一键提供 CUDA（NVIDIA）和 Vulkan（AMD 与 Intel）运行时，并在 GPU 运行失败时自动回退到 CPU。NVIDIA 转导器在 CPU 上已经足够快，对听写而言 GPU 加速意义不大。"},"decision":{"title":"应该选哪一个？","headers":{"situation":"您的需求","use":"建议使用"},"rows":[{"situation":"英语听写，追求最高准确率","use":"Parakeet Unified EN 0.6B"},{"situation":"欧洲语言，快速批量转录","use":"Parakeet TDT 0.6B v3"},{"situation":"说英语时实时显示文字","use":"Nemotron Speech Streaming EN"},{"situation":"以西班牙语、日语、韩语、阿拉伯语、印地语等实时显示文字","use":"Nemotron 3.5 ASR Streaming"},{"situation":"NVIDIA 模型未覆盖的语言","use":"Whisper large-v3 或 turbo"},{"situation":"旧设备或内存较少，记录简短笔记","use":"Whisper tiny 或 base"}],"privacyPrefix":"如果隐私是您选择本地方案的原因，三者同样合格：音频在本机处理，绝不上传。我们在","privacyLink":"本地转录与云端转录对比","privacySuffix":"中进一步讨论了这种取舍。"},"cta":{"title":"两分钟试完三个模型","paragraphPrefix":"最客观的选择方法，是用每个模型听写同一段话。OpenWhispr 在 macOS、Windows 和 Linux 上免费开源，并通过一个设置下拉菜单提供本文全部模型：六种大小的 Whisper、两款 Parakeet 和两款 Nemotron。下载模型，按下快捷键，然后说话。我们的","modelsLink":"模型页面","paragraphSuffix":"列出了全部模型的大小和取舍。","github":"在 GitHub 上加星"},"faqSection":{"title":"常见问题"}},"blogHipaaNoteTakers":{"metadata":{"title":"符合 HIPAA 的 AI 会议记录与听写软件（2026） | OpenWhispr","description":"哪些 AI 会议记录工具会签署 BAA，哪些又把 AI 功能排除在外。2026年的清晰对比，以及我们在 SOC 2 Type II 与 ISO 27001 上的立场。","keywords":["符合hipaa的ai会议记录","符合hipaa的笔记应用","hipaa ai 医疗记录助手","符合hipaa的转录","符合hipaa的听写软件","符合hipaa的语音转文字","otter ai 是否符合 hipaa","hipaa ai 记录心理治疗师","ai 会议记录 soc 2 hipaa","符合hipaa的会议转录"]},"breadcrumb":{"home":"首页","blog":"博客","article":"符合 HIPAA 的 AI 会议记录工具"},"schema":{"title":"2026年符合 HIPAA 的 AI 会议记录工具与听写软件","description":"哪些厂商会签署业务伙伴协议，哪些把 AI 功能排除在外，签约前该问什么，以及本地转录如何改变整个判断。"},"hero":{"title":"符合 HIPAA 的 AI 会议记录工具：BAA 究竟覆盖什么","subtitle":"多数自称「符合 HIPAA」的工具确实会签 BAA，然后把 AI 功能留在外面。以下是在患者数据交给厂商之前的核查方法，以及我们在 SOC 2 Type II、ISO 27001 和 HIPAA 上的立场。","date":"2026年8月11日","tag":"合规"},"sections":[{"id":"where-we-stand","title":"我们的立场：SOC 2、ISO 27001、HIPAA"},{"id":"what-hipaa-requires","title":"HIPAA 实际要求什么"},{"id":"ai-exclusion","title":"陷阱：当 BAA 排除了 AI"},{"id":"comparison","title":"哪些工具会签署 BAA"},{"id":"checklist","title":"向厂商提出的五个问题"},{"id":"local-processing","title":"本地处理为何改变这道算式"},{"id":"frameworks","title":"SOC 2 和 ISO 27001 不等于 HIPAA"},{"id":"faq","title":"常见问题"}],"opening":{"lead":"并不存在「通过 HIPAA 认证」的 AI 会议记录工具。","body":"没有任何机构会把软件认证为符合 HIPAA（美国《健康保险流通与责任法案》）。当厂商网站写着「已通过 HIPAA 认证」时，这句话出自市场团队，而不是审计师。真正重要的问题更为具体，也容易核实得多：他们是否签署业务伙伴协议（BAA），而该协议是否覆盖你打算用在患者数据上的那项功能？","thesis":"团队栽跟头的正是后半句。几家大型平台会为核心产品签下 BAA，却把 AI 功能留在外面——而那恰恰是你为之而来的功能。","updated":"最后更新于 2026年8月。厂商条款变动很快，在向厂商发送任何内容之前请先确认当前的覆盖范围。"},"whereWeStand":{"title":"我们的立场：SOC 2、ISO 27001、HIPAA","disclosure":"OpenWhispr 是我们做的，它也在下面的表格里。这是利益冲突，所以我们先把自己的立场讲清楚：请带着这一点阅读本页其余内容。","body":"我们是一支小团队。我们比同等规模的公司更早完成了 SOC 2 Type II、ISO 27001 和 HIPAA 就绪度评估，因为当有人问患者音频去了哪里时，「相信我们」不是一个答案。具体来说，目前存在的是这些：","items":[{"lead":"SOC 2 Type II 与 ISO 27001。","body":"均由独立机构评估，66 项控制措施持续监控，13 家次级处理方公开列出。最新状态见我们的信任中心。"},{"lead":"HIPAA，附带 BAA。","body":"自 Business 套餐起可应请求提供。我们并非「通过 HIPAA 认证」，因为这种证书根本不存在，对声称如此的人应当保持怀疑。"},{"lead":"第三方渗透测试。","body":"由外部机构执行。报告可应客户请求共享。"},{"lead":"一个报告问题的渠道。","body":"security@openwhispr.com，48 小时内确认收到。公布一个地址，只有在有人回复时才有意义。"},{"lead":"本地模式，它比上述任何一项都重要。","body":"选择本地 Whisper 模型，音频就在你自己的机器上完成转录，从不外传。既然录音从未到达我们这里，也就没有什么需要 BAA 去覆盖。"}],"limitsLabel":"这些认证不能说明的事。","limitsBody":"审计是底线，不是保证。它说明一家公司有控制措施并加以执行，却没有说明某项具体功能是否契合你的工作流程。OpenWhispr 通过全局快捷键可以在任何电子病历字段中听写，因此 Epic、Cerner、athenahealth 和网页版病历都无需集成即可使用。它不做的是通过 API 回写：没有 FHIR 或 HL7 集成，不知道当前打开的是哪位患者的病历，也不产生编码化或结构化数据。它同样不是经过临床验证的环境记录助手。如果你需要把记录归档到某次具体就诊并附带编码建议，专门的记录助手产品会比我们更合适。","trustLine":"SOC 2 报告、ISO 27001 证书、次级处理方名单和受监控的控制措施都发布在我们的<trustLink>信任中心</trustLink>，我们如何处理数据则写在<securityLink>安全页面</securityLink>。"},"whatHipaaRequires":{"title":"HIPAA 对 AI 会议记录工具的实际要求","intro":"如果厂商代表你创建、接收、存储或传输受保护的健康信息（PHI），HIPAA 就把它视为业务伙伴。由此产生一份简短而具体的清单：","items":[{"lead":"一份已签署的 BAA。","body":"这份合同让厂商对你交付的 PHI 承担法律责任。没有它，向厂商发送 PHI 本身就是违规，无论加密做得多好。"},{"lead":"安全规则要求的保护措施。","body":"访问控制、审计日志、传输与静态加密，以及书面的事件响应流程。"},{"lead":"泄露通知。","body":"他们必须不得无故拖延地告知你未受保护 PHI 的泄露，好让你赶上自己的期限。"},{"lead":"最小必要原则。","body":"只能按 BAA 允许的方式使用 PHI。这排除了二次用途，其中最需要盯住的就是用你的患者数据训练 AI 模型。"},{"lead":"向分包方传递义务。","body":"如果你的音频转到第三方模型提供商，那就是分包方，需要签署自己的 BAA。问清楚链路上都有谁。"}],"mythLabel":"关于认证的迷思。","mythBody":"HIPAA 没有认证机构，所以「通过 HIPAA 认证」并不是一个可以核实的说法。诚实的表述是「符合 HIPAA」「我们支持 HIPAA 合规」或「我们签署 BAA」。宣传证书的厂商，要么是在简化「我们完成了第三方就绪度评估」，要么就是没读懂法规。无论哪种，请索取 BAA，而不是那枚徽章。"},"aiExclusion":{"title":"陷阱：一份排除了 AI 的 BAA","intro":"2026年常见的失误不是厂商拒签 BAA，而是签了一份悄悄不覆盖 AI 的 BAA。该品类中两家最大的平台正是如此：","zoomLabel":"Zoom。","zoomBody":"Zoom 为 Zoom for Healthcare 以及符合条件的 Business 和 Enterprise 套餐提供 BAA，覆盖 Meetings、Webinars、Phone、Team Chat 和 Rooms。但启用 HIPAA 模式会停用一部分 AI Companion 功能，剩下的功能中也只有智能录制、会议摘要等少数落在 BAA 范围内。合规的那套配置反而能力更弱。","zoomSource":"Zoom 的 HIPAA 文档","notionLabel":"Notion。","notionBody":"Notion 会签署 BAA，但仅限席位数超过下限的 Enterprise 客户，且 Notion AI 在所有套餐中都在该覆盖范围之外。你可以把 PHI 存进 Notion 数据库。但不能对它使用 Notion AI。","notionSource":"Notion 的 HIPAA 帮助页面","takeaway":"这个模式反复出现，与这些产品的构建方式有关。合规团队审查平台，AI 稍后以另一套法律前提上线，而这项排除从未出现在销售对话中。所以请以书面形式、逐项功能地索取 BAA 的覆盖范围，而不是就整家公司要一个「是」或「否」。"},"comparison":{"title":"哪些 AI 会议记录工具会签署 BAA","intro":"2026年8月的状况，依据各厂商公开文档。最后一栏最关键，也是定价页面最少回答的一栏。","headers":{"tool":"工具","baa":"是否签署 BAA","plan":"所需套餐","aiCovered":"是否覆盖 AI 功能"},"vendors":[{"name":"OpenWhispr","baa":"是，可应请求提供","plan":"Business 套餐及以上","aiCovered":"覆盖，或 PHI 不离开设备","note":"自 Business 套餐起即可应请求提供 BAA，比 Otter 和 Fathom 设置的「仅限 Enterprise」门槛低一档。在本地模式下这个问题根本不会出现：转录使用开源 Whisper 模型在你的机器上运行，音频从不发送给我们。开发 OpenWhispr 的 Gizmo Labs 持有 SOC 2 Type II 和 ISO 27001，并运行 HIPAA 与 GDPR 合规项目。","href":null},{"name":"Otter.ai","baa":"是","plan":"仅限 Enterprise","aiCovered":"覆盖（在 BAA 范围内）","note":"Otter 在独立评估后于 2025年7月宣布符合 HIPAA，并签署覆盖 PHI 使用与披露的 BAA。只有 Enterprise 符合条件。Basic、Pro 和 Business 客户无法获得，因此升级到合规层级意味着实打实的价格跳涨。","href":"/compare/otter"},{"name":"Fathom","baa":"是","plan":"仅限 Enterprise","aiCovered":"覆盖（在 BAA 范围内）","note":"在 Enterprise 层级提供有 BAA 支撑的 HIPAA 合规。其模型提供商 Anthropic、OpenAI 和 Google 在合同上被禁止使用用户数据训练，这正是对分包方问题的正确回答。","href":"/compare/fathom"},{"name":"Fireflies.ai","baa":"是，可应请求提供","plan":"医疗层级","aiCovered":"覆盖（在 BAA 范围内）","note":"Fireflies 于 2025年9月推出面向医疗行业的方案，包含符合 HIPAA 的层级。BAA 不会自动生效：默认套餐不在覆盖范围内，必须主动申请。Fireflies 表示已与自身供应商签署 BAA，禁止其训练或存储 ePHI。","href":"/compare/fireflies"},{"name":"Zoom AI Companion","baa":"部分覆盖","plan":"Healthcare / 符合条件的套餐","aiCovered":"仅部分功能","note":"BAA 覆盖 Zoom 的核心产品，但 HIPAA 模式会关掉 AI Companion 的一系列功能，只有一部分仍在范围内。在依赖它撰写临床文档之前，请逐项确认。","href":null},{"name":"Granola","baa":"否","plan":"—","aiCovered":"不覆盖","note":"Granola 目前既不提供 HIPAA 合规，也不提供 BAA。产品不错，但在这一点改变之前不能用于 PHI。","href":"/compare/granola"},{"name":"Notion AI","baa":"部分覆盖","plan":"Enterprise，100 个席位以上","aiCovered":"不覆盖——AI 被排除","note":"Notion Enterprise 会签署覆盖页面、数据库、Wiki 和文件上传的 BAA，但 Notion AI 在所有套餐中都在其外。存放文档尚可，作为处理 PHI 的 AI 记录助手则不行。","href":null}],"compareLink":"对比 OpenWhispr 与 {name}","disclaimer":"根据 2026年8月各厂商公开文档整理。套餐名称、层级和 BAA 覆盖范围经常变动，请把本表当作自行尽职调查的起点，而非替代。请向厂商及贵方合规顾问核实当前条款。"},"checklist":{"title":"记录 PHI 之前该问的五个问题","intro":"多数安全页面只回答容易的问题，跳过那些决定你能否真正使用该产品的问题。以下五个请以书面形式提出：","questions":[{"lead":"你们签 BAA 吗？需要哪个套餐？","body":"如果答案是「仅限 Enterprise」，在对产品动感情之前先问价。合规层级的费用常常是最初报价的数倍。"},{"lead":"BAA 有没有点名 AI 功能？","body":"不是问「产品是否被覆盖」，而是问转录、摘要以及任何 AI 聊天或智能体功能是否各自在范围内。Zoom 和 Notion 与自家宣传出现落差的地方正在于此。"},{"lead":"哪些次级处理方会拿到 PHI？","body":"如果音频流向 OpenAI、Anthropic、Google 或某个托管语音服务，它们各自都是分包方，需要各自的 BAA 和不用于训练的承诺。请索取名单，而不是一句宽慰。"},{"lead":"我们的数据会用于训练或产品改进吗？","body":"「我们不出售你的数据」是另一回事，而且是弱得多的承诺。你要的是合同层面禁止训练、禁止人工查看你的录音。"},{"lead":"音频保留多久？能否设为零？","body":"实时处理并立即删除，和保留九十天再给一个删除按钮，完全不是一回事。无法配置的保留期，就是将来要在审计中辩解的保留期。"}]},"localProcessing":{"title":"本地处理为何改变这道算式","intro":"上面所有问题之所以存在，都是因为 PHI 离开了你的机构、落在别人的服务器上。当音频从不离开设备，绝大部分分析就不再适用：","points":["不存在向业务伙伴的披露，因为厂商从未收到 PHI。","没有需要审计的分包方链条，因为路径上没有第三方模型提供商。","没有需要谈判的保留策略，因为服务器端不存在副本。","不会因厂商日后改政策而受影响，因为模型跑在你自己掌控的硬件上。","无需联网即可运行，这在临床设备被刻意保持离线的机构里很重要。"],"tradeoff":"取舍是真实存在的，值得说明。本地模型需要一台还算新的机器，最大最准确的那些在旧硬件上会更慢，而且你放弃了让机器人代你参会的便利。对许多临床听写来说这笔账划算。对大型诊所的大批量环境记录来说，未必如此。","closing":"OpenWhispr 正是围绕这一点设计的，也正因如此，我们认为本地这个选项比我们持有的任何证书都更重要。"},"frameworks":{"title":"SOC 2 和 ISO 27001 不等于 HIPAA","intro":"这三者在采购沟通中常被混为一谈，但它们回答的确实是不同的问题。分清楚能为安全审查的双方省下时间：","items":[{"lead":"SOC 2 是一份报告，不是证书。","body":"由独立审计师依据 Trust Services Criteria 出具。Type I 看的是某个时点上控制措施如何设计。Type II 检验它们在一段时间内是否真的有效，通常为三到十二个月，值得索取的是后者。自称「通过 SOC 2 认证」的人，描述的其实是一份根本不以这种方式运作的文件。"},{"lead":"ISO 27001 确实是一项认证。","body":"由认可机构依据标准对组织的信息安全管理体系进行认证，三年为一周期并配合监督审核。请索取证书编号、颁发机构，以及最关键的适用范围声明，它说明实际覆盖了什么。"},{"lead":"HIPAA 是一部法律，根本没有证书。","body":"合规是一项持续义务，通过保护措施、政策、风险分析和已签署的 BAA 来证明。SOC 2 和 ISO 27001 能有力说明厂商认真对待安全。但两者都不能替代 BAA。"}]},"faq":{"title":"常见问题","legalNote":"这只是关于各厂商合规安排的一般性信息，不构成法律意见。你在 HIPAA 下的义务取决于贵机构的情况、风险分析以及法律顾问的判断。"},"faqs":[{"question":"Otter.ai 符合 HIPAA 吗？","answer":"符合，但仅限 Enterprise 套餐。Otter 在独立评估后于 2025年7月宣布符合 HIPAA，并签署覆盖 PHI 使用与披露的 BAA。Basic、Pro 和 Business 客户无法获得 BAA，因此在这些层级上并不符合。如需启动流程，请联系 Otter 的销售团队。"},{"question":"Fathom 符合 HIPAA 吗？","answer":"符合，在 Enterprise 层级并有 BAA 支撑。Fathom 的 AI 提供商 Anthropic、OpenAI 和 Google 在合同上被禁止使用用户数据训练。较低层级不在覆盖范围内，因此在录制任何临床内容之前，请确认你的套餐符合条件。"},{"question":"Notion 符合 HIPAA 吗？","answer":"部分符合。Notion 为席位数超过下限的 Enterprise 客户签署 BAA，覆盖页面、数据库、Wiki 和文件上传。但 Notion AI 在所有套餐中都被排除在该覆盖之外，因此即便你是已签 BAA 的 Enterprise 客户，也不能把 Notion 的 AI 功能用在受保护的健康信息上。"},{"question":"Zoom AI Companion 符合 HIPAA 吗？","answer":"只是部分符合。Zoom 会为 Zoom for Healthcare 以及符合条件的 Business 和 Enterprise 套餐签署 BAA，但启用 HIPAA 模式会停用多项 AI Companion 功能。智能录制、会议摘要等在范围内，其余则不在。在用于临床文档之前，请确认你需要的具体功能确实被覆盖。"},{"question":"Fireflies 符合 HIPAA 吗？","answer":"在其医疗层级上符合。Fireflies 于 2025年9月推出符合 HIPAA 的方案，但默认套餐不在覆盖范围内，必须明确申请 BAA。Fireflies 还表示已与自身次级处理方签署 BAA，禁止其训练或存储 ePHI。"},{"question":"Granola 符合 HIPAA 吗？","answer":"不符合。Granola 目前既不提供 HIPAA 合规，也不提供业务伙伴协议，因此不应用于记录或处理受保护的健康信息。"},{"question":"心理治疗师最适合用哪款符合 HIPAA 的 AI 记录工具？","answer":"这取决于你是否需要会议机器人。个人执业和小型诊所往往更适合设备端听写：会谈音频不离开笔记本电脑，这一步也不需要 BAA。如果需要能加入远程问诊通话的机器人，Otter Enterprise、Fathom Enterprise 和 Fireflies 的医疗层级都会签署 BAA。无论选哪个，都请查看保留设置和次级处理方名单，因为心理治疗记录属于 HIPAA 涵盖范围内最敏感的资料。"},{"question":"如果转录在我自己的电脑上本地运行，还需要 BAA 吗？","answer":"就该步骤而言通常不需要。如果音频完全在设备上转录、从未发送给厂商，那么厂商并没有代表你创建、接收、存储或传输 PHI，因此在这条链路中不充当业务伙伴。但你自己的保护措施依然必要：设备加密、访问控制，以及关于生成记录存放位置的政策。请向贵方合规顾问确认这一判断。"},{"question":"SOC 2 或 ISO 27001 是否意味着某款工具符合 HIPAA？","answer":"不意味着。SOC 2 是审计师关于安全控制的报告，ISO 27001 认证的是信息安全管理体系。两者都能有力说明安全工作是认真的，但都无法建立 HIPAA 所要求的合同关系。只有已签署的 BAA 才能做到。"},{"question":"「符合 HIPAA」和「通过 HIPAA 认证」有什么区别？","answer":"「通过 HIPAA 认证」不是一种真实存在的状态，因为没有任何政府机构或认可机构会依据 HIPAA 认证软件。使用这一说法的厂商通常指的是完成了第三方就绪度评估。你真正能核实的是：他们是否签署 BAA、该 BAA 覆盖什么，以及记录了哪些保护措施。"},{"question":"有没有免费且符合 HIPAA 的 AI 会议记录工具？","answer":"OpenWhispr 的本地模式免费且开源，由于音频从不离开你的设备，也就没有需要覆盖的披露行为。如果你想改用我们的云服务，自 Business 套餐起可签署 BAA。多数云端记录工具都把 BAA 绑定在 Enterprise 定价上，因此免费的云端层级几乎无法用于 PHI。在认定免费选项可行之前，请先确认所需套餐。"},{"question":"AI 会议记录工具会用患者数据训练吗？","answer":"在拟定得当的 BAA 之下不会。协议应当明确禁止厂商及其每一个次级处理方将 PHI 用于模型训练或产品改进。请要求把该条款写进书面文件。隐私政策里关于不出售数据的一句话，远比不上合同层面的训练禁令。"}],"cta":{"title":"留在你机器上的听写与会议记录","body":"OpenWhispr 在本地运行开源 Whisper 模型，你的音频无需离开设备。若确实需要我们的云服务，自 Business 套餐起可签署 BAA。免费、开源，支持 macOS、Windows 和 Linux。","github":"在 GitHub 上点亮星标"}},"useCaseLawyers":{"breadcrumb":{"home":"Home","useCases":"Use Cases","lawyers":"For Lawyers"},"metadata":{"title":"Free Dictation Software For Lawyers & Attorneys - OpenWhispr","description":"Dictation software for lawyers. OpenWhispr offers Cloud transcription or local processing for sensitive and offline work.","keywords":["legal dictation software","lawyer dictation","attorney dictation software","legal transcription","dictation for law firms","offline legal dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Legal dictation software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source dictation software for lawyers. Choose Cloud transcription without a local model download, or local processing for sensitive and offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation software for lawyers & attorneys","title":"Speak the letter. It lands in Word. No typist ever hears the file.","body":"Dictate a letter, a file note or a memo and the clean text drops in wherever your cursor already is: Word, Clio, an email. Choose Cloud for fast transcription or local for highly sensitive and offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Man speaking beside an open laptop on a sunny terrace"}},"trustBar":{"label":"Dictate into any app, including:","names":["Clio","iManage","NetDocuments","MyCase","Smokeball","Microsoft Word"]},"pain":{"label":"What you're choosing between","title":"Saying the letter takes two minutes. Getting it typed is the rest of the afternoon.","subtitle":"Every route from your voice to the file should suit the matter, not keep you typing.","cards":[{"title":"A stranger typing up privileged notes.","body":"With a local model and no Cloud fallback, the dictation is transcribed on the machine you spoke into. No outsourced transcription service handles it."},{"title":"Sensitive matters need a local option.","body":"Choose local models when the matter is highly sensitive or you will work without signal. Your device processes the recording."},{"title":"An hour of typing is an hour you can't bill.","body":"Typing up your own dictation is the slowest way to get a letter out, and you do it after the thinking is already done. Dictate it while the matter is fresh and you review and send instead of type."},{"title":"A tool that only fits one desk.","body":"OpenWhispr runs on macOS, Windows, and Linux. Start with the Free plan, then upgrade when the workflow calls for it."}]},"features":{"label":"From your voice to the file","title":"From your voice to a first draft.","subtitle":"Get the letter out while the matter is fresh. Choose the transcription path that fits the work.","items":[{"image":{"label":"Woman speaking while using a laptop at a desk"},"title":"Cloud transcription without a local model download","body":"Speak a letter, a memo or a file note into your mic. Cloud gives you fast transcription without a local model download, then immediately discards the transcription audio."},{"image":{"label":"Man reviewing a printed document beside his laptop"},"title":"Letters and memos, drafted for you","body":"Tell it what you want and use local AI models for device-only drafting. You read every word before it leaves your desk."},{"image":{"label":"Man speaking and pointing toward a laptop at a table"},"title":"Dictate into Word without an add-in","body":"One hotkey, and the clean punctuated text pastes itself where you're already typing: Word, an email, Clio, your document system. There's nothing for IT to install into any of them."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Privileged audio sent to us","value":"None","footnote":"with a local transcription model the recording never leaves your computer"},{"label":"Cost per lawyer","value":"$$0","footnote":"unlimited local dictation in the Free plan"},{"label":"Processing options","value":"2","footnote":"Cloud transcription or local models"}]},"howItWorks":{"label":"Download to first letter","title":"Three steps from download to draft.","subtitle":"Choose Cloud transcription or download a local model.","image":{"label":"Man standing beside a laptop in a glass-walled office"},"steps":[{"title":"Download it free.","body":"Mac, Windows or Linux. Setup offers OpenWhispr Cloud or local models. Pick local for privileged work and it downloads them once."},{"title":"Dictate the letter.","body":"Speak it the way you'd say it to your assistant, and it comes back as text."},{"title":"Read it, paste it, delete the audio.","body":"Edit the draft, drop it into Word or your document system, and delete the recording. Every one of those is yours to control."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Why firms keep it","title":"Ten reasons your IT or security team can check for themselves.","subtitle":"Choosing dictation software for lawyers is a confidentiality decision before it's anything else. Most of these are facts about the code, so IT or security can check them instead of taking them from us.","items":[{"title":"The recording stays on your machine","body":"Choose a local transcription model for device-only processing. An optional Cloud fallback can send audio if you turn it on."},{"title":"Free for the whole firm","body":"Unlimited local dictation at $0 for every desk, including your assistant and paralegals."},{"title":"Nothing goes to a typing pool","body":"With local transcription and no Cloud fallback, no outsourced transcription service handles the recording."},{"title":"Works with the wifi off","body":"A courthouse hallway, a deposition, a client's office, a flight. Local models need no signal, so nothing stalls mid-sentence."},{"title":"Dictate into Word or any case system","body":"One system-wide hotkey drops clean text where your cursor already is: Word, Clio, iManage, an email. On a Mac it uses the accessibility permission you grant at setup."},{"title":"Letters and memos, not just transcripts","body":"Ask for a case memo and a local AI model can draft it on your machine. Local AI models are included on every plan."},{"title":"You control storage and deletion","body":"Recordings are files on your machine. Audio is deleted after 30 days out of the box, and Settings moves that or switches it off."},{"title":"Nothing to take on faith","body":"Your IT or security team can read every line of the app on GitHub before a partner installs it."},{"title":"Party names and citations, spelled your way","body":"Put party names, case names, statutes, and citations in the custom dictionary to guide transcription and cleanup."},{"title":"100+ languages","body":"Dictate in 100+ languages."}]},"faq":{"label":"Before you install it","title":"Eight questions law firms ask before they install anything.","subtitle":"Straight answers on confidentiality, cost, accuracy, and what IT will say. More questions? Ask in Discord.","items":[{"question":"Does the audio ever leave my device?","answer":"With local transcription, audio is transcribed on your device. Turn off the optional Cloud fallback when every attempt must stay local. Cloud sends audio only to transcribe it, then immediately discards it."},{"question":"Is this consistent with my duty of confidentiality?","answer":"Confidentiality is your professional duty and no tool discharges it for you. Local processing is the deliberate choice when a highly sensitive matter calls for device-only work. Cloud transcription discards audio immediately after transcription. This is not legal advice, and the rules of your bar or law society still apply."},{"question":"Does dictation software for lawyers get case names and citations right?","answer":"Add party names, case names, Latin, statutes, and citations to the custom dictionary. OpenWhispr passes that list to transcription and cleanup to guide the text."},{"question":"Will IT let me install it on the firm laptop?","answer":"It is a desktop app, not an add-in for Word or your document system. On a Mac, grant accessibility permission for auto-paste. Without it, copy the text from your clipboard instead."},{"question":"Will it work in a courthouse or on a flight with no signal?","answer":"Yes, once the models are on your machine. Local models run entirely offline after the first download. Dictate in the courthouse hallway, after a deposition, or on the flight home, then file the note when you are back in range."},{"question":"Does it draft the letter, or just transcribe?","answer":"Both. Dictation turns speech into text, then an AI model can shape a client letter or case memo. Transcription and AI cleanup are separate settings, so choose a local AI model when the draft must stay on-device."},{"question":"Do I need a separate legal dictation tool?","answer":"OpenWhispr runs on macOS, Windows, and Linux. Try the Free plan with a local model, then move to Cloud transcription when it fits your workflow."},{"question":"Is it actually free?","answer":"Free includes unlimited local dictation and local AI models, plus 2,000 Cloud words a week. Pro adds unlimited Cloud transcription, 20 hours of meeting recordings, sync, mobile, and API/MCP access. Business adds unlimited meeting recordings, Agent mode, and chat over data."}]},"reviews":{"label":"What people tell us","title":"Built by people who wouldn't hand their audio to a stranger either.","subtitle":"From our Discord, GitHub and inboxes. None of them is a lawyer, so read them for the tool, not the practice.","score":{"value":"0","label":"Audio stored on our servers with local transcription","body":"With a local transcription model the audio never leaves your device, so there's nothing on our side to lose, breach or hand over."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free to start.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for lawyers","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Lawyers who want local dictation and 2,000 Cloud words a week"}]},{"parts":[{"text":"Lawyers who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Firms documenting at volume with Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local dictation plus 2,000 Cloud words a week"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Enterprise adds SSO, audit logs, retention controls, and dedicated support."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"A client tells you things they'd tell nobody else.","paragraphs":["OpenWhispr started with a simple objection: confidential dictation should have a local option.","We build in public. The app is open source, so your firm can inspect how it works before anyone installs it.","Community support is available in Discord. Paid plans add email or priority support."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"$$0 a seat · open source · no add-in","title":"Speak the next letter while the matter is fresh.","body":"Download OpenWhispr, choose Cloud transcription or local processing, and dictate the next letter.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free on macOS, Windows, and Linux. Paid plans add Cloud transcription, mobile, and workflow tools."}}},"useCaseLegalDocuments":{"breadcrumb":{"home":"Home","useCases":"Use Cases","legalDocuments":"Legal Documents"},"metadata":{"title":"Free AI Dictation For Legal Documents - OpenWhispr","description":"AI dictation for legal documents: talk a clause, a brief section or a motion through and OpenWhispr's agent drafts it. Free forever, and it runs locally.","keywords":["ai dictation for legal documents","dictate a contract","brief and motion dictation software","discovery response drafting tool","deposition summary dictation","paralegal and legal ops dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Legal document drafting software","operatingSystem":"macOS, Windows, Linux","description":"OpenWhispr is an open-source dictation app for the people who produce legal documents: paralegals, legal assistants and lawyers. They dictate contracts, briefs, motions, discovery responses and deposition summaries, and the AI agent redrafts them into structured text, either in place in the document or in the app's assistant panel.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"AI dictation software for legal documents","title":"Stop typing first drafts. Talk the section through.","body":"Talk a contract clause, a motion section or a discovery response through, then tell the AI agent what to make of it: headings, numbered points, a summary. Pick a local model and it's drafted on your own machine, free and unlimited on every plan.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman speaking beside a laptop at an office desk"}},"trustBar":{"label":"Dictate:","names":["Contract clauses","Motion sections","Discovery responses","Deposition summaries","Case memos","Engagement letters"]},"pain":{"label":"Why dictating a contract stalls","title":"Talking it through was never the slow part. Shaping it is.","subtitle":"Speak an engagement letter or a case chronology and you'll get every word back in one unbroken block. Cutting that block into clauses, headings and numbered points is what eats the afternoon, and plain dictation hands that part straight back to you.","cards":[{"title":"One block of speech, not a clause in it.","body":"Dictation cleanup hands back what you said, tidied. It will break the speech into paragraphs and number a list you clearly dictated. What it won't do is impose a heading or restructure a deposition summary into sections. That job belongs to the AI agent, and you have to ask it for it."},{"title":"The motion is the sensitive part, not the audio.","body":"A transcript is words. A motion is the argument you're about to run and the position your client will take. Choose local processing when you want the composing to run on your machine, or cloud tools when they fit your workflow."},{"title":"Paralegals and legal ops draft it too. Per-seat pricing adds up.","body":"An associate dictates the contract, a paralegal checks the cites, an assistant cleans it up, legal ops files it. Charge per seat and the cost follows the draft across four desks. OpenWhispr charges nothing for a seat on Free, so each of them can use it."},{"title":"Party names and statutes come back mangled.","body":"A general-purpose model has never seen your matter. Put the case names, the parties, the Latin and the citations your practice uses in the custom dictionary. A dictation vocabulary only fixes what was heard; this one also reaches the model that drafts, so it sees those names before it writes a word. It moves the odds rather than guaranteeing them, so read what carries legal weight before the document goes out."}]},"features":{"label":"From microphone to document","title":"You dictate a section. You get back something worth editing.","subtitle":"A transcript isn't a document. Dictation, cleanup and the AI agent each hand you something different.","items":[{"image":{"label":"Woman speaking while working on a laptop"},"title":"Say it once, read it back punctuated","body":"Talk an interrogatory response through the way you'd say it out loud, and it comes back transcribed and punctuated with the filler and the false starts taken out. Address the agent instead and it works from the same dictation, handling the disfluencies itself."},{"image":{"label":"Woman reading a printed document beside her laptop"},"title":"Headings and numbered points, not a wall of speech","body":"Cleanup breaks your speech into paragraphs and numbers a list you clearly dictated, and stops there. Ask the AI agent instead, on its own hotkey or by name mid-dictation, and it composes what you asked for: the headings, the numbered points and the lists you talked through."},{"image":{"label":"Man speaking while reviewing a printed document beside a laptop"},"title":"Nothing to install into Word or your DMS","body":"One global hotkey copies the text and pastes it at your cursor, so the clause lands in Word, a matter record, an email or a web form the same way. Auto-paste is on by default, so you edit it where it lands."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Audio kept on your own machine by default","value":"30 days","footnote":"Audio Retention ships on at 30 days. Settings takes it from 1 day to 90, or switches it off"},{"label":"Languages you can dictate a matter in","value":"100+","footnote":"a German framework agreement comes back as German text"},{"label":"Cost to add the paralegal and legal ops","value":"$$0/month","footnote":"no card details, and no per-seat licence to buy"}]},"howItWorks":{"label":"How a contract gets drafted","title":"Talk through the motion, then edit what comes back.","subtitle":"Three steps from talking the argument through to editing it in the document.","image":{"label":"Man speaking beside a laptop in a book-lined office"},"steps":[{"title":"Install it and choose where it runs.","body":"Setup puts a local model and OpenWhispr Cloud in front of you. Going local takes extra model downloads, but keeps the drafting path on your machine. Once that choice is made, place the cursor where the legal text belongs or highlight the passage you want the AI agent to reshape."},{"title":"Talk the clause through, ums and all.","body":"Say the deposition summary the way you'd explain it to a colleague, ums and all. Cleanup hands the same words back with the filler and the false starts stripped out."},{"title":"Ask the agent for headings and numbered points.","body":"Highlight the rough passage, press the agent hotkey and tell it what you want, and it rewrites that passage in place in the document. With nothing highlighted the draft arrives in the app's assistant panel instead, ready to copy across. Either way it's plain markdown text rather than Word styles, so you shape it in the document itself."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Ten claims you can check","title":"What OpenWhispr does to a legal document, and where it stops.","subtitle":"What it does, what it won't do, and where the document ends up. No asterisks.","items":[{"title":"The setting that would upload a contract arrives switched off","body":"Pick a local model at setup and the audio, the transcript and the composed section all stay on the machine you spoke into. Two deliberate acts send content to us: switching on paid cloud sync, which arrives off and stores your notes, transcripts, agent conversations and your custom dictionary, party names included; and sharing a note or moving it into a team space."},{"title":"Cleanup keeps your wording and won't add a sentence you didn't say","body":"Cleanup is told to keep your voice, your wording and your intent, and never to add a sentence you didn't say. It still fixes grammar and punctuation, splits run-on sentences, drops false starts, and writes spoken numbers and dates out in standard form. That's an instruction to a model rather than a guarantee, so read an indemnity back before it goes out."},{"title":"The agent turns talk into a numbered section","body":"Say its name mid-dictation or press its hotkey, tell it what you want, and it composes that from what you talked through: headings, numbered points, a summary of the argument. Highlight a passage first and it rewrites that passage in place; with nothing highlighted the draft lands in the assistant panel. Pick local at setup and it composes on the LLM you downloaded, on your own machine. Styling stays a Word job."},{"title":"One hotkey, and no add-in for IT to approve","body":"The text goes to the clipboard and pastes itself where your cursor is. On macOS, without Accessibility permission it can copy for you but not paste."},{"title":"Know where the ceiling is before you hit it","body":"For selection edits, select 6,000 characters or fewer."},{"title":"Dictate it into a note now, shape it later","body":"Save the case chronology as a note while it's fresh, and a note action turns it into clean markdown when you come back to it."},{"title":"Recordings are files, so deleting one really deletes it","body":"Recordings are audio files on your machine you can open or delete yourself; notes and transcripts sit in a local database the app deletes rows from. Audio Retention ships on at 30 days, and Settings takes it from 1 day to 90 or switches it off. Transcripts are the other way round, kept until you set a window of your own, so set one alongside the audio. OpenWhispr doesn't encrypt that local data itself, so the folder wants the care you'd give any other matter folder."},{"title":"Send procurement the repo instead of a policy PDF","body":"OpenWhispr is open source. When IT asks what happens to a contract you dictated, they can inspect the code rather than take our word for it."},{"title":"The paralegal and the legal ops lead get the same hotkey","body":"Legal assistants and legal ops draft these documents alongside lawyers. Free is a download rather than a licence, so the person who assembles the interrogatory response has the same hotkey as the person who signs it."},{"title":"Dictate the deal in the language it was done in","body":"Whisper-class models cover 100+ languages, so the deal you did in Spanish comes back transcribed in Spanish."}]},"faq":{"label":"Before the first document","title":"What actually comes back when you dictate a contract.","subtitle":"Anything we've missed, bring it to our Discord, where the people who build OpenWhispr answer.","items":[{"question":"Which legal documents can AI dictation draft: a contract clause, a motion section, a discovery response?","answer":"All three, plus case memos, deposition summaries and file notes. Worth knowing which part does what: plain dictation hands back your own words, tidied up. The AI agent, on its own hotkey or called by name mid-sentence, drafts them into a structured section when you ask it for one. Highlight a passage first and it rewrites that passage in place; with nothing highlighted the draft arrives in the assistant panel to copy across. What you get is markdown text, so ## Heading arrives in Word as those literal characters and not a Word style. Pleading captions and court-ready tables aren't its job either, so the formatting your jurisdiction wants still happens in Word."},{"question":"Who else ends up with a copy of an unsigned contract?","answer":"Nobody, as long as you picked a local model at setup and left the cloud alone. Whisper-class transcription runs on your machine, and the composing runs on the LLM you downloaded, served by llama.cpp at 127.0.0.1, so the clause never travels. Two things send content to us, and both are deliberate acts: cloud sync, a paid setting that stays off until you switch it on, which stores your notes, your dictation transcripts, your agent conversations and your custom dictionary, party names included, in our cloud database; and sharing a note or putting it in a team space, which uploads that note on any signed-in account. One more thing if you sign in: the assistant panel can run a web search, which sends the query the model composed, not your document, to us. Stay signed out and it is never offered."},{"question":"Can it write a whole brief in one go?","answer":"For a brief, work section by section: facts, then argument, then relief."},{"question":"Will party names and citations survive a deposition summary?","answer":"Load the custom dictionary first. It holds the words your matters actually turn on: party names, case names, statutes, the Latin, the citations your practice uses. It's a hint the model sees rather than a find-and-replace, so it moves the odds a long way in your favour without promising a clean sweep. It reaches local Whisper transcription, the cleanup pass, the AI agent and translation, across 100+ languages."},{"question":"Can a paralegal take over the contract after I've dictated it?","answer":"Yes. The hotkey copies the text and pastes it wherever the cursor is, so the words land in the file the two of you already share: Word, the matter record, an email. Auto-paste is on by default. On macOS it needs Accessibility permission, or it falls back to leaving the text on your clipboard."},{"question":"What does it cost to run a matter's documents through it?","answer":"Nothing on the local path, where dictation and drafting are unlimited on every plan, Free included, and neither one is metered. The cloud side has two separate meters. On Free you get 2,000 words a week of cloud speech-to-text, and 5 hours a month of cloud meeting recording. Pro at $8 a month makes cloud speech-to-text unlimited and takes meeting recordings to 20 hours a month. Business at $16 per user makes meeting recordings unlimited and adds team features."}]},"reviews":{"label":"Straight from our users","title":"Rambling in, editable text out.","subtitle":"Nobody was asked for these. One of them stumbles mid-sentence and still gets a clean paragraph back, which is what dictating a clause actually sounds like.","score":{"value":"0","label":"Client documents sitting on our servers","body":"In local mode, signed out, with cloud sync off and nothing shared, nothing reaches us. Turn cloud sync on, a paid setting that arrives switched off, and your notes, dictation transcripts, agent conversations and custom dictionary are stored in our cloud database."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free to get started.","Upgrade when you want more."],"caption":"What each plan covers for legal document work","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Paralegals and legal teams drafting contracts, briefs and motions, who don't mind downloading a speech model and an LLM first"}]},{"parts":[{"text":"Legal assistants who'd rather dictate through OpenWhispr Cloud than manage model downloads, at the desk and on an iPhone"}]},{"parts":[{"text":"Legal ops teams handling document work across a lot of matters, and recording calls all month"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Free plan: $0."}]},{"parts":[{"text":"The cloud speech-to-text meter comes off, and device sync and the iPhone app come with it"}]},{"parts":[{"text":"Cloud meeting recordings go unlimited and team features unlock, at $16 per user per month"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud and does not require local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Read the code yourself","title":"A contract shouldn't need a round trip to get written.","paragraphs":["Turning a transcript into clauses and numbered points takes an AI model. OpenWhispr's AI agent can do that work with a model you downloaded yourself, served by the llama.cpp binary that ships inside the app, on 127.0.0.1.","You don't have to take our word for that. OpenWhispr is open source, so your risk team can inspect the code rather than read a policy about it.","Talk the section through while the argument is fresh, have the AI agent shape it into headings or numbered points, then read what carries legal weight before the document goes out."],"image":{"label":"The people who build it"}},"finalCta":{"tagline":"$$0 a seat · No card details","title":"Say the clause out loud. Get a numbered section back.","body":"Download it free, pick a local model at setup, then hand the AI agent your rough dictation. Highlight the passage and it rewrites the clause or the case memo in place, as plain markdown text. No card details, no seat to buy.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Available on macOS, Windows, and Linux. Get community support in Discord."}}},"useCaseVeterinary":{"breadcrumb":{"home":"Home","useCases":"Use Cases","veterinary":"For Vets"},"metadata":{"title":"Free Offline Veterinary Dictation Software - OpenWhispr","description":"OpenWhispr is free, open-source veterinary dictation software that turns spoken findings into SOAP notes on your device, even on a farm call with no WiFi.","keywords":["veterinary dictation software","vet dictation app","veterinary transcription","SOAP note dictation","veterinary practice software","offline vet dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Veterinary dictation software","operatingSystem":"macOS, Windows, Linux","description":"Local-first, open-source veterinary dictation software. Dictated findings are transcribed on the vet's own device and never leave it. Free for unlimited local dictation, and works fully offline.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Veterinary dictation software for vets & vet techs","title":"Say it once. Get the SOAP note.","body":"OpenWhispr turns that dictation into a SOAP note, a discharge or a client email on your own computer. No per-DVM subscription, and it still works in a barn with no signal.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Veterinarian speaking beside a dog and laptop in a clinic"}},"trustBar":{"label":"Dictate into:","names":["ezyVet","Cornerstone","Avimark","Covetrus Pulse","Shepherd","Microsoft Word"]},"pain":{"label":"The documentation squeeze","title":"Documentation shouldn't cost your evenings and $99 a vet a month.","subtitle":"You want the note written for you. You don't want a subscription per DVM, a recorder in the exam room, or a tool that dies the moment you leave WiFi.","cards":[{"title":"Charting follows you home.","body":"Turn dictated text into a structured note with a custom action."},{"title":"A per-DVM subscription is another monthly line item.","body":"The Free plan includes unlimited local AI models and 100+ language support at $0."},{"title":"No signal, no scribe.","body":"Barn calls, farm visits, a back room with one bar. A cloud scribe can stall exactly where you are, leaving you to rebuild the note from memory that night. OpenWhispr runs on your device, fully offline."},{"title":"A recorder in the euthanasia room.","body":"Some conversations shouldn't be uploaded to a stranger's server to become a note. You dictate the findings afterwards, and with a local model everything is processed on your machine. Nothing sits on the table while you talk a client through it."}]},"features":{"label":"What you get","title":"Records closed before you lock up. No per-DVM bill.","subtitle":"An AI scribe listens to the room so you never have to say it. OpenWhispr waits for you to dictate, then a local model drafts the note from a prompt you saved yourself, on your own laptop.","items":[{"image":{"label":"Veterinarian speaking beside a dog on an exam table and a laptop"},"title":"Dictated at the treatment table, transcribed on your laptop","body":"Signalment, history, exam findings, assessment and plan, straight into the mic. Pick a local model and it's transcribed on your laptop, never uploaded, and you decide what's kept and when it goes."},{"image":{"label":"Veterinarian handing paperwork to a dog owner at a clinic doorway"},"title":"One dictation, two documents","body":"Save your own SOAP, discharge and client-email prompts as one-click actions, then run whichever one the case needs. The same dictation becomes the record for the file and the instructions the client reads in the car park. You check both before either is filed."},{"image":{"label":"Veterinarian using a laptop beside a dog in a rural clinic"},"title":"Types into the PIMS field you're already in","body":"Press a hotkey and dictate into the Cornerstone note field, an ezyVet consult tab, a discharge template or a referral email to the specialist. Clean, punctuated text, processed locally."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Client conversations uploaded","value":"None","footnote":"in local mode, not the estimate conversation, not the goodbye"},{"label":"Cost per DVM, per tech","value":"$$0","footnote":"unlimited local dictation, including the relief vet who works Saturdays"},{"label":"Setup takes","value":"No fixed time","footnote":"Install the app, then choose a local model or OpenWhispr Cloud."}]},"howItWorks":{"label":"How it works","title":"Dictate on the walk back to treatment.","subtitle":"This is one app on the practice laptop, with no PIMS migration or IT project.","image":{"label":"Veterinarian carrying a tablet past a dog in a treatment room"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Pick a local model during setup."},{"title":"Dictate the case.","body":"Signalment, presenting complaint, exam findings, assessment and plan. In the room, in the treatment area, or in the truck between farms. It's transcribed on your device, and one click turns it into the draft."},{"title":"Review, paste, close the record.","body":"Fix what it got wrong, drop it into the patient's record in ezyVet or Cornerstone, and delete the audio. All of it happens on your laptop."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons to choose OpenWhispr","subtitle":"In a three-vet practice this is a budget decision the owner signs off on, not an IT project. Here are the facts, and the practice manager can hand the repo link to whoever does your IT.","items":[{"title":"Choose local when dictation must stay on your device","body":"Dictation, and note drafts with a local model, are processed on your machine. The open-source code shows there's no upload step."},{"title":"Every DVM and every tech, $0","body":"Unlimited local dictation, forever. Scribes charge per DVM and leave the techs writing records for free anyway."},{"title":"Works in the barn and in the truck","body":"The local model runs with the laptop in airplane mode. Dictate in the truck, file the note when you're back in range."},{"title":"Dictate into the PIMS you already run","body":"A system-wide hotkey drops clean text wherever your cursor is: ezyVet, Cornerstone, Avimark, a discharge template, a referral email. No integration to build, and no reason to change PIMS."},{"title":"One dictation, the SOAP note and the discharge","body":"Save the prompt once. A local model, or your own AI key, turns the same dictation into the record for the file and the instructions the client takes home."},{"title":"No recorder in the room","body":"You dictate afterwards. Nothing sits on the table during the estimate, and nothing is running when you put a hand on a client's shoulder."},{"title":"Choose how long local records stay on the practice laptop","body":"When you enable local audio or transcript retention, files stay on this device and are deleted after the retention period."},{"title":"Read the code before the owner signs anything","body":"Every line is on GitHub. The practice manager can send the link to whoever does your IT and have an answer the same week."},{"title":"Knows cephalexin from Cerenia","body":"The custom dictionary learns the drug names, breeds and procedures you correct, and you add the abbreviations your clinic already writes: OHE, TPR, BAR, QAR."},{"title":"100+ languages","body":"Take the history from a client who would rather explain it in Spanish, transcribed on your laptop."}]},"faq":{"label":"Before you download","title":"The questions vets ask first.","subtitle":"Straight answers on cost, terminology and what happens when you leave WiFi. If yours isn't here, ask us in Discord.","items":[{"question":"Do you record the appointment, like a scribe does?","answer":"No. Nothing joins the consult and nothing sits on the table. You dictate afterwards, at the treatment table or in the truck, and in local mode that dictation is transcribed on your own laptop with no upload step. The code is public, so you can check that instead of taking our word for it."},{"question":"Do vets even need this? HIPAA doesn't cover animals.","answer":"Right. HIPAA covers human patients, not pets. But the record is still confidential, and the audio carries the client's name, their address and the hardest conversation of your day. Local-first keeps all of it off a vendor's server."},{"question":"Does veterinary dictation software get drug names and breeds right?","answer":"Yes, once you teach it yours. It runs Whisper- and Parakeet-class models trained on a wide range of accents and languages, and the custom dictionary learns the terms you correct: Apoquel, Cerenia, brachycephalic, Bichon Frise, OHE, plus whatever abbreviations your clinic already writes."},{"question":"Will it work on a farm call with no signal?","answer":"Yes. Local models run entirely offline. Dictate in the barn or the truck, and file the note once you're back in range. Cloud scribes can't do this, because they stall the moment you leave WiFi."},{"question":"Can it write the discharge as well as the SOAP note?","answer":"Yes, from the same dictation. Run a local model (Llama, Qwen, Gemma) or connect your own AI key, then save your own SOAP, discharge and client-letter prompts as one-click actions and run whichever one the case needs. A local model keeps it all on your machine; your own key means the model and data path are ones you already trust."},{"question":"Is it actually free?","answer":"Yes. The Free plan includes unlimited local AI models, 100+ languages, and desktop availability on macOS, Windows and Linux at $0. Pro is $8/month and adds unlimited OpenWhispr Cloud transcription, sync across devices, and the iPhone and iPad app."}]},"reviews":{"label":"What users say","title":"Built by people who wouldn't put a recorder in the room either.","subtitle":"Three OpenWhispr users describe how they use it.","score":{"value":"0","label":"Consults on a server","body":"You dictate afterwards, so there is no appointment recording to begin with. In local mode the dictation is never uploaded either."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for veterinarians","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Vets and techs who want free, private, offline dictation and are fine with a quick local setup"}]},{"parts":[{"text":"Vets who want easy, reliable cloud transcription"}]},{"parts":[{"text":"Practices and teams documenting at volume"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Free offline dictation, with consult audio staying on your device"}]},{"parts":[{"text":"Unlimited cloud transcription without managing local models or keys"}]},{"parts":[{"text":"Unlimited recordings plus advanced workflow features for a team"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Enterprise includes team features and admin controls, plus dedicated support."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"The goodbye appointment doesn't belong on a vendor's server.","paragraphs":["OpenWhispr started with a simple objection: the three conversations that take the most out of you, the diagnosis, the estimate and the goodbye, shouldn't have to pass through a stranger's server to become a note.","We build in public. The code is on GitHub. When you choose local processing, dictation stays on your device. A practice can verify that in the code.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Last appointment at six. Records closed at six.","body":"Download OpenWhispr and dictate the next case on the walk back to treatment. Choose local models when dictation needs to stay on your laptop and work offline.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Unlimited local AI models are free on macOS, Windows and Linux."}}},"useCaseJournalists":{"breadcrumb":{"home":"Home","useCases":"Use Cases","journalists":"For Journalists"},"metadata":{"title":"Free AI Speech To Text Tool For Journalists - OpenWhispr","description":"Free, open-source AI speech to text for journalists. In local mode the interview is transcribed on your laptop, never uploaded. No per-minute bill.","keywords":["ai speech to text tool for journalists","transcription software for journalists","interview transcription software","offline interview transcription","private transcription for reporters","local speech to text","free interview transcription"]},"schema":{"name":"OpenWhispr","applicationCategory":"Journalism transcription software","operatingSystem":"macOS, Windows, Linux","description":"Local-first, open-source speech to text software for journalists. In local mode interview audio is transcribed on the reporter's own device and never leaves it. Free for unlimited local transcription.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"AI speech to text for journalists & reporters","title":"Drop in the tape. Get the whole transcript.","body":"Get the whole transcript without uploading the tape. In local mode, OpenWhispr transcribes on your machine, free and open source, with no per-minute bill or signal once the model is downloaded.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Person typing on a laptop beside a field recorder in an open vehicle"}},"trustBar":{"label":"Paste your quotes into:","names":["Google Docs","Microsoft Word","WordPress","Arc XP","Slack","Substack"]},"pain":{"label":"What an upload actually costs you","title":"Source protection shouldn't end at the upload button.","subtitle":"You want the tape transcribed on deadline. You don't want it living in someone else's account to get there.","cards":[{"title":"Keep source audio on your device in local mode.","body":"Upload an interview and a vendor holds the recording. Some services route the audio to human transcriptionists. In local mode OpenWhispr transcribes on your device, so there is no server for the tape to sit on."},{"title":"Your tape becomes someone else's record.","body":"Every upload leaves unpublished material in an account you don't administer, on a retention schedule you didn't set. In local mode there is no upload and no account, so there is nothing on our side for anyone to subpoena."},{"title":"A quote you can't check.","body":"Speech to text mishears names and crosstalk, and every model will occasionally write a sentence nobody said. A summary will not show you where that happened. Record the interview in the app and the transcript is stamped line by line. To check a quote before you file, compare it with the original recording you keep."},{"title":"\"Trust us\" is not source protection.","body":"Closed tools ask you to take a privacy policy on faith. OpenWhispr's code is public on GitHub under the MIT license. Your security desk, or any developer you trust, can read exactly what it does and what it doesn't."}]},"features":{"label":"What you get","title":"Filed faster. The tape stays yours.","subtitle":"Transcription tools usually trade source protection for speed, or bill you by the minute. This does neither. In local mode the interview is transcribed on your machine, at no cost per minute.","items":[{"image":{"label":"Two people talking across a table with an open laptop"},"title":"Interview files transcribed on-device","body":"Drop in an MP3, WAV, M4A or half a dozen other formats, one file or a whole batch. In local mode it's transcribed on your machine, with no length limit and no per-minute bill."},{"image":{"label":"Person reviewing printed pages beside a laptop and phone"},"title":"The whole transcript, never a summary","body":"You get the full transcript, not a digest. Record the interview in the app and every line is stamped on your device, ready to export as .txt, .md, .json or .srt. On an upload, speaker detection stamps each turn locally; per-line stamps need your own Groq or Mistral key, or OpenAI set to whisper-1."},{"image":{"label":"Two people talking at a desk with a field recorder and notebook"},"title":"Live transcript from the room","body":"Put your laptop on the table and it transcribes an in-person interview as you talk, on any platform. On a remote interview it captures your computer's own audio, so no bot joins and nothing shows in the participant list. That capture needs Windows, macOS 14.2 or later, or a Linux desktop whose audio portal allows it. Recording is its own setting, so point that one at a local model too."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Source audio sent to us","value":"None","footnote":"in local mode, your audio is transcribed on your own device"},{"label":"Cost per hour of tape","value":"$$0","footnote":"unlimited local transcription, every reporter, forever"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first transcript, plus the one-time model download"}]},"howItWorks":{"label":"How it works","title":"Transcribed before you write the lede.","subtitle":"No account or credit card required. Choose a local model in Settings, then drop in the tape or record in the app.","image":{"label":"Person at a desk with a laptop, notebook, and audio recorder"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Skip the account and setup hands you a local model. Either way, point uploads and recording at one in Settings."},{"title":"Drop in the tape.","body":"MP3, WAV, M4A, FLAC and more, one file or a batch. In local mode it's transcribed on your device, however long the interview ran."},{"title":"Check it, then file.","body":"Read the transcript in the app and check any line against the audio. A recorded interview exports to .txt, .md, .json or .srt with every line stamped."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"On the record","title":"10 reasons to choose OpenWhispr","subtitle":"Picking an AI speech to text tool is a source-protection decision before it's anything else. Here are the facts, and the ones about your audio you can check in the code yourself.","items":[{"title":"Local mode keeps source audio on your device","body":"Transcription runs on your machine, with no upload step."},{"title":"No bot joins a remote interview","body":"It records your computer's own audio, on Windows, macOS 14.2 or later, or a Linux desktop that allows it. Nothing appears in the participant list, and on a local model no vendor gets a copy of the call."},{"title":"Offline, once the model is downloaded","body":"Courthouse steps, a jail visit, a rural county meeting. Once the model is downloaded, no signal required."},{"title":"No import limits, no per-minute bill","body":"Local transcription is unlimited at $0, on every plan."},{"title":"Your security desk can read the code","body":"MIT licensed, every line on GitHub. A newsroom security review can read exactly what happens to the audio, before you put a source's tape through it."},{"title":"Recorded interviews, stamped line by line","body":"Export to .txt, .md, .json or .srt with every line stamped, so you can point standards or the lawyers at the exact moment."},{"title":"No length limit on your own machine","body":"A ninety-minute interview costs the same as a nine-minute one, and takes as long as your laptop needs."},{"title":"Who said what, on every plan","body":"Two voices in one interview, labelled. Speaker detection runs on your device, so it works on a local model with nothing uploaded."},{"title":"Nothing rewrites your quotes","body":"An uploaded interview is never rewritten. The note is the transcript as the model heard it. In dictation the untouched text sits beside the cleaned-up version, and you can switch cleanup off entirely."},{"title":"100+ languages","body":"Interview a source in their own language, not their second one. In local mode, transcription stays on your machine."}]},"faq":{"label":"Before you download","title":"The questions reporters ask first.","subtitle":"Straight answers, including where it falls short. Anything missing, ask in Discord.","items":[{"question":"Does my source's audio ever leave my device?","answer":"Not in local mode. Transcription runs on your machine and the open-source code proves there's no upload step. Cloud transcription is the default, so switch to a local model in Settings before your first interview. Dictation, recordings and uploaded files each carry their own transcription setting, so change the one you use for tape. If you sign in, point the AI cleanup model at a local one too, or switch cleanup off: that setting writes the note title and it defaults to the cloud."},{"question":"Can it transcribe a file I already recorded?","answer":"Yes. Drag in an MP3, WAV, M4A, FLAC, AAC, OGG, Opus or WebM file, pick it from a file dialog, or queue a batch. On a local model there's no length limit and no per-minute charge. Local video files aren't supported, but you can paste a YouTube or video link and it pulls the audio, up to six hours or 500 MB."},{"question":"Is it accurate enough to quote from?","answer":"It runs Whisper-class and NVIDIA Parakeet models, and no speech to text is perfect on accents, names or crosstalk. That's why you get the whole transcript, never a summary: record the interview in the app and it's stamped line by line, so you can check the quote against the tape before you file. Treat it as a first pass you verify, not a transcript you publish unread."},{"question":"Can it tell my source's voice from mine?","answer":"Yes. On a local model the speaker detection runs on your device too, with nothing uploaded. It separates the voices in an in-person interview and labels who said what, on every plan; the transcript is live as you talk and the labels land when you stop. A recording expects two speakers and goes up to eight, enough for a panel. On an uploaded file, switch labels on before you transcribe and it works out the count for you."},{"question":"If we get a subpoena, what do you have?","answer":"In local mode, nothing. There is no upload and no account, so there is no tape and no transcript on our side to hand anyone. What exists is on your laptop, under your own retention and source-protection rules. That is also why offline works: once the model is downloaded, a courthouse or a jail visit needs no connection."},{"question":"Is it actually free?","answer":"Yes. Unlimited local transcription, speaker labels, 100+ languages, all three desktop platforms, $0 forever. Pro ($8/month) adds unlimited OpenWhispr Cloud speech-to-text and mobile dictation. Business ($16/user/month) adds unlimited cloud meeting recordings and priority support."}]},"reviews":{"label":"What users say","title":"Different jobs. The same instinct about their own audio.","subtitle":"From our Discord, GitHub and inboxes. None of them is a reporter, so read them for the tool, not the beat.","score":{"value":"0","label":"Interviews on a vendor's server","body":"In local mode the interview never leaves your device. There is no copy on our side, because nothing is sent to us."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for journalists","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Reporters who want free, offline transcription and don't mind setting up a local model"}]},{"parts":[{"text":"Reporters who'd rather not manage a local model and are fine with cloud transcription"}]},{"parts":[{"text":"Desks that transcribe at volume, or need unlimited cloud meeting recordings"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"The tape never leaves your laptop, at $0"}]},{"parts":[{"text":"Unlimited cloud transcription without managing local models or keys"}]},{"parts":[{"text":"Unlimited cloud meeting recordings and priority support for a desk"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some conversations only happen off the record. We built for those.","paragraphs":["OpenWhispr started with a simple objection: a tape of a source who asked for confidentiality shouldn't end up on a stranger's server just so you can read back what they said.","We build in public. The code is on GitHub. When we say the audio stays local, that isn't just a promise in a privacy policy. Anyone on your desk can inspect the code and verify what happens to the audio.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Your tape stays between you, your source and your laptop.","body":"Download OpenWhispr, point it at a local model, and your next interview is transcribed on your machine and nowhere else.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseContentCreators":{"breadcrumb":{"home":"Home","useCases":"Use Cases","contentCreators":"For Content Creators"},"metadata":{"title":"Free Dictation Software For Content Creators - OpenWhispr","description":"OpenWhispr is free, open-source dictation software for content creators. Paste your own video link or an audio file, get the transcript, talk out the rest.","keywords":["dictation software for content creators","dictation for youtubers","youtube video transcript software","transcribe your own youtube video","voice to text for creators","speech to text for video scripts"]},"schema":{"name":"OpenWhispr","applicationCategory":"Dictation software for content creators","operatingSystem":"macOS, Windows, Linux, iOS","description":"Local-first, open-source transcription and dictation software for content creators. Paste the link to a video you posted yourself, or drop in an audio file, and get the transcript in a note you can search and quote. Then talk out the title, the description and the captions and the clean text lands wherever the cursor already is. With a local model chosen the audio is transcribed on the creator's own device. Free for unlimited local transcription and dictation on macOS, Windows and Linux, and it works fully offline.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation software for content creators","title":"Paste your own video link. Get the transcript.","body":"Get a transcript of your own YouTube video or audio file on your laptop for $0. Then use the same hotkey to dictate the title and description.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"A person speaks to a camera in a home studio with a ring light and laptop nearby"}},"trustBar":{"label":"Dictate into any app, including:","names":["YouTube Studio","CapCut","Notion","Patreon","Buffer","Discord"]},"pain":{"label":"What you're choosing between","title":"Everything you said is in the video. None of it is in text.","subtitle":"A transcript is what turns one upload into chapters, a newsletter and a thread. Getting one means paying per minute, or sitting through your own video with a keyboard.","cards":[{"title":"Your back catalogue is locked inside the videos.","body":"Turning last year's uploads into a newsletter means sitting through them again. Paste the link to one of your own videos and OpenWhispr pulls the audio down and transcribes it. Paste a list of your links, up to 50 at a time, and it works through the queue."},{"title":"The interview is on your disk and nowhere else.","body":"The raw recording, the voice memo, the podcast cut that never went up. Drop the audio file in and it comes back as a transcript in a note you can search. mp3, wav, m4a, flac, aac, opus, ogg and webm, straight off the disk."},{"title":"The description box, still blank at midnight.","body":"The export is done and the title, the description and the pinned comment are not. Press a hotkey in the tab you already have open and say them. Clean, punctuated text lands where the cursor is."},{"title":"Another subscription before the channel pays for itself.","body":"Editor, thumbnails, scheduler, transcripts, newsletter. Five tools billing one channel that does not cover them yet. Unlimited local transcription and dictation on OpenWhispr is $0, on every machine you own, and the free plan is not a trial."}]},"features":{"label":"From the upload to the words around it","title":"The transcript is there before you sit down to write.","subtitle":"Pick local models and both the transcription and the clean-up run on your own computer, unlimited, at no cost, in 100+ languages.","items":[{"image":{"label":"A person works at a laptop in a home editing setup with audio equipment nearby"},"title":"A transcript out of a video you already posted","body":"Paste the link to a video you posted yourself, or a list of them one per line, and OpenWhispr downloads the audio and transcribes it into a note you can search and quote. Up to 50 links at a time, 500 MB and six hours a video. Playlists and live streams are out."},{"image":{"label":"A person speaks beside a camera and laptop at a kitchen table"},"title":"The audio file that never got posted","body":"Interviews, voice memos, the podcast cut, the raw take. Drop in mp3, wav, m4a, flac, aac, opus, ogg or webm and it comes back as a transcript. On a local model there is no per-minute bill, and a file already on your disk has no size limit at all."},{"image":{"label":"A person speaks while using a laptop beside handwritten notes"},"title":"One hotkey, every tab you publish from","body":"Talk into YouTube Studio, your script doc, Notion, an email to a sponsor. The clean text pastes itself where you were already typing, with no plugin in any of them. On a Mac you grant accessibility once, and that is the part that lets it paste at your cursor."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Sent to our servers","value":"None","footnote":"with a local transcription model the audio is transcribed on your own computer"},{"label":"Cost per creator","value":"$$0","footnote":"unlimited local transcription and unlimited local dictation, on every machine you own"},{"label":"Setup takes","value":"A few extra steps","footnote":"download, pick a model, paste a link or set a hotkey"}]},"howItWorks":{"label":"Download to first transcript","title":"Three steps from your posted video to searchable text.","subtitle":"Nothing to connect to the platforms you publish on.","image":{"label":"A person speaks into a microphone at a desk with a laptop and camera gear nearby"},"steps":[{"title":"Download it free.","body":"Mac, Windows or Linux, whatever the edit rig runs. Setup offers OpenWhispr Cloud or local models, and picking local downloads them once."},{"title":"Paste the link, or drop the file.","body":"Your own video's URL, a list of them one per line, or an audio file off your disk. OpenWhispr pulls the audio down and transcribes it into a note."},{"title":"Read it, use it.","body":"Search it, quote it, cut it into chapters and a newsletter. Then hit the hotkey in YouTube Studio and talk the title and the description out."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Why creators keep it","title":"10 reasons content creators choose OpenWhispr","subtitle":"Every one is in the app or in the repo. Check before you download anything.","items":[{"title":"Paste your own video link, get the transcript.","body":"Give OpenWhispr the YouTube URL of a video you posted, a direct audio URL, or an audio file from your disk. It transcribes the audio and can queue up to 50 links; playlists and live streams are not supported."},{"title":"It types into everything you publish from.","body":"Use the system-wide hotkey in YouTube Studio, your script doc, newsletter or sponsor email. On a Mac, one accessibility permission lets OpenWhispr paste at the cursor."},{"title":"Unlimited local transcription costs nothing.","body":"The free plan is not a trial. With a local model there is no per-minute meter on a file, no word cap on dictation, and no card to enter."},{"title":"Your niche words come out spelled right.","body":"The custom dictionary takes your channel name, your recurring guests and the vocabulary only your audience uses. Dictation hands that list to the transcription model as a hint, and to the clean-up model after it."},{"title":"The ums and the false starts come out.","body":"The clean-up model strips filler, stutters and restarts, and is told to keep your voice, your wording, your formality and your intent, with your jargon and names exactly as spoken."},{"title":"100+ languages.","body":"Dictate the description in the language your audience actually comments in. Same hotkey, same model, no separate tool."},{"title":"Location shoots with no signal.","body":"A rented studio, a hotel, a plane, a shoot in the middle of nowhere. Local models need no network at all."},{"title":"Nothing is uploaded when you keep it local.","body":"Pick a local transcription model and the audio is transcribed on your own machine, so none of it reaches us. Point the clean-up at a local model too and the text stays put as well."},{"title":"The code is public.","body":"OpenWhispr is open source on GitHub, where 5,000+ engineers have starred it. Every privacy claim on this page is a file you can open and read."},{"title":"Mac, Windows and Linux, one free download.","body":"Whatever you edit on. The same app, the same free price, and a hotkey you choose yourself."}]},"faq":{"label":"Before you install it","title":"The questions creators ask before they download it","subtitle":"Short answers, and where in the app or the repo you can check them.","items":[{"question":"Can I get a transcript of a video I already published?","answer":"Paste the link to your own video. OpenWhispr accepts a YouTube URL or a direct audio file URL, downloads the audio and transcribes it into a note. Paste a list of your links, one per line and up to 50 at a time, and it works through them as a queue. Playlists and live streams are not supported, and the download caps at 500 MB and six hours a video. On a local model that cap is the only one. On OpenWhispr Cloud the free plan transcribes files up to 25 MB, and past that it is a paid plan."},{"question":"Can I upload the video file itself?","answer":"Not the video file. The upload panel takes audio: mp3, wav, m4a, webm, ogg, oga, flac, aac and opus. For something you already posted, paste its link instead and OpenWhispr pulls the audio down for you. For a cut that is not posted anywhere yet, export the audio out of your editor and drop that in."},{"question":"Is my audio uploaded anywhere?","answer":"Not if you pick a local transcription model: the audio is transcribed on your own computer and nothing is sent. One thing worth knowing, because it is two settings and not one. Transcription and the AI clean-up are chosen separately, and if you are signed in the clean-up still defaults to OpenWhispr Cloud. On that default the transcript text, never the audio, is sent to us to be tidied. Set the clean-up model to a local one too and neither leaves the machine."},{"question":"Is it really free, or is the free plan a trial?","answer":"Free is free, with no card. Unlimited local dictation and unlimited local transcription are $0 on every plan, forever. Pro ($8/month) makes OpenWhispr Cloud transcription unlimited, and Business ($16/month) adds unlimited cloud meeting recordings and agent mode. Nothing on the free plan expires."},{"question":"Will it get my channel name and my niche jargon right?","answer":"Add them to the custom dictionary once. Your channel name, your recurring guests, the in-jokes and the product names your audience uses. Dictation passes that list to the transcription model as a hint, and the clean-up model gets the same list. One exception worth knowing: the local Parakeet model does not take the hint, so use a local Whisper model if the dictionary matters."},{"question":"Does it work on a shoot with no WiFi?","answer":"Yes, once the local models are downloaded. Local transcription and a local AI model both run on your own machine with no network, so a hotel, a plane or a location with no signal is the same as your desk. Cloud transcription is the option that needs a connection."}]},"reviews":{"label":"What people tell us","title":"They tried it on their own machine and stopped typing.","subtitle":"From our Discord, GitHub and inbox, word for word. None of them runs a channel, so read them for the tool.","score":{"value":"0","label":"Plugins to install in YouTube Studio","body":"The hotkey works the same in a browser tab, your edit suite and your newsletter draft. Nothing to connect, nothing to log into, no integration to keep alive."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for content creators","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"One-person channels that want unlimited local transcription and dictation at no cost and will set up a local model"}]},{"parts":[{"text":"Creators who would rather pay $8 than look after a local model"}]},{"parts":[{"text":"Channels where an editor, a writer and a producer all touch the same content"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Transcripts of your own videos, and everything you write around them, for nothing, with the audio staying on your machine"}]},{"parts":[{"text":"Cloud transcription that never runs out, with no models or API keys to look after"}]},{"parts":[{"text":"Unlimited cloud meeting recordings and agent mode for everyone on the channel"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"The fastest way to sound like yourself is to say it out loud.","paragraphs":["OpenWhispr started with an objection: talking is how people are fluent and typing is how they get stiff, and the software that turns one into the other kept routing everything through somebody else's server.","We build in public. Every release and every line of the app is on GitHub. So when we say the dictation can stay on your machine, you do not have to take our word for it. You can open the file and read it.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"$$0 · open source · no plugin to install","title":"Get the transcript, then write around it.","body":"Download OpenWhispr, pick a local model at setup, and paste the link to the video you posted this morning. No card, no per-minute bill, no second subscription.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free forever on Mac, Windows & Linux."}}},"useCasePodcasters":{"breadcrumb":{"home":"Home","useCases":"Use Cases","podcasters":"For Podcasters"},"metadata":{"title":"Best Speech-to-Text Software For Podcasting - OpenWhispr","description":"OpenWhispr is free, open-source speech-to-text for podcasters. Transcribe unlimited episode audio on your own machine and never pay per minute again.","keywords":["speech to text for podcasting","podcast transcription software","free podcast transcription","dictation software for podcasters","podcast show notes software","offline podcast transcription","open source transcription software"]},"schema":{"name":"OpenWhispr","applicationCategory":"Podcast transcription software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Local-first, open-source speech-to-text for podcasters. With a local model chosen, episodes and show notes are transcribed on the creator's own device. Free for unlimited local transcription and dictation, and works fully offline.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text software for podcasters","title":"Record the episode. Get the transcript. Pay nothing per minute.","body":"Turn each recorded episode into a transcript, then dictate show notes, descriptions and posts into any app. Choose local transcription for offline work, or Cloud for high accuracy and speed without downloading models.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Person speaking into a studio microphone beside a laptop and handwritten notes"}},"trustBar":{"label":"Write show notes in:","names":["Notion","Buzzsprout","Transistor","Spotify for Creators","Substack","YouTube Studio"]},"pain":{"label":"The problem with per-minute transcription","title":"Transcription shouldn't cost more than your hosting.","subtitle":"A transcript and show notes are part of shipping many episodes. Many tools bill by the minute or cap usage by the month.","cards":[{"title":"Paying by the minute to transcribe your own voice.","body":"A weekly one-hour show, metered per minute, turns into a real monthly bill that never ends. OpenWhispr transcribes on your machine: unlimited hours, $0, and the Free plan is not a trial."},{"title":"Show notes take longer than the episode.","body":"You talked for an hour, then typed for two. Press a hotkey and speak the summary, the guest bio and the links straight into your host or your docs. Clean, punctuated text, at speaking speed."},{"title":"Guest names turn show notes into a correction job.","body":"When your guest's name or show title comes out wrong in dictated notes or AI-drafted summaries, you have another correction job. Add those terms to OpenWhispr's dictionary to improve how they appear."},{"title":"Pulling one quote means scrubbing the whole episode.","body":"Clips, quote graphics and newsletters all start with finding the line. With a transcript on your machine you search text, copy the quote and move on. No scrubbing."}]},"features":{"label":"What you get","title":"Transcripts, show notes, summaries. No meter on any of them.","subtitle":"Make transcripts and show notes on your own machine in 100+ languages, without adding another monthly transcription bill.","items":[{"image":{"label":"Person monitoring audio at a microphone with a laptop and recording interface"},"title":"Unlimited local transcription","body":"Import the episode you already cut, and it is transcribed on your device with local Whisper-class models. No per-minute fees or monthly hour caps for local transcription, with nothing uploaded in local mode."},{"image":{"label":"Person speaking into a desktop microphone while working on a laptop"},"title":"Show notes at speaking speed","body":"Press a hotkey and speak your summary, description or newsletter straight into Notion, Buzzsprout, Substack, wherever your cursor is."},{"image":{"label":"Two people recording with microphones while one reviews notes beside a laptop"},"title":"A dictionary that knows your show","body":"Add your show's title, recurring guests and niche vocabulary to the custom dictionary. It improves how those names and terms are handled in dictation and AI-drafted notes."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Unreleased audio uploaded","value":"None","footnote":"in local mode the episode never leaves your machine"},{"label":"Cost per episode","value":"$$0","footnote":"unlimited local transcription and dictation, every episode, forever"},{"label":"Transcription mode","value":"Local or Cloud","footnote":"Local runs on your device. Cloud prioritizes accuracy and speed with no model download."}]},"howItWorks":{"label":"How it works","title":"A transcript and show notes before you close the editor.","subtitle":"Choose local transcription when you need offline work or want audio to stay on your machine. Choose Cloud for high accuracy and speed without downloading local models.","image":{"label":"Person holding headphones beside a closed laptop and handwritten notes"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Pick a local model during setup."},{"title":"Record or dictate.","body":"Record a session for a local transcript, or press the hotkey and speak your show notes into any app."},{"title":"Review, paste, publish.","body":"Edit the text, paste it into your host or your docs, and ship the episode. You control all of it."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons podcasters choose OpenWhispr","subtitle":"Transcription cost recurs with every episode. These are the practical details to compare.","items":[{"title":"Unlimited local transcription","body":"No per-minute fees or monthly hour caps for local transcription on the Free plan."},{"title":"Free for your whole show","body":"Unlimited local dictation and transcription at $0: you, your co-host, your editor. No per-seat licence."},{"title":"Dictate straight into your host","body":"A system-wide hotkey drops clean text wherever your cursor is: Buzzsprout, Notion, Substack, YouTube Studio."},{"title":"Summaries drafted, not just transcribed","body":"Connect your own AI key, or run a local model, to turn a transcript into an episode summary or description."},{"title":"A dictionary for your guests' names","body":"Teach it the show title, recurring guests and niche terms once. The custom dictionary improves how it handles them in dictated notes and drafted summaries."},{"title":"Works fully offline","body":"Transcribe on the flight home from the conference. Local models never need a signal."},{"title":"Pull quotes without scrubbing","body":"The transcript is text on your machine. Search it, copy the line, make the clip."},{"title":"Unreleased audio stays on your machine","body":"In local mode nothing is uploaded. Embargoed interviews and unshipped episodes never touch a server."},{"title":"Open source, auditable by anyone","body":"OpenWhispr is open source under the MIT License."},{"title":"100+ languages","body":"Record in the language your audience actually speaks, transcribed locally."}]},"faq":{"label":"Before you download","title":"The questions podcasters ask first.","subtitle":"Straight answers on cost, accuracy and privacy. If yours isn't here, ask us in Discord.","items":[{"question":"What's the best speech-to-text software for podcasting?","answer":"OpenWhispr is a strong option if you want to transcribe recorded episodes locally without per-minute fees. It can turn imported audio into a transcript and lets you dictate show notes wherever your cursor is. Its open-source code makes the workflow inspectable. Trying it costs a download."},{"question":"Is it really free?","answer":"Yes. Local dictation and transcription are included at $0 on macOS, Windows and Linux. The Free plan also includes 2,000 words a week of OpenWhispr Cloud transcription and 5 hours a month of meeting recordings. Pro is $8/month with unlimited Cloud transcription and 20 meeting-recording hours. Business includes unlimited meeting recordings."},{"question":"Will it get my guests' names and niche terms right?","answer":"OpenWhispr supports 100+ languages. Its custom dictionary lets you add your show's title, guest names and niche vocabulary to improve how they appear in dictation and AI-drafted show notes."},{"question":"Does it transcribe my recorded episodes, or only what I say live?","answer":"Both. Import an episode you've already recorded, or record a session in the app, and the transcript is made on your device. Press the hotkey to dictate show notes into any app, and connect your own AI key, or a local model, to draft summaries from the transcript."},{"question":"Does it work offline?","answer":"Yes, once the models are downloaded at setup. After that they run entirely offline. Transcribe an interview on the plane and publish when you land."},{"question":"Do unreleased episodes stay private?","answer":"In local mode, the audio stays on your machine. OpenWhispr is open source under the MIT License. Cloud transcription is optional: choose local for offline work or Cloud for high accuracy and speed without downloading models."}]},"reviews":{"label":"What users say","title":"Built for people who'd rather talk than type.","subtitle":"What people say about OpenWhispr.","score":{"value":"0","label":"Invoices for your own words","body":"Local transcription runs on your machine. There is no meter, no hour cap, and no bill at the end of the month."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for podcasters","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Get unlimited Cloud transcription"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Podcasters who want free, unlimited local transcription and are fine with a quick local setup"}]},{"parts":[{"text":"Podcasters who want easy cloud transcription and mobile dictation"}]},{"parts":[{"text":"Networks and teams producing shows at volume"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local transcription, with your audio staying on your machine"}]},{"parts":[{"text":"Unlimited cloud speech-to-text without managing local models or keys"}]},{"parts":[{"text":"Unlimited cloud recordings plus advanced workflow features for a team"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Your words are the show. You shouldn't rent them back.","paragraphs":["OpenWhispr started with a simple objection: turning your own voice into text shouldn't cost by the minute or pass through someone else's server.","We build in public on GitHub. Local transcription is not just a privacy-policy promise. It is a property of the code you can inspect.","And if anything breaks the night before an episode drops, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Stop paying by the minute for your own voice.","body":"Download OpenWhispr, pick a local model at setup, and your next episode gets transcribed, summarized and shipped on your own computer.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free desktop app for macOS, Windows and Linux."}}},"useCaseProductTeams":{"breadcrumb":{"home":"Home","useCases":"Use Cases","productTeams":"For Product Teams"},"metadata":{"title":"Free AI Meeting Notes For Product Teams, No Bot - OpenWhispr","description":"OpenWhispr is free, open-source AI meeting notes for product teams. No bot joins the call, and with a local model the roadmap never leaves your machine.","keywords":["AI meeting notes","meeting notes for product teams","local meeting transcription","private meeting notes","bot-free notetaker","offline meeting notes"]},"schema":{"name":"OpenWhispr","applicationCategory":"Meeting notes software","operatingSystem":"macOS, Windows, Linux","description":"Local-first, open-source AI meeting notes for product teams. With a local model the call is transcribed on the user's own machine with no bot joining, and written up as decisions, action items and owners. Local meeting notes are unlimited and free, and work fully offline.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"AI meeting notes for product teams","title":"Run the meeting. Get the decisions written up.","body":"Get a clean meeting write-up with decisions and next steps. OpenWhispr captures the call on your own computer; pick a local model and the recording never reaches a server, ours or anyone's.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Two people listen to a colleague during a meeting around a table"}},"trustBar":{"label":"Paste the notes into:","names":["Linear","Jira","Notion","Confluence","Slack","Productboard"]},"pain":{"label":"The trade-offs product teams notice","title":"Product decisions shouldn't disappear when the call ends.","subtitle":"The PM should be facilitating, not typing. OpenWhispr turns the discussion into decisions, action items and owners, without adding a bot. A local model keeps transcription on your computer.","cards":[{"title":"A bot shows up to every meeting.","body":"Meeting bots can make a sensitive discovery call feel different. OpenWhispr captures your computer's audio, so it does not join the call."},{"title":"Your strategy on a vendor's server.","body":"Roadmap, pricing, competitive analysis, reorg talk: product meetings can cover sensitive work. Choose a local model when you want transcription to run on your device."},{"title":"Someone stops contributing to take notes.","body":"The PM types instead of facilitating, and the decision still gets lost. OpenWhispr writes up the decisions, action items and owners, so everyone stays in the discussion."},{"title":"Per-seat pricing for the whole org.","body":"Choose the mode and plan that fit the team. Local transcription is unlimited on every plan; Free includes 5 hours of meeting recordings a month, Pro adds unlimited Cloud transcription and 20 recording hours, and Business includes unlimited meeting recordings."}]},"features":{"label":"What you get","title":"Turn the product conversation into decisions and next steps.","subtitle":"Capture the call without a bot, review a structured write-up, then paste it into the tools where the team works. Choose a local model when the discussion needs to stay on your computer.","items":[{"image":{"label":"A person gestures while speaking to a colleague across a table"},"title":"Captures the call from your computer, no bot","body":"Listens to your computer's audio on Zoom, Meet, Teams or a Slack huddle, or to the mic for a standup in the room. Transcribed on your device with a local model, and nothing joins the call."},{"image":{"label":"Two colleagues review a laptop together during a group discussion"},"title":"Decisions, action items and owners","body":"Turns your rough notes and the transcript into a structured write-up: what was decided, what happens next, who owns it. You review before it's shared."},{"image":{"label":"A person gestures while working on a laptop at a table"},"title":"Paste it where the work lives","body":"Copy the clean write-up into Linear, Jira, Notion or Slack. It's text, so there's nothing to connect to your tracker."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Roadmap calls sent to us","value":"None","footnote":"pick a local model at setup and there is no upload step in the code"},{"label":"Cost per person on the call","value":"$$0","footnote":"unlimited local meeting notes, the whole org, forever"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first meeting written up"}]},"howItWorks":{"label":"How it works","title":"Leave the call with the decisions already written down.","subtitle":"Download the app, choose a model, then start the call.","image":{"label":"A person holding a tablet speaks in an open office"},"steps":[{"title":"Download for free.","body":"macOS 14.2+, Windows or Linux. Pick a local model during setup."},{"title":"Start the meeting.","body":"Open Zoom, Meet or Teams as usual. OpenWhispr captures your computer's audio and transcribes it on your device."},{"title":"Review, paste, move on.","body":"Hit Generate Notes, edit the write-up and paste it where the work lives. With a local model, transcription runs on your device."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons product teams choose OpenWhispr","subtitle":"Choosing a notetaker is a security decision before it's a note-taking one. Here are the facts, and your security team can verify the privacy ones line by line in the repo.","items":[{"title":"The roadmap call never reaches a server","body":"Pick a local model and transcription runs on your machine. The open-source code shows no upload step on that path."},{"title":"No bot in the call","body":"It captures your computer's audio, so nothing appears in the participant list. Your customer never sees a notetaker."},{"title":"Every seat is free, including the quiet ones","body":"Unlimited local meeting notes at $0, for everyone who ever sits in a call. Only the cloud side is metered."},{"title":"Works with any meeting app","body":"Zoom, Meet, Teams, Slack huddles, Webex, a standup in the room. It captures audio, not a specific app."},{"title":"Decisions and owners, not just transcripts","body":"The write-up is structured with decisions, action items and follow-ups. Where an owner is clear in the transcript, it uses that speaker label. You don't write the prompt."},{"title":"No meeting-app integration to approve","body":"No meeting-app plugin or bot account to approve. It's a desktop app that listens to your computer's audio."},{"title":"The recording is deleted after the meeting","body":"Temporary diarization audio is removed after processing. You control the recordings and transcripts saved in the app."},{"title":"Read the code before you pick the tool","body":"OpenWhispr is open source. Your security team can read the code on GitHub."},{"title":"No signal, same write-up","body":"A standup in a room with bad WiFi, a customer site, a plane. Local models need no signal."},{"title":"Who said what, worked out on your machine","body":"Speaker labels are computed on your own device, free on every plan. The write-up says who owns cutting scope, rather than merely that someone said it."}]},"faq":{"label":"Before you download","title":"The questions product teams ask first.","subtitle":"Straight answers on bots, your data and cost. If yours isn't here, ask us in Discord.","items":[{"question":"Does the meeting audio ever leave my device?","answer":"Not in local mode. Pick a local model during setup and transcription runs on your machine with no upload step. Cloud transcription is optional, and we never train on your data."},{"question":"Does a bot join the call?","answer":"No. OpenWhispr captures your computer's audio, so nothing appears in the participant list. Zoom, Meet, Teams, a Slack huddle or a standup in the room all work the same way because it listens instead of joining. With a local model, transcription runs on your device. Tell people you're recording, same as you would with a bot."},{"question":"Will it get our feature names and acronyms right?","answer":"Add them to the custom dictionary, free on every plan. The write-up is told to use those exact spellings, so codenames and feature names come out right in the notes you paste."},{"question":"Do the AI meeting notes capture decisions and owners, not just a transcript?","answer":"Yes. Jot rough notes during the call, then run the write-up when it ends. AI combines them with the transcript into decisions, next steps and action items. Where an owner is clear, it uses that speaker label. You review before anything is shared."},{"question":"Will it work for an in-person standup with no call?","answer":"Yes. It records from your mic, and local models run entirely offline. Capture the standup, then paste the write-up into Linear or Notion when you're done."},{"question":"Is it actually free?","answer":"Yes. Unlimited local meeting notes, 100+ languages, all three desktop platforms: $0 forever, for every seat. Only the cloud side is metered: cloud meeting recordings run 5 hours a month free, 20 on Pro ($8/month), unlimited on Business ($16/user/month)."}]},"reviews":{"label":"What users say","title":"Built by people who don't want a bot in their meetings either.","subtitle":"From our Discord, GitHub and inboxes, the same instinct that matters when the roadmap's on screen.","score":{"value":"0","label":"Roadmap calls on a vendor's server","body":"In local mode, transcription runs on your device. Cloud transcription is optional, and its audio is discarded after transcription."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for product teams","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Product teams who want private meeting notes and are fine with a quick local setup"}]},{"parts":[{"text":"Product teams who want easy, reliable cloud transcription"}]},{"parts":[{"text":"Product orgs in calls all day"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Free local meeting notes, with the audio staying on your device"}]},{"parts":[{"text":"Unlimited cloud transcription without managing local models or keys"}]},{"parts":[{"text":"Unlimited recordings plus advanced workflow features for a team"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for top accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","tooltip":"Cloud meeting recordings. Recording and transcribing meetings with a local model is unlimited on every plan.","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some meetings decide the roadmap. We built for those.","paragraphs":["OpenWhispr started with a simple objection: the meeting where you decide the roadmap shouldn't have to pass through a stranger's server to become a write-up.","OpenWhispr is open source. Your security team can inspect how local transcription works before putting it in roadmap meetings.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Take the decisions into the work.","body":"Download OpenWhispr, pick a local model, and your next meeting gets transcribed, written up and ready to paste on your computer.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseStudents":{"breadcrumb":{"home":"Home","useCases":"Use Cases","students":"For Students"},"metadata":{"title":"Free Speech-to-Text Software For Students - OpenWhispr","description":"OpenWhispr is free, open-source speech-to-text software for students. Record the lecture, get the notes written up on your own laptop, at $0 with no card.","keywords":["speech to text software for students","free dictation software for students","lecture transcription app","record lectures and get notes","speech to text for university students","free speech to text for students","student lecture notes app"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text software for students","operatingSystem":"macOS, Windows, Linux","description":"Local-first, open-source speech-to-text software for students. Lectures are recorded and transcribed on the student's own laptop, written up as structured notes with a local AI model, and essays are dictated the same way. Unlimited local transcription is free, and it works fully offline.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text software for students","title":"Hit record in the lecture. Get the notes written up.","body":"Turn a lecture into readable notes on your laptop. OpenWhispr records the room or the stream, and one tap on Generate Notes turns the transcript into headings and key points for $0 with a local model.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"A person smiling with a laptop while relaxing on a sofa."}},"trustBar":{"label":"Record lectures from:","names":["Zoom","Microsoft Teams","Panopto","Echo360","Google Meet","YouTube"]},"pain":{"label":"The problem with lecture notes","title":"You cannot listen and write at the same time.","subtitle":"A lecture runs at the speed of speech. Writing notes pulls your attention away from following the argument.","cards":[{"title":"The notes you take are the lecture you missed.","body":"Every sentence you write down is a sentence you stop listening to. Let the recording take the words and keep your attention on the whiteboard instead."},{"title":"A two-hour recording takes two hours to replay.","body":"Audio is hard to revisit when you cannot skim it. OpenWhispr hands back text and structured notes you can scan before replaying anything."},{"title":"Pick the right mode for exam month.","body":"Use local transcription for unlimited on-device transcription at $0, or Cloud for top accuracy and speed without downloading a model. Free includes 2,000 words of Cloud transcription a week and 5 hours of meeting recordings a month; Pro adds unlimited Cloud transcription and 20 recording hours."},{"title":"Your lectures, sitting on a company's server.","body":"Most transcription tools upload the audio to process it. In local mode OpenWhispr keeps the recording on your laptop, and the code is public on GitHub so you can check that yourself."}]},"features":{"label":"What you get","title":"The lecture, written up before you get home.","subtitle":"Recording is the easy half. What makes a lecture useful in week nine is text you can search and notes you can read.","items":[{"image":{"label":"A person listening beside a laptop in a tiered lecture hall."},"title":"Records the room or the stream","body":"In the lecture hall it records through the mic. For a Zoom, Teams or Panopto lecture it captures the audio playing on your laptop. Mac, Windows and Linux."},{"image":{"label":"A person sitting beside a laptop in a campus lounge."},"title":"Notes, not a wall of transcript","body":"The transcript goes through a write-up step that returns headings, key points and action items in markdown. With a local AI model that step runs on your laptop too."},{"image":{"label":"A person speaking while working on a laptop at a library desk."},"title":"Dictation for everything after","body":"Press a hotkey anywhere and talk. Clean text lands in Google Docs, Word or Overleaf, so the essay that follows the lecture gets said out loud instead of typed."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Lecture audio sent to us","value":"None","footnote":"in local mode, recording and transcription run on your laptop"},{"label":"Cost per student","value":"$$0","footnote":"unlimited local transcription and write-ups, no card, no expiry"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first recorded lecture"}]},"howItWorks":{"label":"How it works","title":"Three steps, and you leave with the write-up.","subtitle":"The whole setup is one download and one model choice. There is nothing to sign up for.","image":{"label":"A person wearing an earbud beside a laptop and handwritten notes."},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Pick a local model during setup and the whole thing runs offline."},{"title":"Hit record when the lecture starts.","body":"It takes the room through your mic, or the stream through your laptop's audio. Then you just listen."},{"title":"Tap Generate Notes.","body":"The transcript comes back as headings, key points and action items. Search those in revision week instead of scrubbing audio."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons students choose OpenWhispr","subtitle":"On a student budget, cost comes first.","items":[{"title":"Free forever, no card details","body":"Unlimited local transcription and note write-ups at $0, with no trial or semester limit."},{"title":"No bot joins the call","body":"OpenWhispr captures audio directly from your system; it does not join meetings as a participant."},{"title":"Your own notes go in the same file","body":"Type a line during the lecture and the write-up folds it in with what was said."},{"title":"Works with no WiFi","body":"Local models run offline. A basement lecture theatre, the train home, a flight: it still records and still transcribes."},{"title":"Recordings stay on your laptop","body":"In local mode the audio is transcribed on your device, and the open-source code shows the local path never uploads it."},{"title":"Searchable in revision week","body":"Search a term in your write-ups instead of scrubbing audio."},{"title":"Use a custom dictionary in the write-up","body":"Add lecturer names, module codes and technical terms to the custom dictionary to give the write-up the spellings you use."},{"title":"100+ languages","body":"Sit a lecture in your second language and read the notes in it, or dictate your essay in your first."},{"title":"Upgrade for unlimited cloud transcription","body":"The free plan includes 2,000 words a week of OpenWhispr Cloud transcription. Pro is $8/month for unlimited Cloud transcription. Students get 40% off for 12 months; email support@openwhispr.com with proof of enrolment for a code."},{"title":"Open source, under the MIT license","body":"Read the code before you install it. No privacy policy to take on faith."}]},"faq":{"label":"Before you download","title":"The questions students ask first.","subtitle":"Straight answers on cost, permission and what your university will think. If yours is not here, ask us in Discord.","items":[{"question":"Is it actually free, or free for a month?","answer":"Actually free. Local transcription, local note write-ups, 100+ languages and all three desktop platforms cost $0 forever, with no card details. The free plan also includes 2,000 words a week of OpenWhispr Cloud transcription and 5 hours a month of cloud meeting recordings. Pro is $8/month if you want unlimited Cloud transcription. Students get 40% off for 12 months; email support@openwhispr.com with proof of enrolment for a code."},{"question":"Am I allowed to record a lecture?","answer":"That is your university's call, not ours, and it varies by institution and by country. Most have a policy on recording teaching, often tied to accessibility. Check yours, and ask the lecturer. OpenWhispr does not record anything until you press the button."},{"question":"Does it work for online lectures as well as in-person ones?","answer":"Yes. In a lecture hall it records through your mic. For a Zoom, Teams, Google Meet or Panopto lecture it captures the audio playing on your laptop, on Windows, Linux and macOS 14.2 or newer. Already have the file, from your phone or a lecture-capture download? Drop it in and it transcribes that too, with no size limit on a local model."},{"question":"Will it get my subject's terminology right?","answer":"It runs Whisper-class models trained on a wide range of accents and 100+ languages. For the words a general model has never seen, add them to the custom dictionary in Settings: lecturer names, module codes and technical terms. Those spellings carry into your dictation and into the write-up of your notes."},{"question":"Is using the AI write-up cheating?","answer":"Transcription is not: the words are the ones your lecturer said, and reading them back is no different from reading your own notes. The write-up step is different, because a model summarises the transcript for you, so check your university's rules on AI before any of it goes into work you hand in."},{"question":"Where does my recording actually go?","answer":"It depends on the mode you pick during setup. OpenWhispr Cloud is the default: audio is sent only to transcribe it, and is not stored afterwards. Choose a local model instead and the answer is nowhere. It is transcribed on your laptop, and the open-source code shows the local path has no upload step. You can switch in Settings any time."}]},"reviews":{"label":"What users say","title":"Built by people who take notes in bad WiFi too.","subtitle":"From our Discord, GitHub and inboxes.","score":{"value":"0","label":"Lecture recordings kept on our servers","body":"In local mode the recording never leaves your laptop. On cloud, audio is sent only to transcribe it and is not stored after that."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for students","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Students recording lectures on their own laptop, with unlimited local transcription and notes"}]},{"parts":[{"text":"Students who want cloud transcription that just works, with no local models to download"}]},{"parts":[{"text":"Research groups and societies recording at volume"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"No meter and no card on the local side, with the recording staying on your laptop"}]},{"parts":[{"text":"Unlimited cloud speech-to-text without managing local models or keys"}]},{"parts":[{"text":"Unlimited cloud meeting recordings plus speaker labels for group work"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"A lecture you can read shouldn't need a subscription.","paragraphs":["OpenWhispr is for students who want to turn lectures into readable notes without a subscription.","We build in public. The code and releases are on GitHub, where 5,000+ engineers have starred the project. When we say the recording stays on your laptop, that's not a promise in a privacy policy. It's a property of the code, and you do not have to take our word for it.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Turn your 9am into notes with one tap.","body":"Download OpenWhispr, hit record when the lecture starts, and read the notes on the way home. On your laptop, at $0, with no card details.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free forever on Mac, Windows & Linux."}}},"useCaseSpanishVeterinary":{"breadcrumb":{"home":"Home","useCases":"Use Cases","spanishVeterinary":"For Spanish-speaking vets"},"metadata":{"title":"Free Spanish Veterinary Dictation Software - OpenWhispr","description":"OpenWhispr turns Spanish veterinary dictation into text for QVET, Provet Cloud, or any clinical record, with free local models available.","keywords":["spanish veterinary dictation software","veterinary dictation software spanish","spanish vet dictation","veterinary transcription spanish","dictado veterinario en español","software de dictado veterinario"]},"schema":{"name":"OpenWhispr","applicationCategory":"Spanish veterinary dictation software","operatingSystem":"macOS, Windows, Linux","description":"Free, open-source dictation software for veterinary teams who document in Spanish. Select Spanish, dictate, review the transcript, and paste it into the clinical record you already use.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Spanish veterinary dictation software","title":"Dictate in Spanish. Get text ready for your PIMS.","body":"Put the cursor in the historia clínica, press the hotkey, and dictate in Spanish. Review the text before you file it.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ happy users","image":{"label":"Veterinarian working at a clinic computer beside a dog"}},"trustBar":{"label":"Use alongside:","names":["QVET","Provet Cloud","GVET","WinVet","Wirevet","Microsoft Word"]},"pain":{"label":"The problem with Spanish clinical documentation","title":"The consult ends. The historia clínica is still blank.","subtitle":"You still owe the anamnesis, exploración, diagnóstico, tratamiento, informe de alta, and sometimes a prescription. Every line still needs your review.","cards":[{"title":"The Spanish note starts after the patient leaves.","body":"The findings are fresh, but the historia clínica is empty. Dictate the note while you still remember the sequence, then review it before filing."},{"title":"Drug names, doses, and accents need a final check.","body":"Select Spanish before you dictate. Store recurring patient names, drug names, and clinic abbreviations in the custom dictionary, then check every dose and clinical term before filing."},{"title":"Your PIMS does not need a new integration.","body":"OpenWhispr does not connect to QVET, Provet Cloud, GVET, WinVet, Wirevet, or Word. It produces text you can paste into a standard editable field, so you keep the workflow you already use."},{"title":"A prescription still needs a veterinarian.","body":"Dictation can type the details. It cannot examine the animal, choose treatment, validate required fields, or sign the prescription. You stay responsible for reviewing the record in the approved system."}]},"features":{"label":"What you get","title":"Pick Spanish. Keep your PIMS.","subtitle":"Select Spanish, press the hotkey, then review what lands in the field. No migration and no new record system.","items":[{"image":{"label":"Veterinarian examining a cat beside an open laptop"},"title":"Spanish is an actual speech setting","body":"Select Spanish during setup or in settings. The shipped app lists Spanish for local Whisper and Parakeet models as well as supported cloud transcription."},{"image":{"label":"Veterinarian using a laptop beside a cat in a carrier"},"title":"From the microphone to a PIMS-ready transcript","body":"Use the system-wide hotkey in a clinical history, discharge, referral, email, or Word document. Automatic paste is a setting and needs the required OS permission. The clipboard option keeps text available for manual paste."},{"image":{"label":"Man using a laptop in a stable tack room"},"title":"A local option when the connection drops","body":"Download a local model once, then Spanish dictation can run on the computer without internet. Local models are unlimited and included on every plan."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"New PIMS required","value":"None","footnote":"paste into the veterinary record system you already use"},{"label":"Cost to start","value":"$$0","footnote":"unlimited local models are included on the Free plan"},{"label":"Setup takes","value":"5 min","footnote":"from download to the first Spanish dictation"}]},"howItWorks":{"label":"How it works","title":"From spoken Spanish to the right clinical field.","subtitle":"OpenWhispr handles the dictation step. You keep the PIMS, the clinical judgment, and the final review.","image":{"label":"Veterinarian using a laptop while comforting a dog on an exam table"},"steps":[{"title":"Download for free.","body":"Install on Mac, Windows, or Linux. Select Spanish as a spoken language."},{"title":"Put the cursor in the record.","body":"Open the right QVET, Provet Cloud, GVET, WinVet, Wirevet, or Word field, then press the hotkey and speak."},{"title":"Review every clinical term.","body":"Check patient names, drug names, doses, accents, and required fields before you save or sign anything."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons Spanish-speaking vets choose OpenWhispr","subtitle":"Spanish is a speech setting in the app. The hotkey, local models, dictionary, and automatic-paste or clipboard workflow stay the same across your desktop.","items":[{"title":"Spanish appears in the language picker","body":"Spanish is a selectable spoken language in the shipped onboarding and settings flow."},{"title":"Two local model families list Spanish","body":"The app's model registry includes Spanish for local Whisper and multilingual Parakeet transcription."},{"title":"No dedicated PIMS integration required","body":"If a PIMS field accepts pasted text, use automatic paste or the clipboard. No dedicated QVET, Provet Cloud, GVET, WinVet, or Wirevet integration is claimed."},{"title":"Use the sections your team already knows","body":"Dictate anamnesis, exploración, diagnóstico, tratamiento, evolución, or informe de alta without changing the record's structure."},{"title":"Keep recurring terms in one list","body":"The custom dictionary stores the patient names, drug names, procedures, and abbreviations you add. You still verify each result."},{"title":"Local Spanish dictation can work offline","body":"After the local model is downloaded, transcription runs on the computer without an internet connection."},{"title":"Unlimited local models cost $0","body":"The Free plan includes unlimited local models, Spanish support, and the custom dictionary."},{"title":"Mac, Windows, and Linux","body":"Use the same Spanish dictation workflow across all three supported desktop platforms."},{"title":"The cleanup instruction is set for Spanish","body":"The instruction asks for Spanish grammar, spelling, and punctuation while keeping naturally spoken English technical terms. Review the result before filing."},{"title":"It does not pretend to prescribe","body":"OpenWhispr types what you dictate. The veterinarian still examines the animal, chooses treatment, checks the required fields, and signs in the approved system."}]},"faq":{"label":"Before you download","title":"The questions Spanish-speaking vets ask first.","subtitle":"Straight answers on Spanish support, terminology, PIMS workflow, offline use, prescriptions, and cost.","items":[{"question":"Does OpenWhispr actually support Spanish dictation?","answer":"Yes. Spanish is a selectable speech language in the shipped app. The code lists it for local Whisper, local multilingual Parakeet, and supported cloud transcription. This page does not claim equal accuracy for every accent or vocabulary."},{"question":"Will it recognize every veterinary term and drug name?","answer":"No blanket accuracy claim. The closed product sources do not document veterinary-specific or Spanish-veterinary tuning for the included local models. Add recurring terms to the custom dictionary, then review patient names, drugs, doses, abbreviations, and diagnoses before filing."},{"question":"Does it work with QVET, Provet Cloud, GVET, WinVet, and Wirevet?","answer":"No dedicated integration is claimed. If the field accepts pasted text, press the system-wide hotkey and dictate. Automatic paste needs the relevant setting and OS permission; the clipboard option supports manual paste."},{"question":"Can Spanish veterinary dictation work without internet?","answer":"Yes, in local mode after you download a model. Local models run on the computer with no internet required. Cloud transcription still needs a connection."},{"question":"Can OpenWhispr create or submit a veterinary prescription?","answer":"No. It can type the details you dictate into the active field, but it does not prescribe, validate a jurisdiction's required fields, submit to a prescription platform, or sign on your behalf. The veterinarian reviews and completes that workflow."},{"question":"Is it actually free?","answer":"Yes. The Free plan is $0 and includes unlimited local models, Spanish support, the custom dictionary, and Mac, Windows, and Linux. OpenWhispr Cloud on Free is limited to 2,000 words a week; Pro is $8 a month and makes cloud transcription unlimited."}]},"reviews":{"label":"What users say","title":"Real users. Their words, unchanged.","subtitle":"These reviews are not presented as veterinary endorsements. They describe the dictation experience the page is built on.","score":{"value":"ES","label":"Selectable speech language","body":"Spanish appears in the shipped language registry for local and cloud transcription."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for Spanish-speaking veterinary teams","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Vets and techs who want unlimited local Spanish dictation and are fine with a quick model download"}]},{"parts":[{"text":"Vets who want unlimited OpenWhispr Cloud transcription in Spanish"}]},{"parts":[{"text":"Veterinary teams that want unlimited recordings and advanced workflows"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Spanish dictation with local models at $0"}]},{"parts":[{"text":"Unlimited cloud transcription without downloading a local model"}]},{"parts":[{"text":"Unlimited recordings plus Agent mode and chat over your data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" stores terms you add"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" stores terms you add"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" stores terms you add"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"Spanish should be a setting, not a workaround.","paragraphs":["OpenWhispr gives veterinary teams a direct path from spoken Spanish to the clinical field already open on screen. The app handles dictation. The veterinarian keeps the terminology review, clinical judgment, and signature.","We build in public. Every release and every line of the app is on GitHub, where 5,000+ engineers have starred and forked the project. So when we say the dictation can stay on your machine, nobody in your practice has to take our word for it. They can read the code.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Put the cursor in the historia clínica. Dictate in Spanish.","body":"Review the text while the case is still fresh, then file it in the system you already use.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCasePhysicians":{"breadcrumb":{"home":"Home","useCases":"Use Cases","physicians":"For Physicians"},"metadata":{"title":"Free Physician Dictation Software for EHR Notes - OpenWhispr","description":"OpenWhispr is free physician dictation software for HPI, exam and assessment text at your EHR cursor, with on-device transcription.","keywords":["physician dictation software","medical dictation for physicians","EHR dictation software","clinical note dictation","voice recognition for physicians","offline physician dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Physician dictation software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source physician dictation software with optional on-device transcription. Put HPI, exam and assessment text at the active cursor with unlimited local dictation at no cost.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Physician dictation software for clinical notes","title":"Press the hotkey. Fill the chart field you're in.","body":"Dictate the HPI, exam, assessment and plan into the EHR field you already opened. Choose a local model so transcription runs on your computer. OpenWhispr is free and open source.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Physician dictating at a computer in an exam room"}},"trustBar":{"label":"Physicians chart in:","names":["Epic","Oracle Health","athenaOne","eClinicalWorks","MEDITECH","Microsoft Word"]},"pain":{"label":"The problem with charting by keyboard","title":"The visit ended. The chart did not.","subtitle":"HPI, exam, assessment, plan, inbox. Every empty field is another piece of the visit waiting to be typed.","cards":[{"title":"Open charts stack up between patients.","body":"The encounter is over, but the HPI and assessment are still blank. Dictate while the details are fresh."},{"title":"Every field breaks your train of thought.","body":"Click HPI, type. Click exam, type. Click assessment, type again. OpenWhispr leaves the EHR open and puts the transcript at the active cursor."},{"title":"One drug name becomes five corrections.","body":"Generic dictation trips over medication names, eponyms and local abbreviations. Add the terms you use to OpenWhispr's custom dictionary."},{"title":"A five-minute trial needs an IT ticket.","body":"You want to test dictation on one chart, not provision a practice. Local mode starts with a desktop download and no account."}]},"features":{"label":"What you get","title":"Less keyboard. Fewer open charts.","subtitle":"Physician dictation software should fit the note template already on screen. OpenWhispr turns speech into text at your cursor without making you rebuild the workflow around it.","items":[{"image":{"label":"Physician dictating into a desktop microphone at a workstation"},"title":"HPI and assessment at the active cursor","body":"Open the note, place the cursor and press the hotkey. Your dictation is transcribed and pasted into that field, ready for review."},{"image":{"label":"Physician holding a medication bottle beside a computer"},"title":"Your medication names in the dictionary","body":"Add the drugs, procedures, clinician names and abbreviations that keep coming back wrong. OpenWhispr passes those terms to transcription."},{"image":{"label":"Clinician dictating beside a laptop inside a van"},"title":"Keep dictating when ward Wi-Fi drops","body":"Download a local model once. Then transcribe on the workstation in clinic, at home or on rounds without an internet connection."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Audio uploaded in local mode","value":"None","footnote":"in local mode, transcription runs on the physician's device"},{"label":"Cost per physician","value":"$$0","footnote":"unlimited local dictation on every desktop plan"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first dictated chart field"}]},"howItWorks":{"label":"How it works","title":"Finish the chart you already have open.","subtitle":"Keep the EHR workflow you already know. OpenWhispr adds a hotkey and local transcription, with no account or card required to start.","image":{"label":"Physician working on a laptop at night"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Choose a local model during setup."},{"title":"Put the cursor in the chart.","body":"Open the HPI, exam or assessment field, press the hotkey and dictate."},{"title":"Review and sign.","body":"Read the inserted text, correct what needs correcting, then finish the note in the EHR."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons physicians choose OpenWhispr","subtitle":"Try one workstation, one chart field and one local model. These are the ten things to check.","items":[{"title":"No copy-and-paste detour","body":"Dictate inside the open chart field. Review the inserted text there, then continue the note."},{"title":"One hotkey across desktop apps","body":"Use the same dictation control in the EHR, a portal reply, an email or a document."},{"title":"A dictionary for your clinical vocabulary","body":"Add drug names, procedures, eponyms and abbreviations as recognition hints for supported transcription and cleanup paths."},{"title":"Local mode stays on the workstation","body":"Choose a local model and dictation is transcribed on the device. Cloud fallback is a separate opt-in setting."},{"title":"Works fully offline","body":"Once the local model is downloaded, dictation keeps working without internet in clinic, at home or on a ward."},{"title":"Unlimited local dictation at $0","body":"Every desktop plan includes unlimited use of local speech-to-text models."},{"title":"Mac, Windows and Linux","body":"Use it on the desktop operating system your workstation already runs."},{"title":"Open source, inspectable by IT","body":"Your organization can review the code on GitHub instead of taking the local-mode claim on faith."},{"title":"100+ languages","body":"OpenWhispr supports 100+ languages. Availability varies by local model and cloud provider."},{"title":"Local or cloud dictation in Settings","body":"Choose Local or Cloud for dictation during setup, and change that dictation setting later when your workflow changes."}]},"faq":{"label":"Before you download","title":"What physicians ask before changing dictation software.","subtitle":"Straight answers on EHR fields, terminology, patient data, offline use and cost.","items":[{"question":"What is the best physician dictation software for EHR notes?","answer":"If you want a hotkey that puts the transcript at your active cursor, unlimited local dictation at $0, and code your IT team can inspect, try OpenWhispr on one workstation. It does not replace the physician's review or the EHR's signing workflow."},{"question":"Does OpenWhispr integrate with my EHR?","answer":"OpenWhispr is not an EHR integration. It transcribes your speech and pastes the text at the active cursor, so you can use it in editable chart fields and other desktop apps. Test it with your organization's EHR and virtual-desktop setup before a wider rollout."},{"question":"How does it handle medication names and clinical abbreviations?","answer":"OpenWhispr includes a custom dictionary. Add the medication names, procedures, eponyms and abbreviations you use, and those terms are supplied to transcription. You still review the note before signing."},{"question":"Can I dictate patient information?","answer":"Your organization decides which tools are approved for patient information. In local mode, transcription runs on the device without uploading the audio. Confirm your organization's policy before use."},{"question":"Will it work when the hospital network or Wi-Fi does not?","answer":"Yes, after you download a local model. Local transcription runs on Mac, Windows and Linux without internet. Cloud transcription remains a separate mode you can choose in Settings."},{"question":"Is it actually free?","answer":"Yes. Every desktop plan includes unlimited local dictation, a custom dictionary and 100+ languages at $0. Pro is $8 per month and adds unlimited cloud transcription and the iPhone and iPad app."}]},"reviews":{"label":"What users say","title":"What users notice when voice replaces the keyboard.","subtitle":"Verbatim notes from our Discord, GitHub and inboxes.","score":{"value":"0","label":"Per-physician licences for local dictation","body":"Unlimited local dictation is included at $0 on every desktop plan."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Start free with local dictation.","Upgrade for cloud or team features."],"caption":"Compare OpenWhispr plans for physicians","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Physicians who want free offline dictation at the active cursor and can set up a local model"}]},{"parts":[{"text":"Physicians who want unlimited cloud transcription and mobile dictation"}]},{"parts":[{"text":"Practices and teams documenting at volume"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited on-device dictation with no per-physician fee"}]},{"parts":[{"text":"Unlimited cloud dictation without downloading a local model"}]},{"parts":[{"text":"Unlimited recordings and priority support for a practice"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"The path from microphone to chart should be inspectable.","paragraphs":["Physicians should not have to guess what a dictation tool does between the microphone and the chart. OpenWhispr makes that path visible in public code.","Your organization can inspect the local-mode implementation on GitHub instead of relying only on a privacy page.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"OpenWhispr founder and community"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Finish the chart before you open the next one.","body":"Review the inserted text, sign the note and move on.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, ask us for help in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseDoctors":{"breadcrumb":{"home":"Home","useCases":"Use Cases","doctors":"For Doctors"},"metadata":{"title":"Free Doctor Dictation Software, No Middleman - OpenWhispr","description":"Doctor dictation software with no middleman: pick a local model and OpenWhispr types your letters on your own machine in seconds. Free and open source.","keywords":["doctor dictation software","physician dictation software","dictation app for doctors","voice to text for doctors","referral letter dictation","offline dictation for physicians"]},"schema":{"name":"OpenWhispr","applicationCategory":"Physician dictation software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Local-first, open-source dictation software for doctors. With a local model chosen, referral letters, reports and notes are transcribed on the doctor's own device and never leave it. Free for unlimited local dictation, and works fully offline.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation software for doctors & physicians","title":"Speak the referral. Send it before the patient leaves.","body":"Choose a local model at setup and the referral, report or patient letter is typed and drafted on your own computer. No transcription service, no per-seat licence, no one else hearing the audio.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ happy users","image":{"label":"Doctor speaking into a desk microphone at a computer"}},"trustBar":{"label":"Dictate into:","names":["DrChrono","Practice Fusion","Tebra","AdvancedMD","Outlook","Microsoft Word"]},"pain":{"label":"The problem with dictation services","title":"You already dictate. The middleman is the problem.","subtitle":"Doctors had dictation figured out decades ago. What's broken is the channel: a service that bills per line and returns the letter tomorrow, or a licence that bills per doctor forever.","cards":[{"title":"A stranger hears the patient's history.","body":"Send a dictation to a transcription service and someone you have never met hears the patient's name, history and diagnosis. In local mode, OpenWhispr transcribes on your own machine."},{"title":"The letter comes back tomorrow.","body":"Outsourced transcription returns your referral in a day or two, and then it still needs checking. OpenWhispr types it the moment you stop talking, so the referral goes out in the same sitting."},{"title":"A licence fee for every doctor, every year.","body":"Medical dictation suites are sold per seat, per year, and the quote climbs at renewal. OpenWhispr's local dictation is $0 for every doctor in the practice, forever."},{"title":"The code is the confidentiality agreement.","body":"Closed tools hand you a privacy policy and ask for faith. OpenWhispr's code is public on GitHub, so anyone in the practice can verify that in local mode, patient audio never leaves the machine."}]},"features":{"label":"What you get","title":"Instant turnaround. Nobody else in the loop.","subtitle":"Doctor dictation software used to mean a slow service or a per-seat licence. Keep the typed letter and drafted report without sending dictation through a service or buying another seat.","items":[{"image":{"label":"Doctor using a desktop computer in an office"},"title":"Typed on your machine, in seconds","body":"Dictate the referral, report or reply into your mic. The words land on your screen seconds after you stop talking, transcribed locally and never uploaded."},{"image":{"label":"Doctor writing in a notebook beside a computer"},"title":"The referral, already in letter form","body":"Speak the gist and get back a structured referral, report or patient instruction sheet. You read it, fix a word, and send it. The typist never existed."},{"image":{"label":"Doctor typing at a clinical workstation"},"title":"One hotkey. Any window, any EHR.","body":"Press the hotkey and speak into whatever is open: the chart, a reply to a colleague, a report in Word. The text lands at your cursor, punctuated, processed locally."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Audio sent to a transcription service","value":"None","footnote":"in local mode, dictation is transcribed on your device and goes nowhere else"},{"label":"Cost per doctor","value":"$$0","footnote":"unlimited local dictation with no per-doctor licence"},{"label":"Setup takes","value":"5 min","footnote":"install it yourself, no IT ticket, no contract"}]},"howItWorks":{"label":"How it works","title":"Turnaround time: the length of the dictation.","subtitle":"Old dictation goes out to a service and comes back tomorrow. Local mode runs on your machine, with no account or card required to start.","image":{"label":"Doctor opening a laptop at a desk"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Choose a local model during the five-minute setup."},{"title":"Dictate as you always have.","body":"Speak the referral, report or note. It's typed on your device seconds after you stop."},{"title":"Review, send, delete.","body":"Check the draft, paste it into the EHR or an email, and delete the audio. You keep control of all of it."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons doctors choose OpenWhispr","subtitle":"Your dictation carries a patient's history, so check the code and the settings, not just the brochure.","items":[{"title":"No transcription service in the loop","body":"The recording stays on your device instead of going out for overnight transcription."},{"title":"The audio has nowhere to go","body":"Local mode has no upload step. The recording is transcribed on your machine and stays there."},{"title":"Free for every doctor in the practice","body":"Unlimited local dictation at $0, every seat, forever. No per-doctor licence to renew each year."},{"title":"Works fully offline","body":"Home visits, a locum shift on a ward with no signal, a flight to a conference. Once the model is on your machine, dictation never stalls."},{"title":"Types into the chart, the inbox and Word","body":"One hotkey puts the text at your cursor: a referral template, a colleague's inbox, the record itself."},{"title":"Letters and reports, not just transcripts","body":"Run a local AI model, or connect your own key, to turn a dictated summary into a formatted referral, report or patient letter."},{"title":"Your letters are your files","body":"Recordings and drafts sit on your disk. Keep, encrypt or shred them on your schedule, not a vendor's retention policy."},{"title":"Auditable by whoever runs your IT","body":"Your IT team can inspect the public code instead of relying only on a vendor claim."},{"title":"Learns your drug names and eponyms","body":"A custom dictionary picks up the drugs, procedures and abbreviations your specialty uses, so they come out spelled right."},{"title":"100+ languages","body":"Dictate in the language you think in, transcribed locally. Useful when the note is in one language and the referral goes out in another."}]},"faq":{"label":"Before you download","title":"What doctors ask before switching dictation software.","subtitle":"Confidentiality, cost, accuracy, answered straight. Anything else, ask in Discord; a human replies within a day.","items":[{"question":"What's the best dictation software for doctors?","answer":"For doctors who want the letter typed the moment they stop talking, at $0 per doctor, with patient audio staying on the machine in local mode: OpenWhispr. It's open source, works offline, and drafting is built in, with a local AI model like Qwen or Llama or your own AI key. If you currently pay a transcription service or a per-seat licence, this is the thing that replaces it."},{"question":"Who hears my dictation?","answer":"In local mode, no one. It's transcribed on your machine and the open-source code shows there is no upload step. Cloud transcription is the other mode, one global setting you pick at setup or in Settings: audio is sent only to be transcribed and never stored."},{"question":"Can I use it with patient information under HIPAA?","answer":"In local mode, patient audio is transcribed on your device and never reaches a third party, so there is no vendor to cover with a BAA. For cloud transcription of PHI, OpenWhispr signs a BAA on the Business plan and up, and holds a SOC 2 Type 2 report and ISO 27001 certification. Whether your workflow is compliant still depends on how you use it; no tool can grant that on its own."},{"question":"Will it get the drug names right?","answer":"It runs Whisper-class models that handle a wide range of accents, and the custom dictionary learns the drug names, procedures and abbreviations your specialty uses. Add a term once and it comes out spelled right from then on."},{"question":"Does it work on home visits, or on a ward with no signal?","answer":"Yes. Once the local model is on your machine it runs entirely offline. Dictate at the bedside or in the car between visits, and send the letter when you're back in range. Cloud dictation tools stall the moment you lose signal."},{"question":"Is it really free for the whole practice?","answer":"Yes. Unlimited local dictation on Mac, Windows and Linux, in 100+ languages, at $0, for every doctor in the practice. Pro at $8/month adds unlimited cloud transcription and the iPhone app when you want them."}]},"reviews":{"label":"What users say","title":"Nobody here mails their voice to a stranger either.","subtitle":"From our Discord, GitHub and inboxes.","score":{"value":"0","label":"Consults heard by a transcriber","body":"In local mode the audio never leaves your device. No server stores it, and no transcriber ever hears it."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for doctors","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Doctors who want free, private, offline dictation and can spare five minutes for a local setup"}]},{"parts":[{"text":"Doctors who want easy, reliable cloud transcription and mobile dictation"}]},{"parts":[{"text":"Practices documenting at volume, or that need a signed BAA"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Free offline dictation, with patient audio staying on your device"}]},{"parts":[{"text":"Unlimited cloud transcription with no local models to manage"}]},{"parts":[{"text":"Unlimited recordings plus advanced workflow features for a practice"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"A patient's history is not training data.","paragraphs":["OpenWhispr started with a simple objection: turning speech into text shouldn't mean the audio lands on someone else's server, or in someone's training set. A patient's history least of all.","We build in public on GitHub. When we say local-mode dictation stays on your device, your practice can inspect the code instead of relying only on a privacy policy.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Keep dictating. Just stop paying someone to type it.","body":"Your next referral goes from your voice to the recipient without passing through anyone else's hands. No service, no seat licence, no waiting.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseProductivity":{"breadcrumb":{"home":"Home","useCases":"Use Cases","productivity":"For Productivity"},"metadata":{"title":"Free Dictation App For Productivity - OpenWhispr","description":"OpenWhispr is a free, open-source dictation app for productivity. Press one hotkey to put tasks, notes and replies in the app already open.","keywords":["dictation app for productivity","productivity dictation app","voice typing for productivity","speech to text productivity app","dictate tasks and notes","offline dictation app"]},"schema":{"name":"OpenWhispr","applicationCategory":"Productivity dictation app","operatingSystem":"macOS, Windows, Linux, iOS","description":"Local-first, open-source dictation app for productivity. Press a system-wide hotkey and speak tasks, notes and replies into the app already open. With local models selected, transcription and clean-up run on your own computer and work offline. Unlimited local dictation is free on macOS, Windows and Linux.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation app for productivity","title":"Say the task. Get it into the list.","body":"Capture the thought before it disappears, right in the app you already have open; no separate inbox to check later. Use OpenWhispr Cloud for fast, accurate transcription, or local models to keep the audio on your machine.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman speaking while working at a laptop beside a notebook and material samples"}},"trustBar":{"label":"Dictate into the app already open, including:","names":["Todoist","Notion","Obsidian","Gmail","Google Docs","Slack"]},"pain":{"label":"The capture problem","title":"The thought is quick. Typing it down is the slow part.","subtitle":"A task arrives while you are already doing something else. The longer capture takes, the more likely the thought stays in your head.","cards":[{"title":"The task arrives in the middle of another task.","body":"You are in Slack when you remember the invoice, or in a doc when you spot tomorrow's follow-up. Put the cursor in your task list, press the hotkey and say it before the thought disappears."},{"title":"The note is longer than the time you have.","body":"A useful idea becomes three thin bullet points because typing the whole thought feels like work. Talk it into Notion or Obsidian instead. OpenWhispr removes filler and false starts, then pastes the cleaned text at the cursor."},{"title":"Five two-minute replies become an hour of avoidance.","body":"The answer is already in your head. Press the same hotkey in Gmail or Slack, say the reply once and review the text before you send it. No separate recorder to empty later."},{"title":"Your system has names only your system understands.","body":"Project codenames, client names, acronyms and recurring places get mangled by generic dictation. Add them to the custom dictionary once so the transcription and clean-up get the spellings you use."}]},"features":{"label":"One hotkey wherever you type","title":"Keep the system you built. Replace the typing.","subtitle":"One system-wide hotkey works wherever the cursor already is. Choose local models and it keeps working without a connection.","items":[{"image":{"label":"Man speaking into a desk microphone beside a monitor and open notebook"},"title":"Your apps stay exactly where they are","body":"Todoist for tasks, Notion for projects, Obsidian for notes, Gmail for replies. OpenWhispr types into all four. It pastes the finished text into the active field, so there is no new inbox to process and no integration to maintain."},{"image":{"label":"Woman speaking beside a laptop and architectural model"},"title":"A false start does not become a messy task","body":"Say 'Thursday, no, Friday' and the clean-up keeps the correction. It removes filler, fixes punctuation and turns spoken dates into written ones while keeping your wording and intent."},{"image":{"label":"Man working on a laptop and notebook aboard a train"},"title":"The capture habit still works offline","body":"Download local speech-to-text and a local clean-up model once. Then tasks, notes and replies can be transcribed and tidied on your computer with no connection and no per-word bill."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"AI processing location","value":"On your device","footnote":"when local transcription and local clean-up models are selected"},{"label":"Cost to capture locally","value":"$$0","footnote":"unlimited local dictation on every desktop you use"},{"label":"Local setup","value":"A few steps","footnote":"download, choose models, set the hotkey and start speaking"}]},"howItWorks":{"label":"Download to first captured task","title":"Three steps from a thought to the right inbox.","subtitle":"Nothing to migrate and no productivity system to rebuild.","image":{"label":"Woman typing on a laptop backstage beside camera equipment"},"steps":[{"title":"Download it free.","body":"Install OpenWhispr on Mac, Windows or Linux. Choose OpenWhispr Cloud for speed, or download local models for offline use."},{"title":"Put the cursor where the thought belongs.","body":"Open the task field, note, email or message you already use. Press your hotkey and say the whole thought, including the context you would normally skip."},{"title":"Review it and move on.","body":"OpenWhispr cleans the wording and pastes it at the cursor. Check the text, save or send it, and get back to the work you were doing."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Why it earns a place in the stack","title":"10 reasons people use OpenWhispr in the apps they already use","subtitle":"Each one removes friction from capture without asking you to replace the apps that already work.","items":[{"title":"One hotkey across your apps.","body":"Tasks in Todoist, project notes in Notion, linked thoughts in Obsidian, replies in Gmail and Slack. If automatic paste is unavailable, the text stays on the clipboard."},{"title":"No new inbox to empty.","body":"The words land in the tool where they belong. You do not record a voice memo now and remember to move it later."},{"title":"You approve the text before it becomes a task.","body":"The cleaned text lands in the field first. Fix a name, date or label before you save or send it."},{"title":"Dates look like dates.","body":"Spoken dates, times, numbers and currency are converted to standard written forms before the text reaches your task, note or reply."},{"title":"Your project vocabulary follows you.","body":"Add client names, codenames, acronyms and product terms to the custom dictionary. The same list helps across the apps you dictate into."},{"title":"Unlimited local dictation is free.","body":"No trial clock, no word cap and no card on every desktop plan when you run the speech-to-text model on your own machine."},{"title":"Offline means the capture habit does not break.","body":"Local transcription and a local clean-up model keep working on a plane, during an outage or anywhere the connection drops."},{"title":"100+ languages.","body":"Capture the thought in the language you actually think in. The same app and the same hotkey support more than 100 languages."},{"title":"The privacy claim is code you can inspect.","body":"OpenWhispr is open source. You can inspect the local transcription and paste paths yourself."},{"title":"Mac, Windows and Linux.","body":"Use the same free app on the desktop at work and the laptop at home, with a hotkey you choose for each machine."}]},"faq":{"label":"Before you add voice input to your productivity stack","title":"What people ask before they put OpenWhispr in their stack","subtitle":"What it does, what it does not do and what stays on your machine.","items":[{"question":"Does OpenWhispr create and schedule tasks for me?","answer":"OpenWhispr types finished text where your cursor is. It does not organise or schedule tasks. Todoist, Notion and other apps still apply their own date and label rules, and you review the text before saving."},{"question":"Do I need an integration for every productivity app?","answer":"No. The system-wide hotkey pastes text into the active app, so there is no Todoist plugin, Notion connection or Gmail extension to set up. On Mac, OpenWhispr needs accessibility permission to paste automatically. If automatic paste fails, the transcription remains available to paste from the clipboard."},{"question":"Will it clean up a rambling thought?","answer":"Yes, when clean-up is enabled. The clean-up prompt removes filler words, false starts, stutters and accidental repetition. It fixes grammar and punctuation, converts dates and times to written forms and is explicitly told to preserve your wording, formality and intent."},{"question":"Can OpenWhispr process my dictation locally?","answer":"Yes. Choose local transcription and local clean-up so audio and transcript processing stay on your computer. Once text is pasted, the destination app's own storage and privacy rules apply. OpenWhispr Cloud audio is transcribed and discarded, not stored."},{"question":"Is unlimited dictation really free?","answer":"Unlimited local dictation is $0 on every desktop plan and does not expire. The Free plan also includes 2,000 words a week through OpenWhispr Cloud. Pro is $8 per user each month for unlimited cloud transcription, and Business is $16 per user each month for unlimited cloud meeting recordings and agent mode."},{"question":"Will it spell project names and acronyms correctly?","answer":"Add the exact spellings to the custom dictionary. OpenWhispr passes that vocabulary into supported transcription paths and the clean-up step so client names, project codenames and internal acronyms have more context. You can edit the dictionary any time."}]},"reviews":{"label":"What people tell us","title":"Better than typing. Clean after a stumble. Local without WiFi.","subtitle":"Read them for the tool, not for a productivity method.","score":{"value":"0","label":"Productivity apps to replace","body":"OpenWhispr is the voice input across the stack you already use. Your tasks, notes, docs, email and chat stay where they are."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for productivity","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"People who want unlimited local dictation across their personal productivity stack for $0"}]},{"parts":[{"text":"People who want unlimited cloud dictation without managing local models"}]},{"parts":[{"text":"Teams that want agent mode and unlimited cloud meeting recordings alongside dictation"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Capture tasks, notes and replies locally with no word cap, trial clock or card"}]},{"parts":[{"text":"Unlimited OpenWhispr Cloud transcription with no local model to download"}]},{"parts":[{"text":"Unlimited cloud meeting recordings and agent mode for shared workflows"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"A productivity system is only useful if the thought gets into it.","paragraphs":["OpenWhispr turns speech into clean text where the thought belongs: your task list, notes or reply field.","OpenWhispr is open source, so you can inspect the local transcription and paste paths yourself. With local models selected, dictation can stay on your machine.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"$$0 · open source · your apps stay put","title":"Press the hotkey. Get the thought into the app already open.","body":"Download OpenWhispr, choose a model and dictate into your task list, notes or reply field. No card and no migration.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Unlimited local dictation is free on Mac, Windows and Linux."}}},"useCasePathology":{"breadcrumb":{"home":"Home","useCases":"Use Cases","pathology":"For Pathologists"},"metadata":{"title":"Pathology Dictation Software for Reports - OpenWhispr","description":"OpenWhispr is free pathology dictation software for macroscopic descriptions and microscopic findings, with auto-paste in supported report fields.","keywords":["pathology dictation software","pathologist dictation software","pathology report dictation","anatomic pathology dictation","histopathology dictation software","pathology speech recognition","offline pathology dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Pathology dictation software","operatingSystem":"macOS, Windows, Linux","description":"Local-first, open-source pathology dictation software that transcribes spoken report text for automatic paste into supported fields. Free for unlimited local dictation.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Pathology dictation software","title":"Dictate the findings. Get report text at your cursor.","body":"Keep the specimen or slide in view. Dictate the macroscopic description, microscopic findings or final diagnosis. OpenWhispr types into supported fields. You review and sign out.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Laboratory professional reviewing a computer beside a microscope and slide tray"}},"trustBar":{"label":"Reporting environments to test with OpenWhispr:","names":["Epic Beaker","Oracle Health PathNet","Clinisys PowerPath","Clinisys WinPath","Sectra Digital Pathology","Microsoft Word"]},"pain":{"label":"The problem with typing pathology reports","title":"The slide is in focus. The keyboard pulls you away.","subtitle":"The finding is clear. The field is empty. Every trip to the keyboard interrupts the read.","cards":[{"title":"Your eyes leave the slide for the keyboard.","body":"You see the finding, then look away to type it. OpenWhispr lets you speak the report text while the morphology is still in view."},{"title":"The gross description waits on your hands.","body":"Measurements, margins, sections and the block key arrive in sequence at the cut-up bench. Dictate each observation while it is in front of you, then review the text before it enters the final report."},{"title":"Structured fields still need precise prose.","body":"Synoptic fields capture element-response pairs. Free-text microscopic descriptions, comments and addenda can still need your words. OpenWhispr transcribes those words without building or validating the protocol."},{"title":"One unfamiliar term becomes a correction loop.","body":"Specimen labels, eponyms and local abbreviations recur case after case. Add them to the custom dictionary as context hints, then check every occurrence before sign-out."}]},"features":{"label":"What you get","title":"Keep the case in view. Put the words in the report.","subtitle":"Pathology dictation software should fit the reporting screen already open. OpenWhispr turns speech into text at the active cursor. It does not replace your LIS, protocols or judgment.","items":[{"image":{"label":"Person typing beside magnified tissue images on two monitors"},"title":"Report text at the active cursor","body":"Open an editable report field, place the cursor and press the hotkey. OpenWhispr pastes the transcript when the field and session support automatic paste. Otherwise it copies the text for manual paste."},{"image":{"label":"Laboratory professional using a desktop microphone at a workstation"},"title":"Local transcription when you choose it","body":"Download a local speech model and select Local in Settings. With cloud fallback disabled, dictation stays on the workstation and continues without internet after the model is installed."},{"image":{"label":"Laboratory professional holding a specimen tube and smartphone beside a microscope"},"title":"Your recurring terms as context hints","body":"Add specimen names, eponyms, stains and local abbreviations to the custom dictionary. OpenWhispr feeds them to supported speech models as hints. It does not promise specialist terminology accuracy."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Direct LIS integrations included","value":"None","footnote":"OpenWhispr uses active-cursor paste in compatible fields instead"},{"label":"Cost per pathologist","value":"$$0","footnote":"unlimited local dictation on every desktop plan"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first dictated report text"}]},"howItWorks":{"label":"How it works","title":"Findings in the report before you leave the case.","subtitle":"Speech becomes editable text in the report field. You check it against the case and complete sign-out in the reporting system.","image":{"label":"Healthcare worker in blue scrubs opening a laptop in a hospital corridor"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Pick a local model during setup."},{"title":"Dictate the findings.","body":"Put the cursor in an editable field, press the hotkey and speak the macroscopic description, microscopic findings, ancillary-study results, addendum or diagnosis text."},{"title":"Review, then sign out.","body":"Check the text against the specimen, slides, ancillary studies and required report elements. The final report remains your call."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons OpenWhispr fits pathology reporting","subtitle":"A pathology report carries exact observations and a final professional judgment. OpenWhispr stays in the narrower lane of turning your speech into editable text.","items":[{"title":"Automatic paste in supported fields","body":"Use the same hotkey in a compatible report field, an email or a document. If automatic paste is unavailable, OpenWhispr leaves the text on your clipboard."},{"title":"No direct LIS plug-in required","body":"Compatible editable fields use desktop dictation and automatic paste. OpenWhispr does not claim a direct integration with Epic Beaker, PathNet, PowerPath, WinPath or another LIS. Test your reporting system and virtual-desktop setup."},{"title":"A local transcription path","body":"Choose Local in Settings and keep cloud fallback disabled. The speech model runs on the workstation and the dictation audio is not uploaded for transcription on that path."},{"title":"Works offline after model setup","body":"Once a local model is downloaded, dictation continues without internet at the microscope, gross bench or home workstation."},{"title":"A dictionary for recurring terms","body":"Add specimen names, stains, eponyms and abbreviations as context hints. The pathologist still checks the transcript before sign-out."},{"title":"Unlimited local dictation at $0","body":"Every desktop plan includes unlimited use of local speech-to-text models."},{"title":"Mac, Windows and Linux","body":"Use OpenWhispr on the desktop operating system your reporting workstation already runs."},{"title":"Open source, inspectable by IT","body":"The desktop app code is public on GitHub, so your technical team can inspect the local and cloud transcription paths."},{"title":"100+ languages","body":"OpenWhispr supports 100+ languages. Availability varies by local model and cloud provider."},{"title":"Local or cloud in Settings","body":"Choose the transcription path during setup and change it later in Settings. Cloud audio is sent only to transcribe it and is never stored by OpenWhispr."}]},"faq":{"label":"Before you download","title":"What pathologists ask before changing dictation software.","subtitle":"Straight answers on reporting fields, terminology, diagnostic boundaries, local processing and cost.","items":[{"question":"Does OpenWhispr integrate with my LIS or digital pathology viewer?","answer":"OpenWhispr is not an LIS or viewer integration. It transcribes speech and pastes text at the active cursor, so you can use it in compatible editable fields and other desktop apps. Test it with your reporting system and virtual-desktop setup before wider use."},{"question":"Does it generate a diagnosis or complete a synoptic report?","answer":"No. OpenWhispr turns what you dictate into editable text. It does not inspect slides, interpret findings, choose a diagnosis, populate required protocol elements or sign out a case. The pathologist reviews every word in the reporting system."},{"question":"Will it recognize pathology terminology?","answer":"OpenWhispr includes a custom dictionary. Add recurring specimen names, stains, eponyms and abbreviations, and supported speech models receive them as context hints. This is not a guarantee of specialist terminology accuracy, so review every report before sign-out."},{"question":"Where is the dictation processed?","answer":"Choose Local and transcription runs on the workstation. Choose OpenWhispr Cloud and the audio is sent only for transcription, then never stored by OpenWhispr. Your laboratory decides which mode and workflow are approved for its data."},{"question":"Will it work without internet?","answer":"Yes, after you download a local speech model. Local transcription runs on Mac, Windows and Linux without internet. Cloud transcription remains a separate mode you can choose in Settings."},{"question":"Is it actually free?","answer":"Yes. Every desktop plan includes unlimited local dictation and a custom dictionary at $0. OpenWhispr supports 100+ languages across available models and providers; local-model availability varies. Pro is $8 per month and adds unlimited cloud transcription, more meeting recording hours, mobile and device sync."}]},"reviews":{"label":"What users say","title":"The dictation is fast. The judgment stays yours.","subtitle":"Existing OpenWhispr user feedback. These testimonials are about the tool, not pathology reporting.","score":{"value":"0","label":"Diagnostic decisions made by OpenWhispr","body":"OpenWhispr transcribes what you say. It does not interpret a specimen or slide, choose a diagnosis or sign out a report."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for pathologists","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Pathologists who want free offline dictation at the active report cursor"}]},{"parts":[{"text":"Pathologists who want unlimited OpenWhispr Cloud transcription"}]},{"parts":[{"text":"Laboratories and teams documenting at volume"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local report dictation at $0"}]},{"parts":[{"text":"Unlimited cloud transcription without managing local models"}]},{"parts":[{"text":"Unlimited recordings and priority support for a team"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users and OpenWhispr features"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" supplies names and terms as context hints"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" supplies names and terms as context hints"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" supplies names and terms as context hints"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" runs user-configured actions on transcripts"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic OpenWhispr team use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, SAML, SCIM, audit logs, retention controls, admin controls, or dedicated support for a larger team? These are enterprise security and data-governance options, not clinical or regulatory certification.","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"The report remains yours. OpenWhispr handles the typing.","paragraphs":["OpenWhispr handles the mechanical part: speech becomes editable text at your cursor. The interpretation, structured elements, final diagnosis and sign-out remain with the pathologist and the laboratory's reporting system.","OpenWhispr is open source under the MIT License, and its source is available on GitHub. When we say dictation can run locally, your laboratory can inspect the code.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"OpenWhispr founder portrait"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Finish the report while the case is still on screen.","body":"Put the cursor in the editable field, dictate the findings and review every word before sign-out.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work with your workstation, ask us for help in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseRadiology":{"breadcrumb":{"home":"Home","useCases":"Use Cases","radiology":"For Radiology"},"metadata":{"title":"Best Radiology Dictation Software, Free & Local - OpenWhispr","description":"OpenWhispr is free, open-source radiology dictation software that transcribes locally and pastes text at the active cursor for review.","keywords":["best radiology dictation software","radiology dictation software","radiologist speech recognition software","offline radiology dictation","local radiology transcription","radiology voice to text"]},"schema":{"name":"OpenWhispr","applicationCategory":"Radiology dictation software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Local-first, open-source dictation software for radiologists. Choose local speech-to-text and the audio is transcribed on the workstation, then text is pasted at the active cursor. Cleanup is configured separately; there is no native radiology-system integration.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Best radiology dictation software for imaging specialists","title":"Keep the images in view. Get dictated text at the cursor.","body":"Choose a local speech-to-text model, place the cursor in an editable report field, and speak. Audio is transcribed on the workstation; use local cleanup or turn cleanup off to keep the text there too.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Radiologist dictating beside chest scans at a reading station"}},"trustBar":{"label":"Common reporting tools, not integrations:","names":["PowerScribe One","Epic Radiant","Sectra IDS7","Visage 7","InteleViewer","Microsoft Word"]},"pain":{"label":"Where radiology dictation breaks","title":"The images need your eyes. The report still needs your words.","subtitle":"A radiology dictation tool has to respect the active field, the vocabulary and the final read. Miss one, and the keyboard comes back.","cards":[{"title":"Your eyes leave the study for the keyboard.","body":"Place the cursor once, then hold the hotkey and dictate the finding. OpenWhispr pastes the transcript into the active field so you can stay with the images."},{"title":"The same term comes back wrong every case.","body":"Add names, modalities, anatomy and local abbreviations to the custom dictionary. They are supplied as hints to supported transcription paths."},{"title":"One missing 'no' changes the report.","body":"Speech recognition can change clinical meaning. OpenWhispr supplies text, not clinical quality control, so the radiologist still proofreads before signing."},{"title":"A simple dictation test turns into an integration project.","body":"OpenWhispr is a desktop speech-to-text layer, not a PACS or RIS replacement. Try it in an editable field without changing the reporting stack."}]},"features":{"label":"What you get","title":"Cursor dictation, not another reporting platform.","subtitle":"OpenWhispr handles speech to text. Your worklist, images, templates, measurements and signing workflow stay in the systems that already own them.","items":[{"image":{"label":"Person working at two monitors beside a desktop computer"},"title":"Findings transcribed on the workstation","body":"Choose a local speech-to-text model and audio is processed on the device. Use local cleanup or turn cleanup off if report text must stay there too; after model download, speech-to-text works offline."},{"image":{"label":"Radiologist checking a reference card beside a medical scan"},"title":"A dictionary for the terms you actually dictate","body":"Add recurring anatomy, eponyms, protocol names and abbreviations. OpenWhispr supplies the list as recognition hints, and you still check the result."},{"image":{"label":"Radiologist typing beside monitors displaying medical scans"},"title":"One hotkey at the active report cursor","body":"Open the report field, place the cursor, hold the hotkey and speak. The transcript is pasted there for review, with no claim of a native PACS, RIS or EHR integration."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Dictated audio uploaded in local mode","value":"None","footnote":"with local-to-cloud fallback off, audio stays on the radiologist's workstation"},{"label":"Cost per radiologist","value":"$$0","footnote":"unlimited local dictation for every radiologist"},{"label":"Setup takes","value":"5 min","footnote":"from download to a test dictation at the cursor"}]},"howItWorks":{"label":"How it works","title":"From spoken finding to text in the open report.","subtitle":"Keep the reporting workflow you already know. OpenWhispr adds a system-wide hotkey and a local transcription option.","image":{"label":"Radiologist working at a laptop beside brain scans"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Choose a local model during setup."},{"title":"Put the cursor in the report.","body":"Open an editable findings or impression field, then press the hotkey and dictate."},{"title":"Proofread and sign.","body":"Check terminology, numbers, negation and laterality, then finish the report in your reporting system."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons radiologists choose OpenWhispr","subtitle":"Use it for the narrow job it ships to do: turn speech into text at the active cursor. Keep the clinical and reporting workflow under your control.","items":[{"title":"Local transcription at the reading station","body":"Choose a local model and, with local-to-cloud fallback off, audio is transcribed on the device without an upload step."},{"title":"Text lands at the active cursor","body":"The desktop app pastes the transcript into the focused editable field. If automatic paste is unavailable, it can leave the text on the clipboard."},{"title":"No reporting-stack replacement","body":"Keep the worklist, viewer, report template and signing system your practice already uses."},{"title":"A dictionary for radiology vocabulary","body":"Add anatomy, eponyms, protocol names and abbreviations as hints for supported transcription and cleanup paths."},{"title":"Keeps working off the hospital network","body":"Once a local model is downloaded, dictation does not need the reading-room network or internet."},{"title":"$$0 local dictation per radiologist","body":"Every desktop plan includes unlimited use of local speech-to-text models."},{"title":"Three desktop operating systems","body":"Run the same desktop dictation app on the operating system your workstation uses."},{"title":"Your IT team can inspect the paste path","body":"Your organization can inspect the transcription and paste paths on GitHub."},{"title":"You read the images. OpenWhispr handles speech to text.","body":"It is not a diagnostic image reader and has no native patient, study or radiology-system context. Agent Mode can use optional screen context, but it does not diagnose or replace review."},{"title":"100+ spoken languages","body":"OpenWhispr supports 100+ languages. Availability and accuracy vary by local model and cloud provider."}]},"faq":{"label":"Before you download","title":"What radiologists ask before testing dictation software.","subtitle":"Straight answers on report fields, terminology, integrations, patient data, offline use and cost.","items":[{"question":"What is the best radiology dictation software?","answer":"It depends on the job. If you need a full reporting platform with patient and study context, templates, structured data, quality checks and enterprise interfaces, use a radiology reporting system built for that. If you want free, local-first speech-to-text that pastes at an active cursor, test OpenWhispr in your actual report field."},{"question":"Does OpenWhispr integrate with PACS, RIS or EHR software?","answer":"No native integration. OpenWhispr does not receive patient or study context from PACS, RIS, EHR, DICOM, HL7 or FHIR. It transcribes speech and pastes text at the active cursor. Optional Agent Mode can use screen context, which is not diagnostic integration. Test your reporting software and virtual desktop."},{"question":"Can it navigate templates or build a structured radiology report?","answer":"OpenWhispr ships no built-in radiology structured-reporting integration, report macros, voice-command field navigation or template logic. Generic Agent Mode can reformat text from your prompt, but it does not know the report fields. Use your existing system template and cursor."},{"question":"How does it handle terminology, numbers and laterality?","answer":"The custom dictionary lets you add recurring anatomy, eponyms, protocol names and abbreviations as recognition hints. It is not a radiology lexicon or clinical error checker. Proofread every report for terminology, measurements, negation and laterality before signing."},{"question":"Can I use it offline with patient information?","answer":"After a local model is downloaded, local speech-to-text runs offline and does not upload audio when the separate local-to-cloud fallback is off. Cleanup is configured separately. Your organization still decides which devices and applications are approved for patient information; cloud transcription is a separate global mode."},{"question":"Is local radiology dictation really free?","answer":"Yes. Every desktop plan includes unlimited local dictation, a custom dictionary and 100+ languages at $0. Pro is $8 per month and adds unlimited OpenWhispr Cloud transcription, more meeting hours, mobile and device sync."}]},"reviews":{"label":"What users say","title":"What users say about local dictation.","subtitle":"These quotes are verbatim. None is rewritten as a radiologist endorsement.","score":{"value":"1","label":"Job in the reporting stack","body":"OpenWhispr supplies speech-to-text at the active cursor. Your reporting system keeps the patient, study, template and signature workflow."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Start free with local dictation.","Upgrade for cloud or team features."],"caption":"Compare OpenWhispr plans for radiologists","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Radiologists who want unlimited local cursor dictation and can run a model on the workstation"}]},{"parts":[{"text":"Radiologists who want unlimited OpenWhispr Cloud transcription"}]},{"parts":[{"text":"Teams that want OpenWhispr Cloud, recordings and priority support"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local speech-to-text at $0, with audio staying on the workstation"}]},{"parts":[{"text":"Unlimited cloud transcription without managing local models or keys"}]},{"parts":[{"text":"Unlimited recordings and priority support for a team"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"We built the speech-to-text layer, not a radiology stack.","paragraphs":["Radiologists already work across a worklist, image viewer and reporting system. OpenWhispr does one narrower job: turn speech into text and paste it at the active cursor.","We build in public. Your reading-room or IT team can inspect the local transcription and paste paths on GitHub instead of relying only on a product page.","And if a report field will not accept the paste, you talk to the people who built it, in Discord."],"image":{"label":"OpenWhispr founder and community"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Test one dictated finding at the cursor.","body":"Start in a blank text field, then confirm cursor paste in an approved report field before clinical use.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, ask us for help in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseHipaaCompliant":{"breadcrumb":{"home":"Home","useCases":"Use Cases","hipaaCompliant":"For HIPAA-Regulated Teams"},"metadata":{"title":"HIPAA Compliant Dictation Software, Local + BAA - OpenWhispr","description":"OpenWhispr is HIPAA compliant dictation software with on-device transcription and a Business-plan BAA for cloud use.","keywords":["hipaa compliant dictation software","hipaa dictation software","hipaa compliant voice to text","secure medical dictation software","private healthcare dictation","baa dictation software","on-device medical transcription"]},"schema":{"name":"OpenWhispr","applicationCategory":"Healthcare dictation software","operatingSystem":"macOS, Windows, Linux, iOS","description":"HIPAA compliant, local-first, open-source dictation software for healthcare teams. Local transcription runs on the user's device. A separate BAA is available on request from the Business plan for cloud Services.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"HIPAA compliant dictation software for healthcare teams","title":"Pick a local model. Get the transcript. Send us no audio.","body":"Patient audio is transcribed on the computer in front of you. The DPA requires a signed BAA before cloud PHI; confirm it covers the exact feature path.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Professional holding a notebook and pointing toward a desktop computer"}},"trustBar":{"label":"Dictate into any text field, including:","names":["Epic","Oracle Health","athenahealth","eClinicalWorks","NextGen","Microsoft Word"]},"pain":{"label":"Choosing HIPAA compliant dictation software","title":"The compliance question starts before anyone presses record.","subtitle":"Ask where the audio goes, what text is saved, who processes it and whether a BAA covers that exact path.","cards":[{"title":"A cloud recording starts before the BAA is signed.","body":"OpenWhispr's DPA says PHI must not go through the cloud Services until a separate BAA is in place. It is available on request from Business, but confirm its scope first."},{"title":"One product, three different data paths.","body":"Local models process audio on your device. OpenWhispr Cloud sends audio to our cloud. Bring your own key sends it to the provider you chose. The setting changes the answer."},{"title":"Audio deleted. Transcript still synced.","body":"Cloud audio is processed in real time and not persisted by OpenWhispr. Transcribed text and notes can still enter cloud sync when you enable it, so map both flows."},{"title":"A HIPAA badge with no configuration.","body":"HIPAA has no software certificate. The useful questions are concrete: which path handles PHI, which agreement covers it and which safeguards your team controls."}]},"features":{"label":"Three choices to make","title":"Three paths. Three different approval questions.","subtitle":"The product gives you local, OpenWhispr Cloud and bring-your-own-key paths. Your workflow decides which one handles PHI.","items":[{"image":{"label":"Clinician dictating into a phone beside a clinical workstation"},"title":"Keep transcription on-device","body":"Pick a local model and patient audio is transcribed on your computer. If you use AI cleanup or drafting, select a local model for that step too."},{"image":{"label":"Professional reviewing a document beside a laptop"},"title":"Put the BAA before the cloud","body":"For PHI in OpenWhispr Cloud, execute the separate BAA and confirm it covers the feature and subprocessors in your path. It is available on request from Business."},{"image":{"label":"Two professionals reviewing a laptop in an office"},"title":"Map the text after transcription","body":"Cloud audio is not persisted after transcription. Notes and transcribed text may still be stored or synced, depending on the settings and tools you choose."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Patient audio sent to us","value":"None","footnote":"when a local transcription model is selected"},{"label":"Cost for local dictation","value":"$$0","footnote":"unlimited on every plan, including Free"},{"label":"Setup takes","value":"5 min","footnote":"then review the data path before using PHI"}]},"howItWorks":{"label":"How it works","title":"Approve the path before the first dictated note.","subtitle":"A HIPAA workflow is a configuration, a contract and your own safeguards. It is not a badge the download can grant.","image":{"label":"Professional reviewing paperwork at a standing desk"},"steps":[{"title":"Download and choose the path.","body":"Pick local models, OpenWhispr Cloud or your own provider. Write down which one your team approved for PHI."},{"title":"Dictate only after the contract check.","body":"Local transcription sends no audio to OpenWhispr. Keep cloud PHI off until the separate BAA is signed and its feature scope is confirmed."},{"title":"Review, file and retain deliberately.","body":"Check the note, put it in the approved record system and set local retention and optional sync to match your policy."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"What your review can verify","title":"10 reasons HIPAA-regulated teams evaluate OpenWhispr","subtitle":"Each line names a specific product or contract fact. None replaces your risk analysis, policies or workforce safeguards.","items":[{"title":"Local transcription sends us no audio","body":"Pick a local speech-to-text model and the recording is processed on your own computer."},{"title":"OpenWhispr does not persist cloud audio","body":"The DPA says cloud voice audio is processed in real time and not persisted by OpenWhispr after transcription."},{"title":"A BAA is available on request","body":"Business-plan and Enterprise customers can request the separate BAA required before PHI enters the cloud Services."},{"title":"Unlimited local dictation is free","body":"Local speech-to-text is unlimited on every plan, including Free, on macOS, Windows and Linux."},{"title":"Local mode works offline","body":"Once the local model is downloaded, transcription needs no internet connection."},{"title":"The app is open source","body":"Security teams can inspect the routing code instead of relying on a marketing description."},{"title":"Customer data is not used for AI training","body":"The DPA prohibits OpenWhispr and its subprocessors from using customer personal data to train or improve models."},{"title":"Transcription and cleanup are separate routes","body":"Review the speech-to-text model and the AI cleanup model independently before PHI use."},{"title":"Bring your own key creates a third path","body":"In BYOK mode, content goes directly from the device to the model provider you chose."},{"title":"Sync is a separate retention decision","body":"Cloud backup defaults off. Turn it on only after deciding where transcribed text and notes may be stored."}]},"faq":{"label":"Before PHI enters the workflow","title":"Six questions for a HIPAA dictation review.","subtitle":"These answers describe the product and its shipped agreements. Your organization still owns the final risk analysis.","items":[{"question":"Is OpenWhispr HIPAA compliant?","answer":"Yes. OpenWhispr is HIPAA compliant. Local transcription sends us no patient audio. For PHI in the cloud Services, its DPA requires a separate signed BAA, available on request from the Business plan. Your organization still needs to configure its chosen path and safeguards correctly."},{"question":"Do I need a BAA for local transcription?","answer":"Generally not for that transcription step. With a local model, audio is processed on the device and OpenWhispr does not receive it. You still need safeguards for the computer, the resulting note, any sync you enable and the system where you file it. Confirm the full analysis with your compliance counsel."},{"question":"Can I use OpenWhispr Cloud with PHI?","answer":"Only after a separate BAA is executed and you confirm it covers the feature path you plan to use. The shipped DPA says end users must not process PHI through the cloud Services without one. Business-plan and Enterprise customers can request it at support@openwhispr.com."},{"question":"What does OpenWhispr store?","answer":"OpenWhispr does not persist cloud voice audio after transcription. Local retention settings can keep dictation audio and transcripts on the device, and optional cloud sync can store transcribed text and notes. Review each setting against your retention policy."},{"question":"Does local transcription keep AI cleanup local too?","answer":"Not by itself. Transcription and AI cleanup use separate models. The local setup can configure both, but verify both routes in Settings before PHI use. A bring-your-own-key cleanup model sends the text to that provider."},{"question":"Is the Free plan enough for a HIPAA workflow?","answer":"Free includes unlimited local dictation on macOS, Windows and Linux. It does not include an OpenWhispr BAA for cloud PHI. If your approved path needs OpenWhispr Cloud, request the BAA on Business or Enterprise before using PHI."}]},"reviews":{"label":"What users say about the tool","title":"Product reviews, quoted word for word.","subtitle":"These are verbatim product reviews. None is presented as evidence for your HIPAA analysis.","score":{"value":"3","label":"Verbatim reviews","body":"The wording and attribution are unchanged from the real source reviews."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for HIPAA-regulated healthcare teams","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Healthcare teams using a local-only transcription path for PHI"}]},{"parts":[{"text":"Individuals using cloud transcription for work that does not contain PHI"}]},{"parts":[{"text":"Teams evaluating OpenWhispr Cloud for PHI under a separate signed BAA"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local dictation, with patient audio processed on your device"}]},{"parts":[{"text":"Unlimited cloud transcription for approved non-PHI workflows"}]},{"parts":[{"text":"A BAA on request, unlimited recordings and advanced team workflows"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"A HIPAA checkbox cannot tell you where the audio went.","paragraphs":["OpenWhispr started with a concrete choice: let people transcribe on their own machine when a cloud data path is the wrong fit.","We build in public. Your security team can inspect the local and cloud routing in the app, then compare it with the DPA and the settings your workforce will actually use.","And if your review finds a question the documents do not answer, you can ask the people who built it in Discord or request the BAA through support."],"image":{"label":"OpenWhispr founder discussing the product's data paths"}},"finalCta":{"tagline":"Local path · cloud path · BAA on Business","title":"Choose the data path before you dictate the patient name.","body":"Start with free local transcription, or request the Business-plan BAA and confirm its scope before PHI enters OpenWhispr Cloud. Then document the settings your team approved.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes about 5 minutes to set up. Free local dictation on Mac, Windows & Linux. Your organization remains responsible for its HIPAA workflow."}}},"useCaseAbaSessionNotes":{"breadcrumb":{"home":"Home","useCases":"Use Cases","abaSessionNotes":"For ABA Therapists"},"metadata":{"title":"Speech-to-Text Tools for ABA Session Notes | OpenWhispr","description":"Dictate ABA session-note details with OpenWhispr, then review and paste the text into your existing record system. Choose Cloud for speed or local processing.","keywords":["speech-to-text for aba session notes","aba session notes software","rbt session note dictation","bcba session notes software","aba dictation software","aba progress notes dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text software","operatingSystem":"macOS, Windows, Linux","description":"Open-source speech-to-text software for dictating ABA session-note details. Review and paste the text into your existing record system; choose Cloud or local processing.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation software for ABA therapists & RBTs","title":"Dictate ABA session notes while the details are fresh.","body":"Dictate ABC observations, interventions, client response and goal progress while the session is still fresh. Review the text, then paste it into your existing record system. Choose Cloud for speed or local processing for sensitive and offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Therapist and a child at a low table with wooden building blocks and an open laptop"}},"trustBar":{"label":"Use dictated text in:","names":["your ABA record system","session-note fields","secure notes","email","any text field"]},"pain":{"label":"What slows a session note","title":"The session is still fresh. Typing the note is the delay.","subtitle":"You need clean dictation between clients, a practical mode for sensitive behavioral data, and a plan that fits your caseload.","cards":[{"title":"A stranger transcribing behavioral data.","body":"Traditional dictation services send the recording to a third party, so someone you've never met hears details about a child's behavior. Choose local processing and OpenWhispr transcribes on your own device instead, so no one else is in the loop."},{"title":"Sensitive client data needs a deliberate path.","body":"Choose local transcription when the session covers sensitive behavioral or health details, or you're between locations without signal. Your device processes the recording, so documentation continues wherever the session happens."},{"title":"Typing is minutes you don't have between clients.","body":"Typing an ABC note eats into your break, or pushes you late into the next session. OpenWhispr drafts it from your dictation while it's fresh, so you review and paste instead of type."},{"title":"The quickest route should still be dependable.","body":"When speed and consistent performance matter most, Cloud avoids a local model download and transcribes with audio discarded immediately after transcription."}]},"features":{"label":"What you get","title":"Faster notes. Choose the mode that fits the session.","subtitle":"Get the note out while the session is fresh. Use Cloud for fast, consistent transcription or local processing for sensitive client data.","items":[{"image":{"label":"Therapist and a child high-fiving across a table with a laptop and picture cards"},"title":"Cloud transcription without a local model download","body":"Dictate an ABC note, progress update or session summary into your mic. Cloud gives you top accuracy and speed with consistent performance across hardware, then discards the transcription audio immediately."},{"image":{"label":"Therapist speaking while working at a laptop in a private office"},"title":"Local processing for sensitive client data","body":"Choose a local model when a session touches details about a client's behavior or health you'd rather keep off a server, or you're documenting offline between locations. You review the drafted note before it goes into the client's file."},{"image":{"label":"Therapist pointing at picture cards beside a child, one hand resting on a laptop"},"title":"Dictation wherever your cursor is","body":"Press a hotkey and dictate straight into CentralReach, Rethink or your EHR. Clean, punctuated text lands where you are already working."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Client data sent to us","value":"None","footnote":"in local mode, transcription never leaves your device"},{"label":"Cost per RBT or BCBA","value":"$$0","footnote":"unlimited local dictation, the whole caseload, forever"},{"label":"Free Cloud allowance","value":"2,000 words","footnote":"per week, alongside unlimited local dictation"}]},"howItWorks":{"label":"How it works","title":"A session note drafted before the next client walks in.","subtitle":"Start free, choose Cloud for the quickest high-performance setup or local for sensitive client data, then get the note drafted before the next session.","image":{"label":"Therapist speaking at a laptop in a therapy room lined with toy shelves"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Choose Cloud for no local model download or local for sensitive client data."},{"title":"Dictate the note.","body":"Speak the ABC data, goal progress and session summary, and it's transcribed in whichever mode you chose: on your device, or via Cloud."},{"title":"Review, paste, delete.","body":"Edit the draft, paste it into CentralReach or your EHR, and delete the audio. You control all of it."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons ABA therapists choose OpenWhispr","subtitle":"Choosing a dictation tool is a confidentiality decision before it's anything else. Here are the facts, and your compliance team can verify every one in the repo.","items":[{"title":"In local mode, audio never leaves your device","body":"Choose local processing and there's no upload step: dictation and drafts are handled entirely on your machine."},{"title":"Free for the whole caseload","body":"Unlimited local dictation at $0, every RBT and BCBA, forever. No per-seat licence to renew."},{"title":"In local mode, nothing goes to a third-party transcriber","body":"Choose local processing and no outsourced service ever hears the file. The dictation stays between you and your machine."},{"title":"Works fully offline","body":"A client's home, the car, a hallway with no wifi. No signal required, so it never stalls mid-dictation."},{"title":"Dictate into CentralReach or any EHR","body":"A system-wide hotkey drops clean text wherever your cursor is: CentralReach, Rethink, WebABA, an email."},{"title":"Optional AI formatting for your draft","body":"Create a custom Note Action to format dictated details into an ABC-style draft, then verify every field against the session record before filing it."},{"title":"You control storage and deletion","body":"Local transcription processes dictation on your machine. Sync, Cloud transcription and connected AI providers use separate data paths, so choose them according to your practice's requirements."},{"title":"Open source, auditable by anyone","body":"Your compliance or IT team can read every line on GitHub. No faith required."},{"title":"Handles ABA terminology","body":"Add ABC, MO, SIB, FBA, program names and practice-specific acronyms to the custom dictionary for better transcription."},{"title":"99 languages, even offline","body":"The local Whisper model alone covers 99 languages, so you can document sessions in the language your team actually speaks with no internet required."}]},"faq":{"label":"Before you download","title":"The questions ABA therapists ask first.","subtitle":"Straight answers on confidentiality, cost and accuracy. If yours isn't here, ask us in Discord.","items":[{"question":"Does the audio ever leave my device?","answer":"With a local transcription model, the recording stays on your device. With OpenWhispr Cloud, audio is sent only to transcribe it, then immediately discarded. Choose the mode that fits the session and your workflow."},{"question":"Is this appropriate for confidential client behavioral data?","answer":"Confidentiality and any HIPAA obligations are your responsibility, and no tool discharges them for you. Local processing is the deliberate choice when a session covers sensitive behavioral or health details. Cloud transcription discards audio immediately after transcription. This is not compliance advice; follow your practice's and payer's requirements."},{"question":"Can it handle ABA terminology and acronyms?","answer":"Yes. Local Whisper models cover a wide range of accents and up to 99 languages (other local models, like Parakeet and Nemotron, cover fewer). Cloud transcription covers 100+ languages. The custom dictionary lets you add ABC, MO, SIB, FBA, client names and the program terms your practice uses so they come out right."},{"question":"Will it work between sessions with no signal?","answer":"Yes. Local models run entirely offline. Dictate in the car, the hallway or a client's home with no wifi, and file the note once you're back online. Choose local before you need that offline path."},{"question":"Does it draft the note, or replace my ABA record system?","answer":"It can transcribe your dictation, and a custom Note Action can format a draft. It does not collect ABA data, manage treatment plans, capture signatures or guarantee payer-compliant documentation. Review the draft and complete the note in your existing system."},{"question":"Is it actually free?","answer":"Yes. Free gives every RBT and BCBA useful local dictation plus a limited Cloud allowance. Upgrade to Pro for unlimited Cloud transcription, more meeting-recording time, sync, mobile, and API/MCP access. Business adds unlimited meeting recordings, Agent mode, and chat over data."}]},"reviews":{"label":"What users say","title":"Built by people who don't want to be recorded either.","subtitle":"From our Discord, GitHub and inboxes, the same instinct that matters when a client's records are on the line.","score":{"value":"0","label":"Confidential client files on a server","body":"In local mode the audio never leaves your device. Nothing is stored on a server, and nothing is sent to a transcriber."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for ABA therapists","tiers":[{"id":"free","name":"Free","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":false},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","badge":"Most Popular","cta":{"label":"Start with Business"},"featured":true}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"RBTs and BCBAs who want a useful free start with local dictation and limited Cloud transcription"}]},{"parts":[{"text":"RBTs and BCBAs who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"ABA practices documenting at volume with Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Useful Free access, including local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some conversations are confidential. We built for those.","paragraphs":["OpenWhispr started with a simple objection: the notes that describe a client's most sensitive moments shouldn't have to pass through a stranger's server to become documentation.","We build in public. The repository has thousands of stars and hundreds of forks, and anyone at your practice can inspect the code. In local mode, dictation is processed on your machine.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · Cloud or local","title":"Dictate the details. Review the ABA session note.","body":"Download OpenWhispr, dictate while the session is fresh, and review the text before pasting it into your existing record system.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free on Mac, Windows & Linux. Community support is available in Discord."}}},"useCaseTherapists":{"breadcrumb":{"home":"Home","useCases":"Use Cases","therapists":"For Therapists"},"metadata":{"title":"Free Therapist Dictation Software - OpenWhispr","description":"OpenWhispr is free therapist dictation software for session notes. Choose Cloud for speed or local processing for sensitive and offline work.","keywords":["therapist dictation software","therapy dictation software","session note dictation","mental health dictation software","offline therapy dictation","clinical note dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Therapist dictation software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source dictation software for therapists. Choose Cloud for top accuracy and speed without a local model download, or local for sensitive and offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation software for therapists","title":"Dictate the session. Get the note written up.","body":"Turn spoken session notes into clean text while the details are still clear. Choose Cloud for no-download speed or local processing for sensitive and offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Therapist listening to a client seated on a sofa across a wood-panelled office, a laptop open on the side table between them"}},"trustBar":{"label":"Dictate into:","names":["SimplePractice","TherapyNotes","Jane","TheraNest","Sessions Health","Microsoft Word"]},"pain":{"label":"What slows a session note","title":"You remember the session now. The keyboard comes later.","subtitle":"Capture the clinical detail while it is fresh, then choose the transcription path that fits the session and your setting.","cards":[{"title":"A full day leaves notes until evening.","body":"Back-to-back sessions leave little room to type between clients. Dictate the note while the session is fresh, then review the clean text before you save it."},{"title":"Client conversations need a deliberate path.","body":"Choose local transcription for highly sensitive sessions or offline work. Your device processes the audio, so you keep control of where the transcription happens."},{"title":"Names and terms need to come out right.","body":"Add client names, medications, modalities, and practice terms to the custom dictionary so OpenWhispr has the spellings you use."},{"title":"The quick setup should stay quick.","body":"When speed matters most, Cloud avoids a local model download and discards transcription audio immediately after transcription."}]},"features":{"label":"What you get","title":"Session notes while the session is still clear.","subtitle":"Use Cloud for fast transcription without a model download. For sensitive notes or an offline day, local processing keeps transcription on your device.","items":[{"image":{"label":"Therapist speaking while seated at an open laptop at a table in a warm office with a bookshelf and armchair"},"title":"Cloud transcription without a local model download","body":"Press your hotkey and dictate the session note. Cloud transcribes without a local model download, then discards the transcription audio immediately."},{"image":{"label":"Therapist gesturing while talking with a client, both in armchairs, a notebook on his lap and a laptop on the side table"},"title":"Local processing for sensitive or offline work","body":"Choose a local model when a highly sensitive session or an offline day calls for device-only transcription. Review the text before it becomes part of the record."},{"image":{"label":"Counsellor working at a laptop along a window counter in a counseling centre, a closed folder beside them"},"title":"Dictation wherever your cursor is","body":"Use a hotkey to dictate into your practice system, a referral letter, an email, or Word. Clean, punctuated text lands where you are already working."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Client session audio sent to us","value":"None","footnote":"in local mode, transcription stays on your device"},{"label":"Cost per therapist","value":"$$0","footnote":"unlimited local dictation, every therapist, forever"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first dictated session note"}]},"howItWorks":{"label":"How it works","title":"A session note written up before the next client.","subtitle":"Start free, choose Cloud for the quickest high-performance setup or local for sensitive and offline work, then dictate while the session is fresh.","image":{"label":"Man speaking while typing at a laptop at a round table in a bright office lounge with plants"},"steps":[{"title":"Download for free.","body":"Mac, Windows, or Linux. Choose Cloud for no local model download or local for sensitive and offline work."},{"title":"Dictate the session note.","body":"Press your hotkey, speak the details you need to record, and OpenWhispr turns your voice into text."},{"title":"Review, save, delete.","body":"Edit the text, save it in your practice system, and delete the audio when you are done. You stay in control of the record."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons therapists choose OpenWhispr","subtitle":"A dictation tool touches sensitive conversations. Here are the product facts, so you can choose the setup that fits your practice.","items":[{"title":"Local audio stays on your device","body":"Choose local transcription when you want the audio processed on your own device, with no internet connection required."},{"title":"Free local dictation for the whole practice","body":"Unlimited local models are included at $0. Start with every clinician who needs to dictate, without a per-seat transcription limit."},{"title":"Works without Wi-Fi","body":"Local models run offline, so a weak clinic connection or a day away from the office does not stop dictation."},{"title":"A custom dictionary for practice language","body":"Add client names, medications, modalities, and acronyms so the words your practice uses have the spellings you expect."},{"title":"Dictate into the record you already use","body":"A system-wide hotkey puts text where your cursor is, whether that is your practice system, Word, or an email."},{"title":"Cloud when setup speed matters","body":"OpenWhispr Cloud delivers top accuracy and speed without asking you to download a local transcription model."},{"title":"Cloud audio is not kept","body":"When you use OpenWhispr Cloud, audio is sent only to transcribe it and is immediately discarded afterwards."},{"title":"100+ languages","body":"Choose from speech-to-text support for more than 100 languages across every plan."},{"title":"Unlimited Cloud transcription when you need it","body":"Pro adds unlimited Cloud transcription, sync, mobile, and API/MCP access. Business adds Agent mode and chat over data."},{"title":"Open source, auditable by anyone","body":"The code is public on GitHub, so you can inspect how OpenWhispr works instead of taking a privacy slogan on faith."}]},"faq":{"label":"Before you download","title":"The questions therapists ask first.","subtitle":"Straight answers on privacy, cost, and how the transcription choices work. If yours is not here, ask us in Discord.","items":[{"question":"Does session audio ever leave my device?","answer":"With a local transcription model, the recording stays on your device. With OpenWhispr Cloud, audio is sent only to transcribe it, then immediately discarded. Choose the mode that fits the session and your workflow."},{"question":"Can I use this for confidential client notes?","answer":"Confidentiality is your professional responsibility, and no tool removes that responsibility. Local processing is the deliberate choice when a highly sensitive session calls for device-only work. Cloud transcription discards audio immediately after transcription. Follow the standards and requirements that apply to your practice."},{"question":"Can it recognize client names and clinical terminology?","answer":"Yes. Add names, medications, modalities, and practice terms to the custom dictionary so OpenWhispr uses the spellings you need. It also supports speech-to-text in more than 100 languages."},{"question":"Will it work when I do not have Wi-Fi?","answer":"Yes. Local transcription models run entirely on your device, without an internet connection. Choose local before you need that offline path."},{"question":"Does it write a note or just transcribe?","answer":"OpenWhispr turns your spoken session note into clean text. You can then review, edit, and save it in the record system you use."},{"question":"Is the free plan actually useful?","answer":"Yes. Free includes unlimited local dictation, a custom dictionary, and a limited Cloud allowance. Upgrade to Pro for unlimited Cloud transcription, more meeting-recording time, sync, mobile, and API/MCP access. Business adds unlimited meeting recordings, Agent mode, and chat over data."}]},"reviews":{"label":"What users say","title":"Built by people who do not want every conversation recorded.","subtitle":"From our Discord, GitHub, and inboxes, the same instinct that matters in a confidential practice.","score":{"value":"0","label":"Client session audio on a server","body":"In local mode the audio never leaves your device. Nothing is stored on a server, and nothing is sent to a transcriber."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for therapists","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Therapists who want a useful free start with local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Therapists who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Practices documenting at volume with Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Useful Free access, including local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Enterprise includes team features and admin controls, SSO, SAML, SCIM, audit logs, retention controls, and dedicated support."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some conversations are private. We built for those.","paragraphs":["OpenWhispr gives you a deliberate local option when you want speech transcribed on your own device.","We build in public. The code is on GitHub, where anyone can inspect how OpenWhispr works. When you choose local dictation, the transcription stays on your device.","Need help? Join the OpenWhispr community on Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Your session note can start with your voice.","body":"Download OpenWhispr and dictate the details while they are fresh. Choose the transcription path that fits your session and your setting.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free local dictation on Mac, Windows & Linux."}}},"useCaseStudentsWithDisabilities":{"breadcrumb":{"home":"Home","useCases":"Use Cases","studentsWithDisabilities":"For Students with Disabilities"},"metadata":{"title":"Speech-to-Text for Students with Disabilities - OpenWhispr","description":"OpenWhispr is free, open-source speech-to-text software for students with disabilities. Dictate coursework in desktop apps, online or offline.","keywords":["speech-to-text software for students with disabilities","dictation software for students with disabilities","assistive technology for students","student speech-to-text","disability accommodation software","offline dictation software"]},"schema":{"name":"OpenWhispr","applicationCategory":"Assistive speech-to-text software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source speech-to-text software for students with disabilities. Dictate coursework with Cloud transcription or local processing for offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text software for students with disabilities","title":"Say the answer. Get it onto the page.","body":"Dictate notes, essays and answers into the apps you already use. Choose Cloud for fast setup or local processing for offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Student in a wheelchair speaking at a library table with an open laptop and notebook"}},"trustBar":{"label":"Dictate into:","names":["Microsoft Word","Google Docs","Canvas","Blackboard","Notion","OneNote"]},"pain":{"label":"When writing gets in the way","title":"You know the answer. Getting it onto the page is the hard part.","subtitle":"Use your voice for notes, essays and short answers. Then review the text in the tools your course already uses.","cards":[{"title":"The idea is there. The sentence will not come out.","body":"When typing takes more effort than thinking, coursework becomes a fight with the keyboard. Speak the sentence, then edit clean text on the page."},{"title":"An accommodation should work in the actual assignment.","body":"Dictate straight into Word, Google Docs, Canvas or the answer field in front of you. No copying between a recorder and your coursework."},{"title":"A slow laptop should not decide whether you can write.","body":"Choose Cloud when you do not want to download a local model. Audio is sent only to transcribe it, then immediately discarded."},{"title":"Wi-Fi is not guaranteed.","body":"Choose local processing when you need to work offline. The transcription runs on your device, so you can keep writing when the connection drops."}]},"features":{"label":"What you get","title":"Speech-to-text where your coursework happens.","subtitle":"Turn spoken answers into editable text. Pick the mode that works for where and how you study.","items":[{"image":{"label":"Student wearing dark glasses speaking at a laptop in a quiet study room"},"title":"Start speaking without a local model download","body":"Choose Cloud for fast setup without a local model download."},{"image":{"label":"Student with a wrist brace speaking at a laptop in a science classroom"},"title":"Work offline with local processing","body":"Download a local model and dictate on your device without internet. It is a practical choice for study spaces with unreliable Wi-Fi."},{"image":{"label":"Student in a wheelchair speaking at a workshop desk while pressing a large accessible switch"},"title":"Put text where your cursor is","body":"Press a hotkey, speak, and place punctuated text in the document, learning platform or notes app you are already using."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Cost to start using local speech-to-text","value":"$$0","footnote":"unlimited local models on every plan"},{"label":"Cloud transcription on Free","value":"2,000","footnote":"words each week, plus unlimited local speech-to-text"},{"label":"Languages supported","value":"100+","footnote":"speech-to-text on every plan"}]},"howItWorks":{"label":"How it works","title":"Speak the answer. Review the text. Submit the work.","subtitle":"Start free, choose Cloud for fast setup or local processing for offline work, then dictate into the assignment in front of you.","image":{"label":"Student speaking at a desk while pressing a large green accessible button beside a monitor"},"steps":[{"title":"Download for free.","body":"Use OpenWhispr on Mac, Windows or Linux. Choose Cloud for no local model download, or local for offline work."},{"title":"Dictate the answer.","body":"Put your cursor in the document, learning platform or notes app, then speak your sentence."},{"title":"Review and submit.","body":"Edit the text in your own words, check the assignment, and hand it in when it is ready."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Built for the work in front of you","title":"10 practical reasons to choose OpenWhispr","subtitle":"A speech-to-text tool has to work where the assignment happens. These are the practical features you can check before you rely on it.","items":[{"title":"Free local speech-to-text","body":"Local AI models are unlimited and free on every plan, so you can dictate without a subscription clock."},{"title":"Works without Wi-Fi","body":"Local models run entirely on your device. Keep working in a library, on a commute, or anywhere the connection is unreliable."},{"title":"Dictate into the app already open","body":"Use a hotkey to place text at your cursor in documents, notes, email, and browser-based coursework."},{"title":"Cloud when fast setup matters","body":"Cloud transcription offers accuracy and speed without a local model download, with audio immediately discarded after transcription."},{"title":"A custom dictionary for your course","body":"Add names, technical terms and acronyms so the terms you use in class have a better chance of appearing correctly."},{"title":"More than 100 languages","body":"OpenWhispr supports speech-to-text in more than 100 languages on every plan."},{"title":"Editable text, not a locked recording","body":"Your speech becomes text in the document you are writing. Review it, revise it, and make it your own before you submit."},{"title":"Use it across your study tools","body":"Dictate into Word, Google Docs, OneNote, Notion, and browser-based learning platforms from the same desktop app."},{"title":"Open source","body":"The project is MIT-licensed, so students and accessibility teams can inspect how the software works."},{"title":"Upgrade only when the workflow needs it","body":"Free includes local models. Pro adds unlimited OpenWhispr Cloud transcription, sync, mobile, and API and MCP access."}]},"faq":{"label":"Before you download","title":"Questions students ask before trying speech-to-text.","subtitle":"Straight answers on accommodations, offline work, accuracy and cost.","items":[{"question":"Can I use this as speech-to-text software for coursework?","answer":"Yes. OpenWhispr turns speech into editable text at your cursor, so you can dictate notes, drafts and answers into the document or learning platform you are using."},{"question":"Does it work without internet?","answer":"Yes, with a local AI model. Local speech-to-text runs on your device without internet. Cloud transcription needs a connection and is the easy option when you want speed with no local model download."},{"question":"Is OpenWhispr free for students?","answer":"Yes. Free includes unlimited local AI models, 2,000 words a week of OpenWhispr Cloud transcription, 5 hours of meeting recordings a month, and unlimited Cloud transcription with your own API keys."},{"question":"Can it handle course vocabulary and names?","answer":"Use the custom dictionary to add names, acronyms and subject-specific terms. Review the final text before submitting, especially for specialised vocabulary."},{"question":"Does Cloud keep my audio?","answer":"Cloud audio is sent only to transcribe it, then immediately discarded. Local processing keeps the transcription on your device. Choose the mode that fits how and where you study."},{"question":"What do I get by upgrading from Free?","answer":"Pro adds unlimited OpenWhispr Cloud transcription, 20 hours of meeting recordings a month, cross-device sync, personal API access, MCP integration, and the iPhone and iPad app. Business adds unlimited meeting recordings, Agent mode, chat over your data, and priority support."}]},"reviews":{"label":"What people say","title":"Speech-to-text that lets the words get out first.","subtitle":"Three comments from OpenWhispr users.","score":{"value":"$$0","label":"Cost of local speech-to-text","body":"Unlimited local AI models are included on every plan."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for students with disabilities","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Students who want free, unlimited local speech-to-text and a useful amount of Cloud transcription"}]},{"parts":[{"text":"Students who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Students who record more meetings and want Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Free local speech-to-text plus limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud with no local model to download, then immediately discards it.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Assistive technology should not make you trade away control.","paragraphs":["OpenWhispr gives students with disabilities a practical way to turn speech into editable text, in the coursework tools they already use.","OpenWhispr is MIT-licensed and its source code is public on GitHub. Local models run on your device, and Cloud audio is immediately discarded after transcription.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team and community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Your next sentence can start out loud.","body":"Download OpenWhispr, dictate the work in front of you, then review it in your own words.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free forever on Mac, Windows and Linux. Get help from the community in Discord."}}},"useCaseKids":{"breadcrumb":{"home":"Home","useCases":"Use Cases","kids":"For Kids"},"metadata":{"title":"Speech-to-Text Software for Teens 13+ - OpenWhispr","description":"OpenWhispr is open-source speech-to-text software for parents and educators helping teens aged 13 and over turn spoken ideas into editable text.","keywords":["speech to text software for kids","speech to text for teens","dictation software for teens","speech to text for parents","speech to text for educators","offline speech to text for teens"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text software for parents and educators of teens","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source speech-to-text software for parents and educators helping teens aged 13 and over turn spoken ideas into editable text. Choose Cloud for speed or local for offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text software for kids","title":"Say the story. Get it on the page.","body":"Help a teen aged 13 and over speak a story, journal entry or note, then let them review and revise the editable text. Choose Cloud for quick setup and fast transcription, or local when working offline.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"A teenager speaking and gesturing at a laptop on a round table in a sunlit library"}},"trustBar":{"label":"Common writing tools:","names":["Google Docs","Microsoft Word","Apple Pages","Microsoft OneNote","Google Slides","Canva"]},"pain":{"label":"When typing slows a teen's first draft","title":"The idea is ready. Help them get it onto the page.","subtitle":"Set up the tool, let a teen aged 13 and over say the words while they are fresh, then give them editable text to review and revise.","cards":[{"title":"The idea is there. The keyboard gets in the way.","body":"A story opening, a journal thought or a list of ideas can be easy to say and slow to type. OpenWhispr turns spoken words into editable text for the teen to review."},{"title":"An offline writing session needs a practical option.","body":"Choose a local model for an offline writing session. The device processes the audio, so the teen can keep working without an internet connection."},{"title":"Speech-to-text is the start, not the finished work.","body":"Help the teen say the opening out loud. The editable text gives them something to reread, change and build on."},{"title":"Cloud is the quick, no-download option.","body":"Cloud avoids a local model download and gives fast transcription. The transcription audio is discarded immediately afterward."}]},"features":{"label":"What you get","title":"Help a teen get the words down. Then let them make them theirs.","subtitle":"Use Cloud for a fast, no-download start, or local processing for an offline writing session. Both leave the teen with editable text to check and improve.","items":[{"image":{"label":"A teenager working at a laptop at a wooden dining table in a bright kitchen"},"title":"Cloud transcription without a local model download","body":"Help a teen speak an opening, a list of ideas or a personal note into the microphone. Cloud is the no-download option for fast transcription, then discards the transcription audio immediately."},{"image":{"label":"A teenager pointing at a laptop screen at a dining table in the evening, with city lights outside the window"},"title":"Local processing for offline work","body":"Choose a local model when working offline. The device processes the audio, then the teen can reread and edit the words."},{"image":{"label":"A teenager speaking at a laptop in a school workshop, with a robotics kit on the bench"},"title":"Set a hotkey to start dictation","body":"Set the hotkey once, then help the teen say what they want to write and review the text before they keep going."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Audio sent to us in local mode","value":"None","footnote":"in local mode, transcription never leaves your device"},{"label":"Cost to get started","value":"$$0","footnote":"unlimited local dictation, free forever"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first dictated sentence"}]},"howItWorks":{"label":"How it works","title":"From a spoken idea to editable text.","subtitle":"Start free, choose Cloud for the quickest no-download setup or local for offline work, then help a teen aged 13 and over turn a spoken idea into text.","image":{"label":"A teenager speaking at a laptop beside an open notebook on a kitchen counter"},"steps":[{"title":"Set it up once.","body":"Download on Mac, Windows or Linux. Choose Cloud for no local model download or local for offline work."},{"title":"Help them say the idea.","body":"The teen speaks the words they want to get down. OpenWhispr turns their voice into editable text."},{"title":"Let them review and revise.","body":"Read the words together, fix what needs fixing, then let the teen keep writing in the document."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 ways OpenWhispr helps parents and educators","subtitle":"Speech-to-text should help a teen aged 13 and over get an idea onto the page, then leave them with words they can still shape themselves.","items":[{"title":"A local mode for offline work","body":"Set up a local model and the audio stays on the device. No internet connection is required for transcription."},{"title":"Free local dictation","body":"Try unlimited local AI models on the Free plan, with no monthly charge."},{"title":"One hotkey to start dictation","body":"Set a hotkey to start dictation while the teen is writing. Review the resulting text before they keep going."},{"title":"No Wi-Fi required in local mode","body":"Choose a local model when the connection drops. Transcription continues on the device."},{"title":"Start with the sentence they can say","body":"Help the teen speak the opening line, the tricky description or the next thought. Then let them edit the text in their own voice."},{"title":"Editable words to reread straight away","body":"OpenWhispr turns what the teen says into text, so they can reread the words before deciding what to write next."},{"title":"Cloud audio is not kept","body":"Cloud audio is sent only to transcribe it, then immediately discarded."},{"title":"Open source, so you can inspect it","body":"The app is MIT-licensed and its code is available on GitHub."},{"title":"A custom dictionary for names and terms","body":"Add names, places and words that matter to the teen so OpenWhispr has a better chance of getting them right."},{"title":"100+ languages","body":"Choose from 100+ supported languages for multilingual speech-to-text."}]},"faq":{"label":"Before you download","title":"Questions parents and educators consider for teens aged 13+.","subtitle":"Straight answers on age requirements, setup, working offline and helping a teen review editable text.","items":[{"question":"What is the age requirement?","answer":"OpenWhispr is intended for users aged 13 and over. If you are under 18, you confirm that your parent or legal guardian has read these Terms and agrees to them on your behalf."},{"question":"What is the difference between Cloud and local?","answer":"Cloud is the quick option: it gives top accuracy and speed without downloading a local model. Local runs on the device and works offline. Choose the setup that fits the writing session."},{"question":"Does OpenWhispr keep Cloud audio?","answer":"No. OpenWhispr Cloud receives audio only to transcribe it, then immediately discards the transcription audio. Local transcription keeps the audio on your device."},{"question":"Will it work without Wi-Fi?","answer":"Yes, when you choose a local transcription model. Local models run entirely on the device, so the teen can keep dictating without an internet connection."},{"question":"What if it gets a name or word wrong?","answer":"Review the text with the teen before they keep writing. You can add names, places and other useful words to OpenWhispr's custom dictionary to help with future transcripts."},{"question":"What does the Free plan include?","answer":"Free includes unlimited local AI models, 2,000 words of OpenWhispr Cloud transcription each week, 5 hours of meeting recordings each month, 100+ languages and a custom dictionary. Pro adds unlimited OpenWhispr Cloud transcription, sync and mobile."}]},"reviews":{"label":"What users say","title":"For parents and educators helping teens get words onto the page.","subtitle":"These are generic OpenWhispr reviews. The words, names and handles are unchanged.","score":{"value":"$$0","label":"to start with local dictation","body":"The Free plan includes unlimited local AI models. Start with the words you want to get onto the page."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for parents and educators supporting teens","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Parents and educators supporting teens aged 13+ who want a useful free start with local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Parents and educators supporting teens aged 13+ who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Teams recording at volume with Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"A useful free start with local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access for adults managing the setup"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some ideas are easier to say. Help a teen get them into editable text.","paragraphs":["For parents and educators helping teens aged 13 and over, OpenWhispr turns a spoken idea into editable text. The teen can then review, change and shape the words.","We build in public. The app is MIT-licensed and its code is available on GitHub. Local transcription runs on your device, and the code is there for anyone to inspect.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"OpenWhispr team and community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Help a teen get the idea onto the page.","body":"Download OpenWhispr, help a teen aged 13 and over say the first draft out loud, then let them make the editable words their own.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseDentists":{"breadcrumb":{"home":"Home","useCases":"Use Cases","dentists":"For Dentists"},"metadata":{"title":"Dental Dictation Software for Clinical Notes - OpenWhispr","description":"OpenWhispr turns spoken dental notes into editable text. Choose Cloud for no-download speed or local processing for sensitive and offline work.","keywords":["dental dictation software","dentist dictation software","dental voice dictation","dental note dictation","dental transcription software","dictation software for dentists"]},"schema":{"name":"OpenWhispr","applicationCategory":"Dental dictation software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source dental dictation software for clinical notes. Use Cloud without a local model download, or local processing for sensitive and offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dental dictation software for dentists","title":"Finish the note before the next patient.","body":"Place your cursor in the patient record and dictate the clinical note as soon as the appointment ends. Choose Cloud for no-download speed or local processing for sensitive and offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"A dentist in green scrubs stands at a counter, speaking while working at a laptop in a treatment room overlooking the sea."}},"trustBar":{"label":"Dictate into:","names":["Patient charts","Clinical notes","Procedure notes","Treatment plans","Referral letters","Microsoft Word"]},"pain":{"label":"What slows clinical notes","title":"The appointment is done. The chart still needs writing.","subtitle":"You need the details while they are fresh, a mode that fits the setting, and text ready to review before you sign.","cards":[{"title":"Notes written after the patient has gone.","body":"A full schedule leaves charting for the gap between patients or the end of the day. Dictate while the treatment details are still clear, then review the text before it enters the record."},{"title":"Procedure names that need checking.","body":"Tooth numbers, materials and procedure names are easy to mistype under time pressure. Add your practice vocabulary to the custom dictionary, then review the dictated note."},{"title":"A local path for sensitive or offline work.","body":"For device-only processing or without a dependable connection, use a local transcription model. Your device processes the recording, so you can keep working where the signal does not reach."},{"title":"A quick setup should still be a choice.","body":"When speed matters most, Cloud avoids a local model download and discards transcription audio immediately after transcription."}]},"features":{"label":"What you get","title":"Clinical notes, ready to review before the next patient.","subtitle":"Cloud transcribes without a model download. A local model is available for sensitive or offline work.","items":[{"image":{"label":"A dentist in navy scrubs sits at a laptop, speaking while holding a pen over a paper chart in an empty treatment room."},"title":"Cloud transcription without a local model download","body":"Dictate an exam, treatment or follow-up note into your microphone. Cloud transcribes without a local model download, then discards transcription audio immediately."},{"image":{"label":"A dentist on a stool beside a laptop talks with a patient sitting upright in the dental chair."},"title":"Local processing for sensitive and offline work","body":"Choose a local model when device-only processing or an offline day is the right fit. Review the editable text before you add it to the patient chart."},{"image":{"label":"A dentist at a laptop talks with an older patient reclined in the chair wearing a dental bib."},"title":"Dictation wherever your cursor is","body":"Press a hotkey and dictate into a dental chart, email or document. Clean, punctuated text lands where you are already working."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Sensitive audio sent to us","value":"None","footnote":"in local mode, transcription never leaves your device"},{"label":"Local dictation","value":"$$0","footnote":"unlimited local AI models on every plan"},{"label":"Languages supported","value":"100+","footnote":"on every plan"}]},"howItWorks":{"label":"How it works","title":"A clinical note transcribed before the next patient.","subtitle":"Start free. Cloud transcribes without a local model download, while a local model is available for sensitive and offline work. Review the text before the next appointment.","image":{"label":"A dentist in a white coat works at a laptop at her office desk, a folder beside her and the treatment room visible behind."},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Use Cloud without a local model download, or use a local model for sensitive and offline work."},{"title":"Dictate the note.","body":"Speak the clinical note, and OpenWhispr transcribes it into editable text."},{"title":"Review and add to the chart.","body":"Check the text, paste it into your dental software, and record it using your practice process."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Reasons to choose","title":"10 reasons OpenWhispr fits dental dictation","subtitle":"A dental dictation workflow needs speed, review and a practical choice of processing mode. Here are the product facts.","items":[{"title":"Unlimited local models at no cost","body":"Every plan includes unlimited local AI models, so local speech-to-text stays free."},{"title":"Cloud audio is never stored","body":"OpenWhispr Cloud receives audio to transcribe it, then immediately discards it."},{"title":"Local transcription works offline","body":"Local models run entirely on your device with no internet required."},{"title":"A custom dictionary for practice vocabulary","body":"Add names, jargon and acronyms so the transcript can better reflect the words your practice uses."},{"title":"Dictate into the active field","body":"Use the desktop app to dictate where your cursor is, whether that is a chart field, email or document."},{"title":"100+ languages supported","body":"Every plan supports transcription in more than 100 languages."},{"title":"Start with 2,000 Cloud words each week","body":"The Free plan includes 2,000 words of OpenWhispr Cloud transcription per week."},{"title":"Use your own API keys","body":"Every plan supports your own OpenAI, Anthropic, Gemini or Groq API keys for unlimited cloud transcription."},{"title":"Upgrade for unlimited Cloud transcription","body":"Pro includes unlimited OpenWhispr Cloud transcription, plus sync, mobile and API/MCP access."},{"title":"Open source under MIT","body":"OpenWhispr is open source, so your technical team can inspect the software."}]},"faq":{"label":"Before you download","title":"The questions dentists ask first.","subtitle":"Straight answers on clinical notes, processing modes, cost and workflow.","items":[{"question":"Does audio ever leave my device?","answer":"With a local transcription model, the recording stays on your device. With OpenWhispr Cloud, audio is sent only to transcribe it, then immediately discarded. Choose the mode that fits your workflow."},{"question":"Can I use this for dental clinical notes?","answer":"OpenWhispr turns your spoken words into editable text. Dictate the note while the appointment is fresh, then review it and enter it into your dental record according to your practice process."},{"question":"Can it handle dental terminology?","answer":"Use the custom dictionary to teach OpenWhispr the names, jargon and acronyms your practice uses. Review the transcript before it becomes part of the clinical record."},{"question":"Will it work without internet?","answer":"Yes. Local models run entirely on your device and do not need an internet connection. Use a local model when you need that offline path."},{"question":"Does it write the clinical record for me?","answer":"It transcribes spoken words into editable text. You remain responsible for reviewing the note and recording it according to your clinical and practice requirements."},{"question":"What do I get by upgrading?","answer":"Free includes unlimited local models and a limited Cloud allowance. Pro adds unlimited OpenWhispr Cloud transcription, more meeting-recording time, sync, mobile and API/MCP access. Business adds unlimited meeting recordings, Agent mode and chat over data."}]},"reviews":{"label":"What users say","title":"Voice input that earns a place in the workflow.","subtitle":"Real OpenWhispr user reviews.","score":{"value":"0","label":"Audio retained after Cloud transcription","body":"Cloud audio is sent to transcribe it, then immediately discarded. Local transcription stays on your device."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Start at $0.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for dentists","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"Unlimited local AI models","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Clinicians who want local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Clinicians who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Teams that need Agent mode, chat over data, and unlimited meeting recordings"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Useful Free access, including local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud without local models to download, then the audio is never stored.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Dictation software for the work between appointments.","paragraphs":["OpenWhispr helps dentists turn spoken clinical notes into editable text, so the details are ready to review while the appointment is still fresh.","It is built for people who want practical control over their workflow, whether you work alone or with a whole dental team.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"OpenWhispr founder photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Dictate the note. Review the text. Move to the next patient.","body":"Download OpenWhispr for free and turn spoken clinical notes into editable text with Cloud or local processing.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Download for Mac, Windows or Linux. If you need help, ask us in Discord."}}},"useCaseHearingImpaired":{"breadcrumb":{"home":"Home","useCases":"Use Cases","hearingImpaired":"For Deaf and Hard-of-Hearing People"},"metadata":{"title":"Free Speech-to-Text for Hearing Impaired - OpenWhispr","description":"OpenWhispr turns recorded speech into text you can read. Use it free with local models, or choose Cloud for fast transcription with no model download.","keywords":["best speech-to-text app for hearing impaired free","free speech-to-text app for hearing impaired","speech-to-text for deaf people","speech-to-text for hard of hearing","free audio transcription app","offline speech-to-text app"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text application","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source speech-to-text software for turning recorded speech into readable text. Choose Cloud for top accuracy and speed without a local model download, or local for private and offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text app for hearing-impaired people","title":"Record the conversation. Read it back as text.","body":"Use Meeting Notes to record conversations and turn them into a transcript you can read, review and keep. Choose Cloud for fast transcription without a model download, or local processing for private and offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Two students sitting together at an outdoor campus table, looking at a laptop screen; one wears a hearing aid"}},"trustBar":{"label":"Save text in:","names":["Microsoft Word","Google Docs","Apple Notes","Notion","Slack","Email"]},"pain":{"label":"When speech is hard to follow","title":"The words disappear before you can ask for them again.","subtitle":"You need text you can return to, a mode that fits the moment, and a free way to start.","cards":[{"title":"The speaker has already moved on.","body":"A missed name, address or next step is hard to reconstruct from memory. OpenWhispr turns recorded speech into text you can reread before the details are gone."},{"title":"You need a private or offline option.","body":"Choose local transcription when the conversation is private or you have no connection. Your device processes the audio, so you can keep working without sending it away."},{"title":"Missing a sentence is tiring.","body":"Instead of piecing a detail together from memory, search and reread the transcript. Keep the text where you can return to it."},{"title":"You want the quickest setup first.","body":"When speed matters most, Cloud avoids a local model download and discards transcription audio immediately after transcription."}]},"features":{"label":"What you get","title":"Text you can read back. A mode that fits the moment.","subtitle":"Use Cloud for fast transcription or local processing for private and offline situations.","items":[{"image":{"label":"Three colleagues talking around a meeting table in a glass-walled office, an open laptop between them; one wears a hearing aid"},"title":"Cloud transcription without a local model download","body":"Add recorded speech and get readable text quickly. Cloud gives you top accuracy and speed without a local model download, then discards the transcription audio immediately."},{"image":{"label":"Two women in conversation across a bright studio table with an open laptop between them; one wears a cochlear implant"},"title":"Local processing for private or offline moments","body":"Choose a local model when you need device-only processing or no connection. Read and review the text on your own device."},{"image":{"label":"A woman wearing a cochlear implant reading on a tablet at a lecture-hall desk, an open notebook and pen beside her"},"title":"Keep the transcript with your notes","body":"OpenWhispr Meeting Notes keeps the conversation as a transcript you can revisit. Ask the Agent to search across your meetings and bring back the details you need. Copy useful text into a note, document, email or chat when you need it elsewhere."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Audio sent away in local mode","value":"None","footnote":"local transcription runs on your device"},{"label":"Cost to start","value":"$$0","footnote":"unlimited local models, forever"},{"label":"Setup time","value":"No fixed time","footnote":"choose the setup that fits you"}]},"howItWorks":{"label":"How it works","title":"A written record before the details fade.","subtitle":"Start free, choose Cloud for a quick high-performance setup or local for private and offline work, then read the words back when you need them.","image":{"label":"Three people leaning over a laptop at a wooden table in a community space, one writing in a notebook"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Start free with unlimited local models that work offline, or choose Cloud when you do not want a model download."},{"title":"Start Meeting Notes.","body":"Record the conversation in Meeting Notes and turn it into a transcript you can read."},{"title":"Read, save, revisit.","body":"Review the transcript, save the useful parts in your notes, and return to the text when you need a detail."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 practical reasons to try OpenWhispr","subtitle":"Speech-to-text should be practical, private when you need it, and free to try. Here are the product facts.","items":[{"title":"Use local transcription when privacy matters","body":"Local AI models run entirely on your device with no internet required. Use them when the conversation calls for device-only processing."},{"title":"Start free with unlimited local models","body":"Free includes unlimited local AI models, plus 2,000 Cloud transcription words each week."},{"title":"Read back the words instead of replaying everything","body":"A transcript gives you a written record to scan, search and save when a detail was missed the first time."},{"title":"Keep going without a connection","body":"Local models run offline. Use them on a commute, in a waiting room, or anywhere reliable internet is not available."},{"title":"Keep useful text in the tools you already use","body":"Copy a transcript into a note, document, email or chat and leave yourself a record you can return to."},{"title":"Choose Cloud when you need speed","body":"Cloud transcription delivers top accuracy and speed without a local model download."},{"title":"Cloud transcription audio is immediately discarded","body":"When you choose OpenWhispr Cloud, audio is sent only to transcribe it, then never stored."},{"title":"Open source means the project is inspectable","body":"OpenWhispr is open source under the MIT license, so anyone can inspect the project rather than take its word for it."},{"title":"Teach it the names you need to get right","body":"Add names, places and recurring terms to the custom dictionary for sharper transcripts."},{"title":"Work across 100+ languages","body":"OpenWhispr supports 100+ languages on every plan, so the spoken language does not have to be English."}]},"faq":{"label":"Before you download","title":"The questions to ask before you download.","subtitle":"Straight answers on privacy, cost and what speech-to-text can do. If yours is not here, ask us in Discord.","items":[{"question":"Does the audio ever leave my device?","answer":"With a local transcription model, the recording stays on your device. With OpenWhispr Cloud, audio is sent only to transcribe it, then immediately discarded. Choose the mode that fits the conversation and your needs."},{"question":"Is this a replacement for hearing aids or assistive listening devices?","answer":"No. OpenWhispr is speech-to-text software for turning recorded speech into readable text. It does not replace hearing aids, cochlear implants, captions provided by a service, or professional accessibility support."},{"question":"Can it handle names, places and unfamiliar words?","answer":"The custom dictionary lets you add names, places, jargon and acronyms for sharper transcripts. OpenWhispr also supports 100+ languages."},{"question":"Will it work with no signal?","answer":"Yes. Local models run entirely offline. Choose local before you need it, then transcribe on your device without an internet connection."},{"question":"Does it provide live captions?","answer":"This page covers turning recorded speech into text you can read and revisit. It makes no claim that OpenWhispr provides live captions for a class, meeting or medical appointment."},{"question":"Is it actually free?","answer":"Yes. Free includes unlimited local AI models, 2,000 Cloud transcription words a week, and five hours of meeting recordings a month. Upgrade to Pro for unlimited Cloud transcription, more meeting-recording time, sync, mobile, and API/MCP access. Business adds unlimited meeting recordings, Agent mode, and chat over data."}]},"reviews":{"label":"What users say","title":"Built for people who want words they can read back.","subtitle":"These users came for a quicker, more dependable way to turn speech into text.","score":{"value":"0","label":"Stored Cloud transcription audio","body":"In local mode the audio stays on your device. With OpenWhispr Cloud, audio is sent only to transcribe it, then immediately discarded."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for hearing-impaired people","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"People who want a useful free start with local transcription and limited Cloud transcription"}]},{"parts":[{"text":"People who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"People transcribing at volume with Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Useful Free access, including local transcription and limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some spoken words need to become text you can keep.","paragraphs":["OpenWhispr started with a simple idea: spoken words should be easy to turn into text you can read, search and keep.","OpenWhispr is open source under the MIT license. Local AI models run on your device with no internet required.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team and community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Keep the spoken words you need in text.","body":"Download OpenWhispr, start Meeting Notes for the next conversation, then read and revisit the transcript when you need the details. Start free on your computer.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free on Mac, Windows & Linux. Community support in Discord."},"byline":{"authorLabel":"Written by","author":"Joshua Padoa","aiDisclosure":"Drafted with AI assistance","reviewerLabel":"Reviewed and fact checked by","reviewer":"Wesley van der Hoop","updatedLabel":"Last reviewed","dateModified":"2026-08-27"}}},"useCaseDyslexia":{"breadcrumb":{"home":"Home","useCases":"Use Cases","dyslexia":"For Dyslexia"},"metadata":{"title":"Best Speech-to-Text App for Dyslexia - OpenWhispr","description":"OpenWhispr is a free, open-source speech-to-text app for dyslexia. Speak naturally, get clean text, and choose local or Cloud transcription.","keywords":["speech to text app for dyslexia","best speech to text app for dyslexia","dyslexia speech to text","voice typing for dyslexia","dictation software for dyslexia","free speech to text for dyslexia"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text software","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source speech-to-text software for people with dyslexia. Choose Cloud for speed without a local model download, or local for device-only and offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text software for people with dyslexia","title":"Focus on what you want to say, not how to spell it.","body":"When spelling or typing interrupts your flow, speak instead. OpenWhispr turns your voice into editable text wherever your cursor is, so you can get the idea down and review it afterward.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Man speaking at his laptop in a design studio, sketches pinned on the wall behind him"}},"trustBar":{"label":"Write in:","names":["Microsoft Word","Google Docs","Notion","Gmail","Canvas","Blackboard"]},"pain":{"label":"When typing breaks the flow","title":"You know what you want to say. Getting it typed is the hard part.","subtitle":"Start with your voice, keep the words moving, and stay with the app in front of you.","cards":[{"title":"The thought arrives faster than the sentence.","body":"When spelling or typing interrupts the thought, the blank page takes over. OpenWhispr turns what you say into text, so you can start with the idea instead."},{"title":"You should not have to change apps to write.","body":"A new writing tool can add another hurdle. Use a hotkey and dictate into the text field you already have open, then keep working from there."},{"title":"A misspelled word can stop the next sentence.","body":"Use the custom dictionary for names, course terms, and words you use often. It helps guide the transcription toward the words you mean."},{"title":"The quick option and the private option should both be useful.","body":"Cloud starts without a local model download. Local keeps transcription on your device and works offline. Pick the path that lets you get the words down."}]},"features":{"label":"What you get","title":"Start with your voice. Get the words down.","subtitle":"Say the word, get editable text, and keep the idea moving. Choose Cloud for no model download, or local transcription for on-device and offline writing.","items":[{"image":{"label":"Student speaking at her laptop in a library, a closed book resting under her hands"},"title":"Speak instead of starting with the keyboard","body":"Hold your hotkey and speak when the keyboard would slow you down. OpenWhispr gives you editable text ready to paste where you are writing, then you can review and correct it."},{"image":{"label":"Woman sitting on a couch in a bright office lounge, speaking at a laptop on a low table"},"title":"Cloud transcription when you want to start fast","body":"Cloud gives you top accuracy and speed without a local model download. The transcription audio is discarded immediately after transcription."},{"image":{"label":"Man speaking at his laptop on a moving train, countryside blurring past the window"},"title":"Local transcription when you need it on your device","body":"Local models run entirely on your device and work without an internet connection. Pick this path for device-only processing or a day with unreliable Wi-Fi."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Audio sent to us in local mode","value":"None","footnote":"local transcription stays on your device"},{"label":"Cost to start","value":"$$0","footnote":"unlimited local dictation, forever"},{"label":"Choose the way you write","value":"Cloud or local","footnote":"start with Cloud or download a local model for offline writing"}]},"howItWorks":{"label":"How it works","title":"Get the thought down before it disappears.","subtitle":"Download free, choose the setup that fits, then say the next sentence before the thought disappears.","image":{"label":"Woman at a kitchen table speaking while working at her laptop, printed pages and reading glasses beside her"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Start with Cloud if you want no model download. Choose local transcription for on-device and offline writing."},{"title":"Speak the sentence.","body":"Press your hotkey and say what you mean. OpenWhispr turns your speech into text."},{"title":"Read and edit.","body":"Keep the words that sound like you, change the ones that do not, and move on to the next idea."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Keep the thought moving","title":"10 product facts for speech-to-text and dyslexia","subtitle":"Say the word instead of stopping at the keyboard. Then review the text and keep the idea moving.","items":[{"title":"Keep moving past the hard word","body":"Use a hotkey to turn the next thought into text. The first pass gives you something to review instead of a blank page."},{"title":"Unlimited local dictation is free","body":"Run local speech-to-text on every plan at $0. There is no minute meter running while you get your words down."},{"title":"Cloud is quick to start","body":"Choose Cloud for top accuracy and speed without a local model download. It is the quickest way to start dictating."},{"title":"Local works without Wi-Fi","body":"Choose a local model when the internet is unreliable or you want device-only processing. Your speech stays on the computer doing the work."},{"title":"Use the apps you already know","body":"Dictate into the text field in front of you, then stay in Word, Docs, email, or the learning platform you already use."},{"title":"Make names and key terms easier to catch","body":"Add names, course terms, and recurring words to the custom dictionary. OpenWhispr sends them to transcription as hints, giving it more context for the words you mean."},{"title":"Cloud audio is not kept","body":"When you use OpenWhispr Cloud, audio is sent only to transcribe it and is immediately discarded after transcription."},{"title":"Open source means you can inspect it","body":"OpenWhispr is open source. You can look at how the app handles your speech instead of taking the claim on faith."},{"title":"Keep writing in more than one language","body":"OpenWhispr supports 100+ languages, so the speech-to-text option can fit the language you need to use."},{"title":"Upgrade only when Cloud and workflow tools help","body":"Free includes local models and a limited Cloud allowance. Pro adds unlimited Cloud transcription, sync, mobile, and API or MCP access."}]},"faq":{"label":"Before you download","title":"Questions about speech-to-text for dyslexia.","subtitle":"Straight answers before you try speech-to-text: writing, privacy, cost, and Cloud or local transcription.","items":[{"question":"What changes when spelling slows me down?","answer":"Speak the sentence when typing or spelling interrupts your flow. OpenWhispr gives you editable text to review and correct afterward. It does not diagnose, treat, or guarantee outcomes for dyslexia."},{"question":"Should I choose Cloud or local transcription?","answer":"Choose Cloud when you want top accuracy and speed without downloading a local model. Choose a local model when you want device-only processing or need to write offline. Cloud audio is sent only to transcribe it, then immediately discarded."},{"question":"How can I give recurring names and terms a better chance of coming through?","answer":"Use the custom dictionary for names, course terms, acronyms, and recurring words. OpenWhispr sends them to transcription as hints, giving it more context for the words you mean."},{"question":"Will it work when I have no internet?","answer":"Yes, with a local model. Local transcription runs entirely on your device after the model download, so you can dictate when the Wi-Fi is unreliable or unavailable."},{"question":"Can I use it in Word, Google Docs, or email?","answer":"Yes. Press a system-wide hotkey and dictate into the text field where you are already working. The text lands at your cursor, so you can keep using the app you know."},{"question":"Is it actually free?","answer":"Yes. Free includes unlimited local transcription, local AI models, and 2,000 Cloud words a week. Upgrade to Pro for unlimited Cloud transcription, more meeting-recording time, sync, mobile, and API or MCP access. Business adds unlimited meeting recordings, Agent mode, and chat over data."}]},"reviews":{"label":"What users say","title":"For people who would rather talk than fight the keyboard.","subtitle":"These are general OpenWhispr reviews, not testimonials about dyslexia.","score":{"value":"0","label":"Audio stored on our servers in local mode","body":"With local transcription, audio stays on your device. With Cloud, audio is sent only to transcribe it, then immediately discarded."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for people with dyslexia","tiers":[{"id":"free","name":"Free","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"}},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","badge":"Most Popular","cta":{"label":"Start with Business"},"featured":true}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"People who want unlimited local transcription and a free start with Cloud"}]},{"parts":[{"text":"People who want unlimited Cloud transcription plus mobile, sync, and API or MCP access"}]},{"parts":[{"text":"People recording at volume with Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local transcription plus a limited Cloud allowance"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Your ideas should not get stuck at the keyboard.","paragraphs":["OpenWhispr started with a simple idea: getting words onto the page should not depend on getting every keystroke right the first time.","We build in public. Every release and every line of code is on GitHub, where people can inspect how local transcription works instead of taking it on faith.","Community support is available in Discord."],"image":{"label":"OpenWhispr team and community"}},"finalCta":{"tagline":"Free · open source · speech-to-text","title":"Say the next sentence instead of stopping at the keyboard.","body":"Download OpenWhispr, speak the next thought while it is fresh, then paste, review, and edit the text before continuing.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free forever on Mac, Windows, and Linux."}}},"useCaseRsi":{"breadcrumb":{"home":"Home","useCases":"Use Cases","rsi":"For People with RSI"},"metadata":{"title":"Voice to Text Apps for RSI - OpenWhispr","description":"OpenWhispr turns speech into editable text in your current app, helping people with RSI reduce sustained keyboard and mouse input.","keywords":["voice to text apps for rsi","voice to text for repetitive strain injury","rsi dictation software","speech to text for wrist pain","hands free typing app","accessibility voice typing"]},"schema":{"name":"OpenWhispr","applicationCategory":"Voice-to-text accessibility software","operatingSystem":"macOS, Windows, Linux","description":"Open-source voice-to-text software that lets people with RSI dictate editable text into the current app, then review and continue with less sustained manual input.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Voice to text apps for people with RSI","title":"Give your hands a break. Speak instead.","body":"Reduce the keyboard and mouse work that can become uncomfortable with RSI. Speak naturally to turn your words into text for messages, forms and notes. Choose fast Cloud transcription or local processing for offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Man in a wrist brace speaking at a laptop in a home office"}},"trustBar":{"label":"Use alongside:","names":["Gmail","Google Docs","Slack","Microsoft Word","Notion","ChatGPT"]},"pain":{"label":"When typing is the hard part","title":"The work is manageable. The keyboard and mouse time adds up.","subtitle":"A long email, form or document can turn into a stop-start job when your hands need a break. Speak instead, then review what lands on the page.","cards":[{"title":"A short reply becomes a long typing session.","body":"In tap mode, press the hotkey once to start and again to stop, then review the transcribed reply before you send it."},{"title":"A flare-up does not wait for a quieter week.","body":"When typing is difficult today, use your voice for the email, note or form in front of you. OpenWhispr gives you another way to get the words down."},{"title":"Cloud should be quick to start.","body":"Choose Cloud when you want fast transcription without downloading a local model. Audio is sent only to transcribe it, then immediately discarded."},{"title":"An internet drop should not end the task.","body":"Choose a local model when you need to work offline. Local transcription runs on your device, so you can keep dictating without a connection."}]},"features":{"label":"What you get","title":"Use your voice for the draft. Keep manual input for review.","subtitle":"Dictate into the app you already have open, then review and refine the returned text instead of entering every word manually.","items":[{"image":{"label":"Man speaking at a laptop beside an ergonomic keyboard"},"title":"Speak while you work","body":"Press a hotkey and dictate the email, document, chat or form in front of you. OpenWhispr pastes transcribed text into the active app when automatic pasting is available, or copies it to your clipboard."},{"image":{"label":"Woman in a wrist brace speaking at a laptop in an open studio office"},"title":"Cloud transcription without a local model download","body":"Start speaking without waiting for a local model download. Cloud transcribes with top accuracy and speed, then immediately discards the audio."},{"image":{"label":"Woman resting her hands on a cushion while working at a laptop at home"},"title":"A local option for offline work","body":"Choose a local model when you need device-only processing or no internet connection. Your computer processes the transcription while you stay in control of the workflow."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Audio sent away in local mode","value":"None","footnote":"local transcription runs on your device"},{"label":"Cost to start","value":"$$0","footnote":"unlimited local AI models, forever"},{"label":"Steps to your first dictation","value":"3","footnote":"download, dictate, review"}]},"howItWorks":{"label":"How it works","title":"A reply in the box with less sustained manual input.","subtitle":"Choose the interaction that suits the task, dictate into the app already open, then review the returned text before continuing.","image":{"label":"Woman in a wrist brace speaking at a laptop in an art studio"},"steps":[{"title":"Download for free.","body":"Mac, Windows or Linux. Start with unlimited local AI models, or use the included Cloud allowance."},{"title":"Choose tap or hold, then speak.","body":"Use tap mode for separate start and stop presses, or hold the hotkey while speaking. Dictate the email, form answer or note in the app already open."},{"title":"Review and send.","body":"Read the text, make any edits you want, then carry on with the task."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Built for the next sentence","title":"10 reasons to try OpenWhispr with RSI","subtitle":"OpenWhispr does not treat RSI. It gives you a practical way to dictate the draft, then use the keyboard or mouse to review and edit the returned text.","items":[{"title":"A hotkey starts dictation","body":"Use a keyboard shortcut to start speaking without hunting for a voice button in every app."},{"title":"Review before you send","body":"Read the transcribed text and make any edits you want before you send or save it."},{"title":"Cloud gets you speaking sooner","body":"Choose Cloud when you want top accuracy and speed without first downloading a local model."},{"title":"Local transcription works offline","body":"Run a local AI model entirely on your device when you do not have an internet connection."},{"title":"Cloud audio is immediately discarded","body":"OpenWhispr Cloud receives audio only to transcribe it, then never stores it."},{"title":"The dictionary learns your words","body":"Add names, acronyms and specialist terms to the custom dictionary for sharper transcripts."},{"title":"100+ languages are supported","body":"Every plan supports 100+ languages."},{"title":"Local models are unlimited and free","body":"Every plan includes unlimited local AI models, so there is no per-minute local dictation bill."},{"title":"Bring your own API keys","body":"Use your own OpenAI, Anthropic, Gemini or Groq keys for unlimited cloud transcription with no OpenWhispr limits."},{"title":"Upgrade when the workflow needs more","body":"Pro adds unlimited OpenWhispr Cloud transcription, sync, mobile, and API/MCP access. Business adds Agent mode and chat over data."}]},"faq":{"label":"Before you download","title":"The questions people with RSI ask first.","subtitle":"Straight answers on voice input, privacy and cost.","items":[{"question":"Is OpenWhispr a treatment for RSI?","answer":"No. OpenWhispr is speech-to-text software, not a diagnosis or treatment. It gives you a way to use your voice for text input when that suits your workflow."},{"question":"Can I dictate into the apps I already use?","answer":"Yes. Press a hotkey and dictate the email, document, chat or form in front of you. OpenWhispr pastes transcribed text into the active app when automatic pasting is available. On Wayland without a paste tool, it is copied to your clipboard for Ctrl+V."},{"question":"Should I choose Cloud or local transcription?","answer":"Choose Cloud when you want top accuracy and speed without downloading a local model. Choose local when you need device-only processing or want to work offline. Both are deliberate options for different situations."},{"question":"Does Cloud keep my audio?","answer":"No. OpenWhispr Cloud receives audio only to transcribe it, then immediately discards it. Local transcription runs on your device and does not need an internet connection."},{"question":"Can it handle names, jargon, and drafting?","answer":"Use the custom dictionary for names, acronyms and specialist terms. OpenWhispr turns speech into text. You can also connect your own AI key to turn dictated text into a draft, then review it yourself."},{"question":"Is it free, and what do paid plans add?","answer":"Free includes unlimited local AI models, 2,000 Cloud words each week, and five hours of meeting recordings each month. Pro adds unlimited Cloud transcription, 20 meeting-recording hours, sync, mobile, and API/MCP access. Business adds unlimited meeting recordings, Agent mode, and chat over data."}]},"reviews":{"label":"What users say","title":"For people who would rather talk than type.","subtitle":"What users told us in their own words.","score":{"value":"$$0","label":"To start dictating","body":"Free includes unlimited local AI models. Choose a paid plan when Cloud transcription or workflow features become useful."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for people with RSI","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"People with RSI who want unlimited local dictation and a useful Cloud allowance"}]},{"parts":[{"text":"People with RSI who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"People who need unlimited meeting recordings, Agent mode, and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local AI models plus a limited Cloud allowance"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Prolonged keyboard and mouse work should not be the only way to get words out.","paragraphs":["OpenWhispr started with a simple objection: someone whose hands need a break should not have to stop getting words onto the page.","OpenWhispr is open source under the MIT license, so the code is available for anyone to inspect. Local AI models run entirely on your device, and you can choose them when that is the right fit for your workflow.","Free includes Community support. Pro includes Email support, and Business includes Priority support."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"When sustained manual input adds friction, let your voice take the next turn.","body":"Download OpenWhispr and dictate the next email, form or note. Review the words, make your edits, and keep the task moving.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Start with three steps: download, dictate, review. Free forever on Mac, Windows & Linux."},"byline":{"authorLabel":"Written by","author":"Joshua Padoa","aiDisclosure":"Drafted with AI assistance","reviewerLabel":"Reviewed and fact checked by","reviewer":"Wesley van der Hoop","updatedLabel":"Last reviewed","dateModified":"2026-08-27"}}},"useCaseIpad":{"breadcrumb":{"home":"Home","useCases":"Use Cases","iPad":"For iPad Users"},"metadata":{"title":"Free Speech to Text on iPad? Check the Plan - OpenWhispr","description":"Capture notes and first drafts by voice on iPad, then continue them on desktop. OpenWhispr’s iPad app and sync are included with Pro.","keywords":["best free speech to text app for ipad","free speech to text app for ipad","ipad speech to text app","ipad dictation app","voice to text for ipad","speech to text ipad"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text app for iPad users","operatingSystem":"macOS, Windows, Linux, iOS","description":"Dictate editable text on iPad and continue synced OpenWhispr notes on desktop. The iPhone & iPad app is included with Pro; Free starts on desktop.","price":"8","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text app for iPad users","title":"Say it on iPad. Finish it on your desktop.","body":"Start the follow-up while the meeting is still fresh. Pro includes the iPhone & iPad app and syncs notes and agent conversations across devices; Free starts on desktop with unlimited local AI models.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"A man speaking to an iPad he is holding at a library table"}},"trustBar":{"label":"Your everyday iPad apps:","names":["Apple Notes","Pages","Microsoft Word","Google Docs","Notion","Obsidian"]},"pain":{"label":"The catch behind the free search","title":"Capture the work while the context is still fresh.","subtitle":"A meeting follow-up, note or rough first draft should not wait until you are back at a full keyboard.","cards":[{"title":"Free should mean useful, not a three-minute trial.","body":"Free includes unlimited local AI models on macOS, Windows and Linux. It also includes 2,000 OpenWhispr Cloud transcription words each week and five hours of meeting recordings each month."},{"title":"The iPad app is not free. It is a Pro feature.","body":"OpenWhispr does have an iPhone & iPad app, but it is included with Pro. That is the answer before you spend time installing anything."},{"title":"Follow up before you are back at your desk.","body":"Start the note while the meeting is still fresh, then pick it up on desktop."},{"title":"Cloud speed needs a clear data path.","body":"OpenWhispr Cloud avoids a local model download and is built for accuracy and speed. Audio is sent only to transcribe it, then never stored."}]},"features":{"label":"A mobile-to-desktop writing workflow","title":"Speak the draft on iPad. Keep working on desktop.","subtitle":"The mobile app turns speech into editable text on iPad; OpenWhispr Notes provide the supported path for carrying that work to desktop.","items":[{"image":{"label":"A woman tapping an iPad at a kitchen table with a laptop open beside her"},"title":"A useful Free plan on desktop","body":"Use unlimited local AI models on macOS, Windows or Linux. You also get 2,000 OpenWhispr Cloud transcription words per week, five meeting-recording hours per month, 100+ languages and a custom dictionary."},{"image":{"label":"A woman speaking toward an iPad in a keyboard case on a sunny rooftop terrace"},"title":"The iPhone & iPad app with Pro","body":"Use the OpenWhispr keyboard to dictate in any text field. When the work needs to continue inside OpenWhispr, start it in an OpenWhispr Note and pick up that synced note on desktop."},{"image":{"label":"A man in a suit speaking to a propped-up iPad in a hotel lobby"},"title":"Unlimited Cloud transcription with Pro","body":"Choose Pro when you want unlimited OpenWhispr Cloud transcription without downloading a local model. Cloud audio is transcribed, then immediately discarded."}]},"stats":{"ariaLabel":"Plan facts for iPad users","items":[{"label":"Free desktop start","value":"$$0","footnote":"unlimited local AI models on macOS, Windows and Linux"},{"label":"iPhone & iPad app","value":"Pro","footnote":"mobile access is included with the Pro plan"},{"label":"Free Cloud allowance","value":"2,000","footnote":"OpenWhispr Cloud transcription words each week"}]},"howItWorks":{"label":"How it works","title":"Capture it on iPad. Finish it on desktop.","subtitle":"Use an iPad text field for immediate editable copy, or an OpenWhispr Note for work that continues inside OpenWhispr on desktop.","image":{"label":"A woman speaking toward a propped-up iPad at an outdoor cafe table with a closed laptop beside her"},"steps":[{"title":"Open the place you want to write.","body":"Use the OpenWhispr keyboard in an email, document or note. Start an OpenWhispr Note when cross-device continuation is part of the job."},{"title":"Speak the follow-up or first draft.","body":"OpenWhispr returns editable text so you can correct names, details and structure while the context is fresh."},{"title":"Continue the supported work on desktop.","body":"Open the synced OpenWhispr Note on desktop and keep editing. OpenWhispr sync covers notes and agent conversations; text inserted into another app is not added to that sync."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No hidden plan comparison","title":"10 reasons iPad users choose OpenWhispr","subtitle":"Looking for free speech-to-text on iPad deserves a straight answer. These are the limits, upgrades and data paths as the plans define them.","items":[{"title":"The iPad app is clearly a Pro feature","body":"OpenWhispr includes an iPhone & iPad app with Pro. Free is desktop-only, so there is no bait-and-switch in the plan comparison."},{"title":"Free is still a real desktop plan","body":"Free includes unlimited local AI models on macOS, Windows and Linux, plus a limited OpenWhispr Cloud allowance."},{"title":"Pro unlocks unlimited OpenWhispr Cloud transcription","body":"Use Pro when the 2,000 free Cloud words per week are not enough and you want unlimited OpenWhispr Cloud transcription."},{"title":"Pro includes sync across devices","body":"Mobile access comes with device sync, so Pro is built for people who move between their computer and iPhone or iPad."},{"title":"Cloud does not need a local model download","body":"OpenWhispr Cloud is the faster setup when accuracy and speed matter and you do not want to download local models first."},{"title":"Cloud audio is never stored","body":"Audio is sent to OpenWhispr Cloud only to transcribe it, then immediately discarded."},{"title":"Bring your own API keys on any plan","body":"Use your own OpenAI, Anthropic, Gemini or Groq keys for unlimited Cloud transcription without OpenWhispr limits."},{"title":"A custom dictionary is included","body":"Teach OpenWhispr your names, jargon and acronyms for sharper transcripts, whether you start Free or use Pro."},{"title":"100+ languages are supported","body":"Every plan supports 100+ languages, so the plan decision is about access and workflow, not a reduced language list."},{"title":"The upgrade cost is stated upfront","body":"Pro is $8 per month or $80 per year. It adds the iPhone & iPad app, unlimited OpenWhispr Cloud transcription and device sync."}]},"faq":{"label":"Before you download","title":"What iPad users need to know first.","subtitle":"The direct answer on price, mobile access, Cloud transcription and the useful free starting point.","items":[{"question":"Is OpenWhispr a free speech-to-text app for iPad?","answer":"Not on iPad. The Free plan is desktop-only. OpenWhispr's iPhone & iPad app is included with Pro, which also includes unlimited OpenWhispr Cloud transcription and sync across devices."},{"question":"What does the Free plan include?","answer":"Free includes unlimited local AI models on macOS, Windows and Linux, 2,000 OpenWhispr Cloud transcription words per week, five hours of meeting recordings per month, 100+ languages, a custom dictionary and community support."},{"question":"What does Pro add for an iPad user?","answer":"Pro adds the iPhone & iPad app, unlimited OpenWhispr Cloud transcription, 20 hours of meeting recordings per month, sync across devices, personal API access, MCP integration in all your apps and email support."},{"question":"What happens to audio sent to OpenWhispr Cloud?","answer":"Cloud audio is sent only to transcribe it, then immediately discarded."},{"question":"Can I use my own API keys instead?","answer":"Yes. Every plan supports your own OpenAI, Anthropic, Gemini or Groq keys for unlimited Cloud transcription without OpenWhispr limits."},{"question":"When should I choose Business?","answer":"Business adds unlimited meeting recordings, Agent mode, chat over your data and priority support. It is for people who need more than Pro's mobile, sync and unlimited Cloud transcription workflow."}]},"reviews":{"label":"What users say","title":"Built for people who would rather speak than type.","subtitle":"Three unedited reviews from OpenWhispr users.","score":{"value":"1","label":"clear boundary","body":"The plan for iPad access is Pro."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free starts on desktop.","Pro carries OpenWhispr from iPad to desktop."],"caption":"Compare OpenWhispr plans for iPad users","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"People who want a useful free desktop start with local AI models and limited Cloud transcription"}]},{"parts":[{"text":"People who capture notes and drafts on iPad, then continue them in OpenWhispr on desktop"}]},{"parts":[{"text":"People who also need unlimited meeting recordings, Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Useful Free desktop access, including unlimited local AI models and limited Cloud transcription"}]},{"parts":[{"text":"The iPhone & iPad app, sync for OpenWhispr notes and agent conversations, and unlimited OpenWhispr Cloud transcription"}]},{"parts":[{"text":"Unlimited meeting recordings plus Agent mode and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device. It works without an internet connection and is unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"A free plan should tell you exactly where it ends.","paragraphs":["OpenWhispr gives you unlimited local AI models on desktop for free. The iPhone & iPad app is part of Pro, and that distinction belongs in the first answer, not the fine print.","We build in public. The code is open source under the MIT license, so you can inspect the product and choose the plan and data path that fit your work.","And if something breaks, ask in Discord."],"image":{"label":"OpenWhispr team and community"}},"finalCta":{"tagline":"Voice on iPad · editable text · continue on desktop","title":"Start the work on iPad. Finish it on desktop.","body":"Capture the follow-up, note or rough first draft while it is fresh. Use an OpenWhispr Note when you want that work to sync to desktop; iPad access and sync are included with Pro.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free includes unlimited local AI models on macOS, Windows and Linux. The iPhone & iPad app is included with Pro."},"byline":{"authorLabel":"Written by","author":"Joshua Padoa","aiDisclosure":"Drafted with AI assistance","reviewerLabel":"Reviewed and fact checked by","reviewer":"Wesley van der Hoop","updatedLabel":"Last reviewed","dateModified":"2026-08-27"}}},"useCaseArthritis":{"breadcrumb":{"home":"Home","useCases":"Use Cases","arthritis":"For Arthritis"},"metadata":{"title":"Free Dictation App for Arthritis - OpenWhispr","description":"OpenWhispr gives people with arthritis free local dictation, or Cloud transcription for speed without a model download.","keywords":["best dictation apps for arthritis","dictation software for arthritis","voice typing for arthritis","speech to text for arthritis","arthritis voice recognition software","hands-free typing for arthritis"]},"schema":{"name":"OpenWhispr","applicationCategory":"Dictation software for arthritis","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source dictation software for people with arthritis. Choose Cloud for accuracy and speed without a model download, or local processing for offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation software for arthritis","title":"Arthritis makes typing harder. Say the reply. Get text to review.","body":"When you are comparing dictation apps for arthritis, check how much keyboard work the app leaves you. OpenWhispr offers free local dictation for offline work, or Cloud transcription for top accuracy and speed without a model download.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Older woman with grey hair speaking at an open laptop, hands resting flat on the desk, in a bright home study"}},"trustBar":{"label":"Dictate into:","names":["Microsoft Word","Google Docs","Gmail","Slack","Notion","Microsoft Teams"]},"pain":{"label":"When the keyboard hurts","title":"When an email takes more out of you than it should.","subtitle":"An email, a form, a document, a message. Arthritis can make every keystroke a decision. Dictation lets your voice do the typing.","cards":[{"title":"A short email takes too long.","body":"When typing hurts, even a few sentences can turn into a stop-start job. Speak the reply, then paste the transcription into the app you already use."},{"title":"Holding a key is uncomfortable.","body":"Use toggle mode: press once to start dictating and once to stop. You do not need to keep a key held down while you speak."},{"title":"Corrections mean more typing.","body":"A custom dictionary lets you add names, places and terms you use often, so there is less to fix after you dictate."},{"title":"A bad connection should not end the sentence.","body":"Local models run on your device without internet. Dictate at home, in a waiting room or wherever the connection drops out."}]},"features":{"label":"What changes","title":"Speak the words. Keep your hands for the rest.","subtitle":"Choose the route that fits the moment: Cloud for fast, accurate transcription without a model download, or local processing when you need to work offline.","items":[{"image":{"label":"Older man in a green polo shirt speaking at a desktop monitor, one hand resting on a large round desk button, in a bright open-plan office"},"title":"Start and stop without holding a key","body":"Set dictation to toggle mode when press-and-hold is not comfortable. One press starts listening. The next press stops it."},{"image":{"label":"Older woman in a blue patterned blouse speaking at an open laptop, forearms resting on a cushion, at a community centre table"},"title":"Put words where you already work","body":"Dictate into a document, email, chat or web form. OpenWhispr copies the transcription so you can paste it where you are already working."},{"image":{"label":"Woman in a green shirt speaking into a phone held near her mouth, laptop and open notebook on a wooden table in a flower shop"},"title":"Choose speed or offline work","body":"Cloud gives you top accuracy and speed with no local model to download. Local models keep working on your device when internet is not available."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Words you need to type to dictate","value":"None","footnote":"speak into the app and review the text when you are ready"},{"label":"Cost for local dictation","value":"$$0","footnote":"unlimited local AI models on every plan"},{"label":"Languages supported","value":"100+","footnote":"available on every plan"}]},"howItWorks":{"label":"How it works","title":"Download it. Speak the email. Send it.","subtitle":"Start with the free desktop app. Choose a local model for offline work, or use Cloud when you want accuracy and speed without downloading a model.","image":{"label":"Woman in a light blue top speaking at a laptop, one hand on folded fabric beside a sewing machine in a sunlit sewing room"},"steps":[{"title":"Download for free.","body":"Install OpenWhispr on Mac, Windows or Linux. Pick a local model during setup, or begin with the included Cloud allowance."},{"title":"Dictate instead of typing.","body":"Choose push-to-talk or toggle mode, then speak into the document, email or message you already have open."},{"title":"Review and send.","body":"Make any edits you want, then send the text as usual. Your keyboard is there when you need it, not for every word."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"The practical reasons","title":"10 practical reasons to try OpenWhispr for arthritis","subtitle":"The point is not another place to write. It is fewer keystrokes in the apps you already use.","items":[{"title":"Toggle mode does not ask you to hold a key","body":"Press once to start dictating and once to stop. Use it when push-to-talk is not comfortable."},{"title":"It works in the apps you already have open","body":"Dictate into documents, email, chat and web forms without moving your work into a separate editor."},{"title":"Local dictation is free","body":"Every plan includes unlimited local AI models. There is no subscription required to keep using local dictation."},{"title":"Cloud is ready when setup needs to be simple","body":"Cloud transcription is built for top accuracy and speed, with no local model download. Free includes 2,000 words each week."},{"title":"Local models work without internet","body":"Run transcription on your device when Wi-Fi is unreliable or unavailable."},{"title":"A custom dictionary cuts repeat corrections","body":"Add names, places, acronyms and terms you use often to help produce sharper transcripts."},{"title":"100+ languages supported","body":"Dictate in the language you use at home or at work."},{"title":"Cloud audio is discarded after transcription","body":"Audio sent to OpenWhispr Cloud is used to transcribe, then immediately discarded."},{"title":"The app is open source","body":"The code is available under the MIT licence for anyone to inspect."},{"title":"Paid plans remove Cloud limits and add workflow tools","body":"Pro adds unlimited Cloud transcription, sync, the iPhone and iPad app, and API/MCP access. Business adds unlimited meeting recordings, Agent mode and chat over your data."}]},"faq":{"label":"Before you download","title":"Questions people with arthritis ask first.","subtitle":"Straight answers about typing less, setup, cost and privacy.","items":[{"question":"Will I need to hold down a key while I dictate?","answer":"No. OpenWhispr supports toggle mode: press once to start dictating and once to stop. You can choose that when holding a key is uncomfortable."},{"question":"Can I dictate into the apps I already use?","answer":"Yes. Put your cursor in a document, email, chat or web form, then dictate. OpenWhispr copies the transcription so you can paste it wherever you can type."},{"question":"Should I choose Cloud transcription or a local model?","answer":"Choose Cloud when you want top accuracy and speed without downloading a model. Choose a local model when you need to work without internet."},{"question":"Does audio leave my device?","answer":"With a local model, transcription runs on your device. With OpenWhispr Cloud, audio is sent only to transcribe it, then immediately discarded."},{"question":"Can it recognise names and words I use often?","answer":"Yes. Add names, acronyms and other recurring terms to the custom dictionary to help produce sharper transcripts."},{"question":"Is it free, or do I need a paid plan?","answer":"Free includes unlimited local AI models, 2,000 Cloud transcription words each week and 5 hours of meeting recordings each month. Pro adds unlimited Cloud transcription, sync, the iPhone and iPad app, and API/MCP access. Business adds unlimited meeting recordings, Agent mode and chat over your data."}]},"reviews":{"label":"What users say","title":"Built for people who would rather speak than keep typing.","subtitle":"Existing quotes carried forward from the reference page.","score":{"value":"$$0","label":"Cost to try local dictation","body":"Every plan includes unlimited local AI models. You can see whether voice input suits your routine before paying for more Cloud transcription or workflow features."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for arthritis dictation","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"People who want free local dictation and a limited Cloud allowance"}]},{"parts":[{"text":"People who want unlimited Cloud transcription, plus sync, the iPhone and iPad app, and API/MCP access"}]},{"parts":[{"text":"People and teams who want unlimited meeting recordings, Agent mode and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local dictation, plus 2,000 Cloud words each week"}]},{"parts":[{"text":"Unlimited Cloud transcription without managing local models or your own API keys"}]},{"parts":[{"text":"Unlimited meeting recordings plus advanced workflow features"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Typing should not be the price of getting words out.","paragraphs":["OpenWhispr started with a simple objection: people who need to type less should not have to give up the apps and documents they already rely on.","OpenWhispr is open source, so you can inspect the code before you install it. Local models run on your device. Cloud audio is transcribed, then immediately discarded.","Free includes community support. Pro adds email support, and Business adds priority support."],"image":{"label":"Team and community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Let your voice carry more of the typing.","body":"Download OpenWhispr, choose Cloud transcription or a local model, and dictate the next email or document.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free on macOS, Windows and Linux. Paid plans add unlimited Cloud transcription, the iPhone and iPad app, sync and workflow tools."}}},"useCaseIphone":{"breadcrumb":{"home":"Home","useCases":"Use Cases","iphone":"For iPhone"},"metadata":{"title":"Dictation App for iPhone: Speak Notes Fast - OpenWhispr","description":"OpenWhispr is a dictation app for iPhone. Speak longer messages instead of thumb-typing, refine the text by touch, and send it from the app you are using.","keywords":["dictation app for iphone","iphone dictation app","voice to text iphone","iphone speech to text","dictation app ios","voice notes iphone"]},"schema":{"name":"OpenWhispr","applicationCategory":"Dictation app for iPhone","operatingSystem":"macOS, Windows, Linux, iOS","description":"OpenWhispr is a dictation app for iPhone with free, unlimited on-device transcription. Pro adds unlimited Cloud transcription, 20 meeting-recording hours per month, and sync across devices.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation app for iPhone","title":"Say more. Type less on your phone.","body":"Your phone is where conversations and ideas already happen. OpenWhispr turns speech into text on iPhone, so you can capture longer thoughts without typing every word, then refine and send them in the app you are already using.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman walking through a city plaza, speaking into the phone in her hand"}},"trustBar":{"label":"Use OpenWhispr on:","names":["iPhone","iPad","Mac","Windows","Linux","OpenWhispr Cloud"]},"pain":{"label":"Why thumb-typing gets in the way","title":"You have more to say than your thumbs want to type.","subtitle":"Long messages, replies, and updates are often easier to say. The hard part is getting those words into the app where the conversation is already happening.","cards":[{"title":"The conversation is already on your phone.","body":"Messages, email, and work threads happen on the device in your hand. Moving to a laptop just to type a longer response breaks the flow."},{"title":"A voice note still makes someone else listen.","body":"Communication moves faster when speech becomes text you can review, refine, and send in the same app."},{"title":"Switching devices should not mean starting over.","body":"If a message turns into longer work, Pro sync lets you continue an OpenWhispr note on the device that suits the job."},{"title":"Mobile access starts free.","body":"OpenWhispr is free to install on iPhone and iPad. Unlimited on-device transcription needs no account and has no usage limit; Pro is optional."}]},"features":{"label":"What you get on iPhone","title":"Speak in the app where you already write.","subtitle":"Use OpenWhispr on iPhone and iPad for free with unlimited on-device transcription, or add Pro for unlimited Cloud transcription, 20 meeting-recording hours per month, and sync across devices.","items":[{"image":{"label":"Man sitting on stone steps outside, speaking into his phone with a backpack beside him"},"title":"Direct insertion across iPhone apps","body":"Use the OpenWhispr keyboard to turn speech into text directly in the iPhone apps where you write."},{"image":{"label":"Clinician in a white coat sitting outdoors, reading her phone with a laptop in the bag beside her"},"title":"Sync across devices","body":"When a message turns into longer work, Pro sync lets you continue an OpenWhispr note on the device that suits the job."},{"image":{"label":"Traveller in an airport lounge armchair, speaking into his phone by the window"},"title":"Unlimited Cloud transcription with Pro","body":"Pro includes unlimited OpenWhispr Cloud transcription. Cloud is the fast, no-local-model-download option, and transcription audio is never stored."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"iPhone and iPad app","value":"Free","footnote":"free to install"},{"label":"On-device transcription limit","value":"None","footnote":"no account or usage limit"},{"label":"US App Store Pro price","value":"$$9.99","footnote":"$$99.99/year; website checkout pricing differs"}]},"howItWorks":{"label":"How it works","title":"Speak the long version. Edit the details.","subtitle":"When something is easier to say than type, use OpenWhispr on iPhone to turn it into text. Review the wording, then continue across your synced devices.","image":{"label":"Woman standing in an empty sunlit apartment holding keys, speaking into her phone"},"steps":[{"title":"Open the app where you want to write.","body":"In a message, email, document, or other iPhone text field, switch to the OpenWhispr keyboard."},{"title":"Say the long version.","body":"Tap the microphone and speak in context instead of typing every word with your thumbs."},{"title":"Speak. Refine by touch. Continue.","body":"Turn speech into text in the iPhone app where you write. For an OpenWhispr note, review the wording by touch, then continue across synced devices."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"The practical reasons","title":"10 things to know before choosing OpenWhispr on iPhone","subtitle":"Start with the plan and device that fit the work. These are the current product facts behind the mobile option.","items":[{"title":"The iPhone app is free to install","body":"Use unlimited on-device transcription without an account or usage limit."},{"title":"Your devices can stay in sync","body":"Pro keeps OpenWhispr notes available across devices when a message turns into longer work."},{"title":"Cloud transcription is unlimited on Pro","body":"Pro includes unlimited OpenWhispr Cloud transcription."},{"title":"Cloud does not need a local model download","body":"OpenWhispr Cloud is the no-local-model-download option when speed and simple setup matter."},{"title":"Cloud transcription audio is never stored","body":"Audio sent to OpenWhispr Cloud is transcribed, then immediately discarded."},{"title":"On-device transcription stays free on iPhone","body":"Private mode runs on your iPhone; on-device transcription is unlimited and does not require an account."},{"title":"Custom dictionary is included","body":"Add names, jargon, and acronyms to improve the words you use most."},{"title":"100+ languages are supported","body":"OpenWhispr supports 100+ languages on every plan."},{"title":"App Store pricing is channel-specific","body":"In the US App Store, Pro is $9.99 a month or $99.99 a year; website checkout pricing is separate."},{"title":"The upgrade has a defined job","body":"Pro adds unlimited Cloud transcription, 20 hours of meeting recording per month, and sync across devices."}]},"faq":{"label":"Before you choose a plan","title":"The questions iPhone users ask first.","subtitle":"Straight answers on free iPhone access, on-device and Cloud transcription, and the optional Pro upgrade.","items":[{"question":"Is there an OpenWhispr app for iPhone?","answer":"Yes. The iPhone and iPad app is free to install, and unlimited on-device transcription works without an account or usage limit."},{"question":"Do I need Pro to use OpenWhispr on iPhone?","answer":"No. Pro is optional. It adds unlimited Cloud transcription, 20 hours of meeting recording per month, and sync across devices."},{"question":"What does the free plan include?","answer":"Free includes 2,000 words a week of OpenWhispr Cloud transcription, five hours of meeting recordings a month, unlimited on-device transcription without an account or usage limit, 100+ languages, a custom dictionary, and zero data retention."},{"question":"Does OpenWhispr Cloud keep my audio?","answer":"No. Audio sent to OpenWhispr Cloud is used to transcribe, then immediately discarded. Cloud sync and transcription-audio retention are separate choices."},{"question":"When should I use on-device transcription?","answer":"Use on-device transcription on iPhone when you want offline, device-only processing. It is free, unlimited, and does not require an account."},{"question":"Why upgrade to Business?","answer":"Business adds unlimited meeting recordings, Agent mode, chat over your data, and priority support on top of Pro."}]},"reviews":{"label":"What users say","title":"For people who would rather speak than stop moving.","subtitle":"Three unedited quotes on local models, travel, and getting clean text back.","score":{"value":"Free","label":"Unlimited on-device transcription","body":"The iPhone and iPad app is free to install. Pro is optional and adds unlimited Cloud transcription, 20 meeting-recording hours per month, and sync across devices."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Website pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr website plans for iPhone users","tiers":[{"id":"free","name":"Free","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"iPhone users who want unlimited on-device transcription without an account"}]},{"parts":[{"text":"iPhone and iPad users who want unlimited Cloud transcription and sync"}]},{"parts":[{"text":"Teams that need unlimited meeting recordings, Agent mode, and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited on-device iPhone transcription plus limited OpenWhispr Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription, 20 meeting hours, and sync"}]},{"parts":[{"text":"Unlimited meeting recordings, Agent mode, and chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Easy: install the iPhone app, download a model, and use on-device transcription without an account"}]},{"parts":[{"text":"Optional upgrade for unlimited cloud speech-to-text and sync"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Mobile app","body":" free on-device dictation on iPhone and iPad"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile app","body":" free on-device dictation on iPhone and iPad"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile app","body":" free on-device dictation on iPhone and iPad"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Talk to us about Enterprise features for your team."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Having more to say should not mean more thumb-typing.","paragraphs":["OpenWhispr is built for the gap between having something to say and finding a keyboard. The iPhone and iPad app is free to install, with unlimited on-device transcription and optional Pro for Cloud transcription and sync.","We build in public. The desktop code is on GitHub, so you can inspect the local models, Cloud option, and the choices behind the plans before you make one.","Free includes community support."],"image":{"label":"OpenWhispr team and community photo"}},"finalCta":{"tagline":"Voice first · touch to refine","title":"Say it on your phone. Send it as text.","body":"Use the OpenWhispr keyboard in the iPhone app where you are already writing. Speak the long version, refine the text by touch, then send it or continue in a synced OpenWhispr note.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Unlimited on-device transcription is free without an account. US App Store Pro is $9.99 a month or $99.99 a year; website checkout pricing is shown above."}}},"careersHub":{"metadata":{"title":"Careers at OpenWhispr: Help Build the Voice Interface for AI","description":"Join a small team building open-source voice AI. We're hiring a Founding Engineering Lead, a Founding Engineer and a Social & Brand Growth Creator, in person in San Francisco or Chicago.","keywords":["OpenWhispr careers","founding engineer jobs","founding engineering lead jobs","AI startup jobs San Francisco","AI startup jobs Chicago","voice AI jobs"]},"breadcrumb":{"home":"Home","careers":"Careers"},"hero":{"eyebrow":"Careers","title":"Help build the voice interface for AI","subtitle":"OpenWhispr turns what you say into action. It's open source, it's been downloaded about 150,000 times, and it's built for people who want a choice: run it locally, bring your own keys, or use our cloud. We're hiring people who want to own big problems end to end."},"roles":{"title":"Open roles","subtitle":"All roles are in person, in San Francisco or Chicago. Visa sponsorship is not available.","viewRole":"View role","items":{"foundingEngineeringLead":{"title":"Founding Engineering Lead","blurb":"A player-coach role. Lead a small team of engineers, set the technical direction, and stay in the code every day.","location":"San Francisco or Chicago","type":"Full-time","compensation":"$$160K–$200K + meaningful equity"},"foundingEngineer":{"title":"Founding Engineer","blurb":"Own big parts of the product, from first idea to production, and help set the engineering bar.","location":"San Francisco or Chicago","type":"Full-time","compensation":"$$140K–$180K + meaningful equity"},"brandGrowthCreator":{"title":"Social & Brand Growth Creator","blurb":"Be the face, voice and storyteller behind OpenWhispr. Build our brand story in public on TikTok and Reels.","location":"San Francisco or Chicago","type":"Full-time","compensation":"$$60K–$120K + meaningful equity"}}},"principles":{"label":"How we work","title":"Our operating principles","subtitle":"These aren't wall posters. They're how we make decisions every day.","items":[{"title":"Bias for action","body":"If something's broken, fix it. If you have an idea, try it. We don't wait for issue reports or perfect specs."},{"title":"Proactive communication","body":"Tell people before they have to ask. That goes for teammates and for users: if we see a problem coming, we reach out first, and if we get a bill wrong, we refund it before anyone complains."},{"title":"No AI slop","body":"AI gives us leverage, not an excuse. Everything we ship gets reviewed, tested and refined, and anything we share with each other is already clear."},{"title":"Context is the bottleneck","body":"We run a lot of workstreams in parallel. The core skill is managing context: giving it, keeping it, and switching between streams without dropping quality."},{"title":"Do the thinking for the user","body":"Simplification is the strategy. Every screen, sentence and default should take work off the user's plate. Opinionated by design, and still built for the tinkerer."},{"title":"Grateful for every user","body":"People caring enough to complain is a privilege. We use the product every day, and we treat taste as something you train on purpose."}]},"closing":{"title":"Don't see your role?","body":"If you think you'd make us better, write to us anyway and show us something you've built.","ctaLabel":"Email us"}},"careerFoundingEngineer":{"metadata":{"title":"Founding Engineer at OpenWhispr","description":"Own big parts of an open-source voice AI product, from first idea to production. In person in San Francisco or Chicago. $140K to $180K plus meaningful equity. Visa sponsorship not available.","keywords":["founding engineer","OpenWhispr jobs","voice AI engineer","AI-native engineer","open source startup jobs","startup engineering jobs San Francisco","startup engineering jobs Chicago"]},"breadcrumb":{"home":"Home","careers":"Careers","role":"Founding Engineer"},"posting":{"hero":{"eyebrow":"Careers","title":"Founding Engineer","summary":"Own big parts of the product, from first idea to production, and help decide what engineering looks like at OpenWhispr.","meta":{"location":"San Francisco or Chicago","employmentType":"Full-time","workplace":"In person","visa":"No visa sponsorship","compensation":"$$140K–$180K + meaningful equity"}},"intro":["Hi, I'm Gabe, the founder of OpenWhispr.","I think voice is about to become one of the main ways people work with AI. Not talking to a chatbot, but saying what you actually mean and having software turn it into action. For that to work, people need to trust it with what they say. So we build in the open, and we give people a real choice: run everything locally on your own machine, bring your own API keys, or use our cloud when you want the best accuracy and speed. We're not anti-cloud. We just think the choice should be yours.","The other thing we believe is that a great consumer product has to be opinionated in its design, and that we should still build for the tinkerer. You can edit your prompts, pick your models, and customise your own voice pipeline. Holding both of those at once is most of the challenge, and most of the fun.","I built the first version on my own. It's open source, it's been downloaded about 150,000 times, and the people who use it tell us loudly when something breaks. That's a privilege. Most products never get anyone to care that much. Now I'm hiring the engineers who'll own big parts of it with me."],"sections":[{"title":"What you'd work on","paragraphs":["You take large, fuzzy problems and drive them all the way to production. No tickets, and no one writing the spec for you. You'll use the product every day, talk to the people who use it, and fix what you find before anyone files an issue.","The engineering problems are genuinely interesting. On our cloud, it's how we get the best accuracy across every language with the fastest inference. On device, it's how we make local speech-to-text and local language models run well on whatever Mac, Windows or Linux machine someone has. In between, it's how we let people bring their own keys without the product getting complicated. And underneath all of it, it's shaving latency out of the voice pipeline while the experience stays simple and beautiful.","Right now that looks like offline meeting recording and speaker identification, the local model pipeline, speech to action inside whatever app you're in, and the calendar and email integrations."]},{"title":"How we work","paragraphs":["We have a strong bias for action. If something's broken, fix it. If a user is about to hit a problem, tell them before they find it. If we get a bill wrong, refund it before anyone complains. The same goes for each other: share what you know before someone has to ask, and flag a blocker the moment it appears.","We lead with context. You get the full picture, then the ownership and agency to go and deliver it with the team. Everyone runs several AI-assisted workstreams at once, and the skill that matters most is keeping context and quality across all of them. No AI slop. Everything gets reviewed, tested and refined before it ships, and anything you share with the team is already clear by the time it lands.","Getting to 90% is easy now. We spend our energy on the last 10%, and we do the thinking for the user so they don't have to."]},{"title":"You'll do well here if","bullets":[{"body":"You've shipped something real that people actually used: a company, a product, a side project with users. You can tell us what you started, what broke, and what you did about it."},{"body":"You already run parallel AI workstreams and can walk us through exactly how you keep quality high while you do it. Not “I use Cursor.” The actual system."},{"body":"You have taste. You notice when something works but doesn't feel right, and you'd rather simplify than add."},{"body":"You communicate before people have to ask, and you write things down."},{"body":"You've got around five years of real engineering behind you. Less is fine if the evidence is unusual."}]},{"title":"Nice to have","paragraphs":["None of these are required, but they help."],"bullets":[{"body":"You've worked at a VC-backed startup before: a YC company, a Series A or B, a16z Speedrun, that kind of environment. If not, you at least know you want a high-ownership, high-velocity role, and you can say why."},{"body":"You've worked on voice in some form: voice AI, speech-to-text, audio pipelines, real-time systems."},{"body":"Desktop apps, open-source work people actually use, or AI agents, evals and model infrastructure."}]},{"title":"The honest bits","paragraphs":["This is in person in San Francisco or Chicago. Not remote. If you're not in one of those cities you'd need to move, and we can't sponsor visas.","The pace is real and our users complain loudly. If that sounds energising rather than exhausting, keep reading."]},{"title":"First 90 days","paragraphs":["In your first 30 days you'll ship a real product surface end to end. By 90 you'll own a major part of the product, and people will be asking you how it should work."]},{"title":"Comp and time off","paragraphs":["$$140K to $180K plus meaningful equity, and unlimited PTO. We'll make sure you actually take it."]},{"title":"How it goes","paragraphs":["A call with me. Then a deep dive on something you've built. Then a day working together in person on a real problem. We move quickly, and we'll tell you where you stand."]}],"apply":{"title":"Apply","body":"Send a few lines to careers@openwhispr.com about the most impressive thing you've built, ideally something nobody asked you to build, plus links. GitHub, demos and weird side projects beat a polished resume.","ctaLabel":"Apply by email","note":"No cover letter needed."}}},"careerFoundingEngineeringLead":{"metadata":{"title":"Founding Engineering Lead at OpenWhispr","description":"A player-coach role: lead a small team of engineers, set the technical direction, and stay in the code every day. In person in San Francisco or Chicago. $160K to $200K plus meaningful equity. Visa sponsorship not available.","keywords":["founding engineering lead","engineering lead jobs","OpenWhispr jobs","voice AI engineer","hands-on engineering leader","open source startup jobs","startup engineering jobs San Francisco","startup engineering jobs Chicago"]},"breadcrumb":{"home":"Home","careers":"Careers","role":"Founding Engineering Lead"},"posting":{"hero":{"eyebrow":"Careers","title":"Founding Engineering Lead","summary":"A player-coach role. Lead a small team of engineers, set the technical direction, and stay in the code every day.","meta":{"location":"San Francisco or Chicago","employmentType":"Full-time","workplace":"In person","visa":"No visa sponsorship","compensation":"$$160K–$200K + meaningful equity"}},"intro":["Hi, I'm Gabe, the founder of OpenWhispr.","I think voice is about to become one of the main ways people work with AI. Not talking to a chatbot, but saying what you actually mean and having software turn it into action. For that to work, people need to trust it with what they say. So we build in the open, and we give people a real choice: run everything locally on your own machine, bring your own API keys, or use our cloud when you want the best accuracy and speed. We're not anti-cloud. We just think the choice should be yours.","The other thing we believe is that a great consumer product has to be opinionated in its design, and that we should still build for the tinkerer. You can edit your prompts, pick your models, and customise your own voice pipeline. Holding both of those at once is most of the challenge, and most of the fun.","I built the first version on my own. It's open source, it's been downloaded about 150,000 times, and the people who use it tell us loudly when something breaks. That's a privilege. Most products never get anyone to care that much. Now I'm hiring the engineers who'll own big parts of it with me, and I need someone to lead them."],"sections":[{"title":"What you'd work on","paragraphs":["This is a player-coach role. You'll spend most of your time in the code, and you'll lead the engineers around you: setting direction, unblocking, reviewing, raising the bar.","You'll own the architecture and the tradeoffs, keep the system simple as it grows, and define how we build: how we review, how we share context, how we keep AI leverage from turning into slop. You'll help me hire the engineers who come after you, and you'll report to me.","The engineering problems are genuinely interesting. On our cloud, it's how we get the best accuracy across every language with the fastest inference. On device, it's how we make local speech-to-text and local language models run well on whatever Mac, Windows or Linux machine someone has. In between, it's how we let people bring their own keys without the product getting complicated. And underneath all of it, it's shaving latency out of the voice pipeline while the experience stays simple and beautiful.","Right now that looks like offline meeting recording and speaker identification, the local model pipeline, speech to action inside whatever app you're in, and the calendar and email integrations."]},{"title":"How we work","paragraphs":["We have a strong bias for action. If something's broken, fix it. If a user is about to hit a problem, tell them before they find it. If we get a bill wrong, refund it before anyone complains. The same goes for each other: share what you know before someone has to ask, and flag a blocker the moment it appears.","We lead with context. You get the full picture, then the ownership and agency to go and deliver it with the team. Everyone runs several AI-assisted workstreams at once, and the skill that matters most is keeping context and quality across all of them. No AI slop. Everything gets reviewed, tested and refined before it ships, and anything you share with the team is already clear by the time it lands.","Getting to 90% is easy now. We spend our energy on the last 10%, and we do the thinking for the user so they don't have to."]},{"title":"You'll do well here if","bullets":[{"body":"You've led a small team of engineers before, set direction, reviewed their work and grown them, while still shipping meaningful code yourself. You took it on because it multiplied what the team could ship, not because you wanted to stop building."},{"body":"You've shipped something real that people actually used: a company, a product, a side project with users. You can tell us what you started, what broke, and what you did about it."},{"body":"You already run parallel AI workstreams and can walk us through exactly how you keep quality high while you do it. Not “I use Cursor.” The actual system."},{"body":"You have taste. You notice when something works but doesn't feel right, and you'd rather simplify than add."},{"body":"You communicate before people have to ask, and you write things down."},{"body":"You've got around five years of real engineering behind you. Less is fine if the evidence is unusual."}]},{"title":"Nice to have","paragraphs":["None of these are required, but they help."],"bullets":[{"body":"You've worked at a VC-backed startup before: a YC company, a Series A or B, a16z Speedrun, that kind of environment. If not, you at least know you want a high-ownership, high-velocity role, and you can say why."},{"body":"You've worked on voice in some form: voice AI, speech-to-text, audio pipelines, real-time systems."},{"body":"Desktop apps, open-source work people actually use, or AI agents, evals and model infrastructure."}]},{"title":"The honest bits","paragraphs":["This is in person in San Francisco or Chicago. Not remote. If you're not in one of those cities you'd need to move, and we can't sponsor visas.","The pace is real and our users complain loudly. If that sounds energising rather than exhausting, keep reading."]},{"title":"First 90 days","paragraphs":["In your first 30 days you'll ship something real end to end and have a clear view of how we build. By 90 you'll own the technical direction and the practices the team runs on."]},{"title":"Comp and time off","paragraphs":["$$160K to $200K plus meaningful equity, and unlimited PTO. We'll make sure you actually take it."]},{"title":"How it goes","paragraphs":["A call with me. Then a deep dive on something you've built. Then a day working together in person on a real problem. We move quickly, and we'll tell you where you stand."]}],"apply":{"title":"Apply","body":"Send a few lines to careers@openwhispr.com about the most impressive thing you've built, ideally something nobody asked you to build, plus links. GitHub, demos and weird side projects beat a polished resume.","ctaLabel":"Apply by email","note":"No cover letter needed."}}},"careerBrandGrowthCreator":{"metadata":{"title":"Social & Brand Growth Creator at OpenWhispr","description":"Be the face, voice and storyteller behind OpenWhispr on TikTok and Instagram Reels, and build our brand story in public. Full-time, in person in San Francisco or Chicago. $60K to $120K plus meaningful equity.","keywords":["social media creator jobs","brand growth creator","content creator startup job","TikTok creator job San Francisco","Instagram Reels creator job Chicago","OpenWhispr jobs","build in public"]},"breadcrumb":{"home":"Home","careers":"Careers","role":"Social & Brand Growth Creator"},"posting":{"hero":{"eyebrow":"Careers","title":"Social & Brand Growth Creator","summary":"Be the face, voice and storyteller behind OpenWhispr, and build our brand story in public.","meta":{"location":"San Francisco or Chicago","employmentType":"Full-time","workplace":"In person","compensation":"$$60K–$120K + meaningful equity","visa":"No visa sponsorship"}},"intro":["Hi, I'm Gabe, the founder of OpenWhispr.","OpenWhispr turns what you say into action. It's an open-source voice product, it's been downloaded about 150,000 times, and the people who use it care enough to tell us loudly when something breaks. We build in the open, and we give people a real choice about how their voice gets processed. That's a story worth telling, and right now nobody is telling it.","If you've spent any time on Instagram or TikTok this year, you've probably seen what Kinso has done: two founders building their company in public, with the arguments, the prototype reviews, the investor calls and the small daily wins all on camera, and a brand people actually root for. That's the kind of narrative I want us to build. Not ads. A story."],"sections":[{"title":"About the role","paragraphs":["You'll be the face, voice and storyteller behind OpenWhispr on TikTok and Instagram Reels. You'll build our presence the way a brand team would, with strategy, consistency and real rigour about what performs, but framed around a personal story: your day-to-day here, the product being built around you, and your point of view on voice, AI and privacy.","The goal is a loyal, engaged following that drives awareness, affinity and, in the end, demand for OpenWhispr. You'll work directly with me and the team, shooting and editing short-form content every week."]},{"title":"What you'll do","bullets":[{"body":"Build and grow OpenWhispr's presence on TikTok and Instagram Reels, positioned as a real person on the team, not a brand account."},{"body":"Develop an on-camera persona, voice and content universe that feels authentically you, with OpenWhispr woven in naturally."},{"body":"Tell the story of building the company: the product decisions, the launches, the bugs, the users, the office. Some of it scripted, most of it not."},{"body":"Shoot and edit new creative every week, used across organic and paid channels."},{"body":"Study what's working: hooks, like-to-view ratio, retention. Then make the next one better."}]},{"title":"What we're looking for","bullets":[{"body":"You've grown a personal, creator-led or brand account before, with follower and engagement growth on TikTok or Reels that you can show us."},{"body":"You're comfortable and charismatic on camera. You know which moments to spotlight, how to hook someone in the first three seconds, and how to leave them wanting the next post."},{"body":"You have taste. You can make a hook that pulls in a wide audience while keeping the brand elevated."},{"body":"You can edit. CapCut, Premiere, Final Cut, After Effects, whatever you use, you know how to tell a story through the cut."},{"body":"You're TikTok and Instagram native. You spend hours a day noticing trends, hooks, CTAs and brand style."}]},{"title":"Nice to have","paragraphs":["None of these are required, but they help."],"bullets":[{"body":"You already film your own life most days."},{"body":"You've made graphics, animations or illustrations."},{"body":"You're the friend who always has a story and loves to overshare."},{"body":"You've worked at an early-stage startup, or you know you want to."}]},{"title":"The honest bits","paragraphs":["This is in person in San Francisco or Chicago. Not remote. If you're not in one of those cities you'd need to move, and we can't sponsor visas.","We're a small team that moves fast. You'll be building this from zero, and a lot of what you make in the first month will be about finding out what works for us."]},{"title":"Comp and time off","paragraphs":["$$60K to $120K plus meaningful equity, depending on experience, and unlimited PTO. We'll make sure you actually take it."]},{"title":"How it goes","paragraphs":["A call with me. Then a look at what you've made and grown. Then a day with us in person, shooting something real. We move quickly, and we'll tell you where you stand."]}],"apply":{"title":"Apply","body":"Make a video under 60 seconds telling us why you're the person to tell OpenWhispr's story. Sell us on yourself the way you'd sell a product to a viewer. Post it publicly or send us the file, and email the link to careers@openwhispr.com along with the accounts you've grown.","ctaLabel":"Apply by email","note":"Energy and editing over polish."}}},"useCaseLaptop":{"breadcrumb":{"home":"Home","useCases":"Use Cases","laptop":"For Laptop Users"},"metadata":{"title":"Speech to Text App for Laptop - OpenWhispr","description":"OpenWhispr turns laptop speech into text at your active cursor, with Cloud or local transcription and a free local option.","keywords":["speech to text app for laptop","laptop speech to text","voice typing app for laptop","dictation software for laptop","speech to text for Mac and Windows","offline laptop dictation"]},"schema":{"name":"OpenWhispr","applicationCategory":"Laptop speech-to-text app","operatingSystem":"macOS, Windows, Linux","description":"Open-source laptop speech-to-text software for live dictation at the active cursor. Choose Cloud for speed without a local model download, or local for sensitive and offline work.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech to text app for laptop users","title":"Speak at the cursor. Get your words written.","body":"Press one hotkey to dictate into the laptop app you already have open, then review and correct the text where it lands. Choose Cloud for quick setup or local for sensitive and offline work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman sitting on a green sofa with a laptop on her lap."}},"trustBar":{"label":"Write in:","names":["Gmail","Slack","Microsoft Word","Google Docs","Notion","Microsoft Teams"]},"pain":{"label":"What breaks laptop voice typing","title":"The thought is ready. The keyboard is the delay.","subtitle":"You need the text in the field you are using, a familiar way to start, and a quick way to correct it.","cards":[{"title":"The cursor is waiting in another app.","body":"A separate dictation window turns one sentence into a copy-and-paste job. OpenWhispr starts from a global hotkey and automatically pastes text into the app you are already using."},{"title":"Corrections pull you back to the keyboard.","body":"Names, punctuation and a changed sentence can turn voice typing into cleanup. Keep the draft at the cursor, review it there, and use your dictionary for words you repeat."},{"title":"Starting should not take you away from the draft.","body":"Set one global hotkey, then start each dictation from the email, message, note or document field already in front of you."},{"title":"Typing all day adds up.","body":"When your hands need a break or a thought is moving quickly, speak the first pass and spend your effort reviewing the words instead."}]},"features":{"label":"What you get","title":"Voice typing that stays in the work you are doing.","subtitle":"Start at the active cursor. Then choose the transcription path that makes sense for this laptop, this connection and this piece of work.","items":[{"image":{"label":"Woman pressing a laptop key at a wooden desk beside an open notebook."},"title":"Dictate where the cursor already is","body":"Press the global dictation hotkey, speak, and let automatic pasting put the text into the app you have open. This is live laptop writing, not a separate transcription queue."},{"image":{"label":"Woman gesturing beside a laptop and an architectural model at a studio desk."},"title":"Choose Cloud for setup speed, local for offline work","body":"Cloud offers top accuracy and speed without a local model download, then discards transcription audio. Local models run on the device when sensitive or offline work calls for that choice."},{"image":{"label":"Man gesturing toward a laptop at a library desk with books and headphones."},"title":"Keep review and correction in the loop","body":"Read the text where it lands and refine it before you send or save. Voice Assistant can use a dedicated hotkey to edit highlighted text in place, without a wake word."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Direct integrations needed to start","value":"None","footnote":"a global hotkey and automatic pasting return text to the focused cursor"},{"label":"Cost to start voice typing","value":"$$0","footnote":"unlimited local AI models are included on every plan"},{"label":"From download to first dictation","value":"5 min","footnote":"then keep writing at the active cursor"}]},"howItWorks":{"label":"How it works","title":"Text in the field before you reach for the keyboard.","subtitle":"Start free, set up the dictation path you want, then speak into the active text field and review the result before you move on.","image":{"label":"Woman using a laptop beside papers and equipment cases backstage."},"steps":[{"title":"Download for free.","body":"Use OpenWhispr on macOS, Windows or Linux. Open a text field you actually use, then choose the hotkey that starts your dictation."},{"title":"Dictate at the cursor.","body":"Open the email, message, note or document you want to write. Press the hotkey, speak, and let the text paste into that field."},{"title":"Review, correct, continue.","body":"Read the draft where it lands, make the edit you need, and keep working in the same app."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons laptop users choose OpenWhispr","subtitle":"A speech-to-text app has to earn its place in the writing loop. These are the choices that make it useful after the first dictation.","items":[{"title":"Start from one hotkey","body":"Use a global hotkey instead of finding a separate dictation window whenever a sentence is ready."},{"title":"Keep the text in the app you opened","body":"Automatic pasting puts dictated text at the focused cursor, so your email, chat, note or document stays the place you review."},{"title":"Write long thoughts before they disappear","body":"Speak a rough first draft when the thought is moving, then shape the words on screen instead of starting with a blank keyboard."},{"title":"Handle short replies without breaking focus","body":"A quick response still begins at the cursor. Dictate it, read it in place, make the small correction, then send."},{"title":"Teach it the names you use","body":"Add names, jargon and acronyms to the custom dictionary, so recurring words need less cleanup in the text you review."},{"title":"Use Cloud when you want to start quickly","body":"Cloud transcription gives you top accuracy and speed without a local model download."},{"title":"Keep local work on the laptop","body":"Local transcription runs entirely on your device with no internet required, for an offline day or work you want to process locally."},{"title":"Know what happens to Cloud audio","body":"Cloud audio is sent only to transcribe it, then immediately discarded. It is a speed choice, not a storage choice."},{"title":"Make a selected edit without retyping","body":"With Voice Assistant's dedicated hotkey, highlighted text can be edited in place. It does not turn your laptop into a hands-free operating system."},{"title":"Start free, then pay for more of the workflow","body":"Free includes unlimited local models and a Cloud allowance. Pro adds unlimited OpenWhispr Cloud transcription, sync, mobile and API/MCP access."}]},"faq":{"label":"Before you download","title":"The questions laptop users ask first.","subtitle":"Straight answers on where text lands, how you correct it, and which transcription path fits the work in front of you.","items":[{"question":"Will it type into the app I am using?","answer":"OpenWhispr uses a global dictation hotkey and automatic pasting, so the text can land at your focused cursor in the laptop app you already have open. That is text insertion, not a direct integration with each app."},{"question":"Which laptop platforms can I use?","answer":"OpenWhispr is available for macOS, Windows and Linux. Start with the app and test it in the text fields where you actually write."},{"question":"What if I need to correct a name or sentence?","answer":"Review the text where it lands. Add names, jargon and acronyms to the custom dictionary for recurring terms. With Voice Assistant's dedicated hotkey, highlighted text can be edited in place."},{"question":"Should I choose Cloud or local transcription?","answer":"Choose Cloud for top accuracy and speed without a local model download. Audio is sent only to transcribe, then immediately discarded. Choose local when sensitive or offline work calls for device-only processing."},{"question":"Is this for live voice typing or recorded audio?","answer":"This page is for live dictation into an active laptop text field. OpenWhispr also supports meeting transcription and importing audio or video, but those are separate workflows."},{"question":"What is free, and what changes on a paid plan?","answer":"Free includes unlimited local AI models, 2,000 Cloud words a week and five hours of meeting recordings a month. Pro is $8 a month or $80 a year for unlimited Cloud transcription, 20 meeting-recording hours, sync, mobile and API/MCP access. Business is $16 a month or $160 a year and adds unlimited meeting recordings, chat over your data and priority support."}]},"reviews":{"label":"What users say","title":"Built for people who would rather speak than type.","subtitle":"Three users describing the practical shift from keyboard-first work to voice input.","score":{"value":"1","label":"Global hotkey to start dictation","body":"Press the hotkey, speak, and let automatic pasting return the text to the focused cursor."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for laptop voice typing","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Laptop users who want unlimited local models and a limited Cloud allowance"}]},{"parts":[{"text":"Laptop users who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"People who also need unlimited meeting recordings, chat over data, and priority support"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local AI models, plus 2,000 OpenWhispr Cloud words a week"}]},{"parts":[{"text":"Unlimited Cloud transcription, 20 meeting-recording hours, sync, mobile, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings, chat over your data, and priority support"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"A laptop dictation app you can inspect.","paragraphs":["OpenWhispr is an open-source desktop app for turning voice into text, notes and actions. It runs on macOS, Windows and Linux.","The writing loop is simple: press a hotkey, speak, see the text at the cursor, then review it in the app where the work already lives.","This page is for live laptop dictation at the active cursor. Meeting recordings and imported audio are separate workflows."],"image":{"label":"OpenWhispr desktop app on a laptop beside a document"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Leave the keyboard for the words that are easier to say.","body":"Download OpenWhispr, put the cursor where you want to write, and start the next draft with your voice.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. Free forever on macOS, Windows and Linux."}}},"useCaseMacbook":{"breadcrumb":{"home":"Home","useCases":"Use Cases","macbook":"For MacBook"},"metadata":{"title":"Dictation App For MacBook, No 30-Second Cutoff - OpenWhispr","description":"OpenWhispr is free, open-source dictation software for MacBook. Hold the hotkey, talk through the pause, and it keeps listening where macOS stops.","keywords":["dictation app for macbook","macbook dictation software","voice to text macbook","macos dictation alternative","speech to text app for mac","dictate on macbook offline","macbook voice typing app"]},"schema":{"name":"OpenWhispr","applicationCategory":"MacBook dictation software","operatingSystem":"macOS, Windows, Linux","description":"Open-source dictation software for MacBook. Local transcription runs on your Apple Silicon chip for offline and battery work, or choose Cloud for top accuracy and speed.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Dictation app for MacBook","title":"Hold the Globe key. Get the whole thought down.","body":"Talk through any pause, nothing to fix afterward. It transcribes on your MacBook's own chip, so it keeps working on battery with no signal.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Person sitting on a floor cushion and using a laptop on a wooden stool"}},"trustBar":{"label":"Dictate into:","names":["Mail","Messages","Notes","Slack","Word","Google Docs"]},"pain":{"label":"What breaks macOS dictation on a MacBook","title":"You paused to think. It stopped listening.","subtitle":"Away from a desk, on battery, on the mic already built into the lid, the built-in tool keeps cutting the thought off before it lands.","cards":[{"title":"It stops after 30 seconds of silence.","body":"macOS Dictation ends automatically when it hears no speech for 30 seconds, Apple's own documented limit. Every pause to think is a chance it quits on you mid-sentence. OpenWhispr's hotkey stays recording for as long as you hold or toggle it. No silence timer to outrun."},{"title":"Names come back mangled.","body":"One MacBook owner who dictates because typing hurts their hands found macOS's own Dictation turning a name they were writing about into a different name entirely, and called the result unusable. OpenWhispr's custom dictionary learns the names, clients and jargon you actually say, so they come out right."},{"title":"Punctuation is a spoken chore.","body":"Say the word 'comma' out loud or it doesn't appear, outside a handful of supported languages. OpenWhispr transcribes full sentences, not commands, so you talk like you talk and the punctuation is already there."},{"title":"A plane, a train, a café with no signal.","body":"A MacBook runs all day on battery, on a built-in mic, in rooms you don't control: a flight, a shared desk, a hotel room with no wifi. Local transcription runs on your MacBook's own Apple Silicon chip, so none of that stops you dictating."}]},"features":{"label":"What you get","title":"Dictation that keeps up with a laptop that moves.","subtitle":"Run it fully offline on your MacBook's own chip, or switch to Cloud when you're at a desk and want the fastest, most consistent transcription.","items":[{"image":{"label":"Person speaking toward a laptop beside an external monitor at a home desk"},"title":"Local transcription that needs no signal","body":"Whisper runs GPU-accelerated on your MacBook's own Apple Silicon chip, with Parakeet alongside it, both entirely on-device. Dictate on a flight, in the car, anywhere the wifi icon has a line through it."},{"image":{"label":"Person speaking beside an open laptop at a shared café table"},"title":"Text that's already finished","body":"No spoken punctuation commands, and nothing for you to clean up afterward. A custom dictionary primes Whisper and OpenWhispr Cloud with the names and jargon you actually use, so they land right the first time."},{"image":{"label":"Person speaking with a finger on a laptop keyboard at a workshop table"},"title":"One hotkey, talk as long as you need","body":"Hold or tap the same Globe key macOS already uses. Nothing here cuts you off after 30 seconds, so a long pause to think doesn't cost you the sentence."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Words lost to a silence cutoff","value":"None","footnote":"hold the hotkey and keep talking, there's no timer"},{"label":"Cost to start","value":"$$0","footnote":"unlimited local dictation on your MacBook, forever"},{"label":"Setup takes","value":"5 min","footnote":"download, grant Accessibility access, start dictating"}]},"howItWorks":{"label":"How it works","title":"From a held key to finished text, before you've moved your hands.","subtitle":"Download, choose Cloud for the quickest setup or local for offline and battery work, then dictate straight into whatever's already open.","image":{"label":"Person speaking while using a laptop at a round table in an office lounge"},"steps":[{"title":"Download for free.","body":"macOS on Apple Silicon or Intel, plus Windows and Linux. Choose Cloud for no model download or local for offline and battery work."},{"title":"Hold the Globe key and talk.","body":"Speak through as long a pause as you need. Nothing stops you at 30 seconds."},{"title":"It's already in the field.","body":"Finished, punctuated text lands where your cursor was. Review it and send."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons MacBook users choose OpenWhispr","subtitle":"Each one is a fact about what the app does on your machine, not an adjective.","items":[{"title":"Never hits a 30-second wall","body":"Hold or tap the hotkey and OpenWhispr keeps recording until you stop it. No silence timer decides when your thought is over."},{"title":"Works offline, on battery","body":"Local transcription runs GPU-accelerated on your MacBook's own Apple Silicon chip. No signal required, so a flight or a dead wifi router doesn't stop you."},{"title":"Free, unlimited, forever","body":"Local dictation costs $0 on every plan, with no word cap and no trial clock. Upgrade if you want unlimited Cloud, Agent mode, sync or mobile."},{"title":"Learns the names spelling gets wrong","body":"A custom dictionary you edit yourself teaches OpenWhispr the client names, product names and jargon your MacBook's autocorrect never learned."},{"title":"Drops text into whatever's already open","body":"One system-wide hotkey pastes into Mail, Messages, Notes, Slack, Word or any other text field, once you grant the Accessibility permission macOS itself asks for."},{"title":"Cloud audio is discarded the moment it's transcribed","body":"Choose Cloud for speed and OpenWhispr sends audio only to transcribe it, then never stores it. Nothing sits on a server afterward."},{"title":"Open source, not a black box","body":"Every line is on GitHub under MIT. If you want to know exactly what happens to your voice, you can read the code instead of a privacy policy."},{"title":"Rewrite the last sentence without touching the keyboard","body":"A separate Voice Assistant hotkey edits your highlighted text in place, no wake word, no cleanup pass. Say what to change, and the selection updates."},{"title":"100+ languages, not just English","body":"macOS auto-inserts punctuation in only a handful of languages, and a browser's voice typing works in English alone. OpenWhispr dictates 100+ languages without either ceiling."},{"title":"Full sentences, not voice commands","body":"No command vocabulary to learn, no 'new paragraph' or 'exclamation point' to remember. Say it the way you'd say it out loud, and the punctuation and capitalization arrive on their own."}]},"faq":{"label":"Before you download","title":"The questions MacBook users ask first.","subtitle":"Straight answers on privacy, cost and whether this actually beats what's already built in. If yours isn't here, ask us in Discord. We reply within 24 hours.","items":[{"question":"Does the audio ever leave my MacBook?","answer":"With a local model, no. Recording and transcription happen on your own Apple Silicon chip. With OpenWhispr Cloud, audio is sent only to transcribe it, then immediately discarded. Pick whichever mode fits where you're working."},{"question":"Will it cut me off mid-sentence like macOS Dictation does?","answer":"No. macOS's own Dictation stops automatically after 30 seconds of silence. OpenWhispr's hotkey keeps recording for as long as you hold or toggle it, so a long pause to think doesn't end the session."},{"question":"Will it actually get my names and jargon right?","answer":"Better than a generic model, because you can teach it. The custom dictionary lets you add the client names, product names and technical terms you use, so OpenWhispr stops guessing at them."},{"question":"Does it work on a flight or somewhere with no wifi?","answer":"Yes, in local mode. Transcription runs entirely on your MacBook's own chip with no network call, so it works exactly the same at 35,000 feet as it does at a desk. Choose local before you lose signal, not after."},{"question":"Is it actually free, or free with a catch?","answer":"Free gives you unlimited local dictation plus a limited Cloud allowance, no card required. Upgrade to Pro for unlimited Cloud transcription, Agent mode, sync across devices, the iPhone and iPad app, and API/MCP access. Business adds unlimited meeting recordings, chat over your data, and priority support."},{"question":"Does it just transcribe, or can it clean up what I said?","answer":"Both. Dictation gets your words down as finished, punctuated text. A separate Voice Assistant hotkey can also edit text you've already written, in place, using a local or cloud AI model, with no wake word needed."}]},"reviews":{"label":"What users say","title":"Built by people who don't want to be recorded either.","subtitle":"From our Discord, GitHub and inboxes, the same instinct that matters on a laptop you carry everywhere.","score":{"value":"0","label":"Voice recordings kept on a server","body":"In local mode the audio never leaves your MacBook. Nothing is stored on a server, and nothing is sent anywhere to be reviewed."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for MacBook users","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"MacBook users who want a useful free start with local dictation and limited Cloud transcription"}]},{"parts":[{"text":"MacBook users who want unlimited Cloud transcription plus sync, the iPhone app, and API/MCP access"}]},{"parts":[{"text":"Anyone dictating at volume across a team, with unlimited meeting recordings and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Useful Free access, including local dictation and limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus Agent mode, sync, mobile, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus chat over your data and priority support"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"We built it for a laptop that leaves the desk.","paragraphs":["OpenWhispr started with a simple objection: a thought you already have in your head, mid-sentence on a laptop away from your desk, shouldn't get cut off by a 30-second silence timer.","We build in public. Every release and every line of code is on GitHub, where 5,000+ engineers have starred, forked and audited the project. When we say the dictation stays local, that's not a promise in a privacy policy. It's a property of the code, and you can verify it yourself.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Your next sentence doesn't stop at 30 seconds.","body":"Download OpenWhispr and dictate straight into whatever's open on your MacBook. On battery, offline, on your own machine.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCasePc":{"breadcrumb":{"home":"Home","useCases":"Use Cases","pc":"For PC users"},"metadata":{"title":"Speech to Text Software for PC - OpenWhispr","description":"OpenWhispr is speech to text software for Windows PCs. Hold a hotkey to dictate into the app you use, with Cloud or local processing.","keywords":["speech to text software for PC","speech to text software for Windows","PC dictation software","voice typing software for PC","Windows speech to text app","dictation software for PC"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text software for PC","operatingSystem":"Windows","description":"Open-source speech-to-text software for Windows PCs. Use a hotkey to dictate into the app you are using, with Cloud transcription or local processing.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech to text software for PC","title":"Hold the hotkey. Get your words in the box.","body":"Put the cursor in an email, document, note, or browser field. With auto-paste on, OpenWhispr puts the transcription there. Choose Cloud for fast transcription without a local download, or local models for offline and sensitive work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Man speaking and gesturing toward a desktop monitor at a wooden desk"}},"trustBar":{"label":"Dictate into:","names":["Microsoft Word","Outlook","OneNote","Google Docs","PowerPoint","Notepad"]},"pain":{"label":"When voice typing stops being enough","title":"The words are ready. The typing is the slow part.","subtitle":"A PC writing tool should get spoken words into the place you already work, without making every sentence a new workflow.","cards":[{"title":"The cursor is waiting. You are still typing.","body":"Put the cursor in the reply, document, note, or browser field first. With auto-paste enabled, OpenWhispr returns the transcription there after you speak."},{"title":"Your voice tool lives in only one app.","body":"OpenWhispr uses a global hotkey for dictation instead of an add-in for one editor. Use it where you write, then review the text in that same place."},{"title":"A name or acronym comes back wrong.","body":"Add the spellings you use to the custom dictionary. It gives transcription and cleanup a better guide for names, jargon, and abbreviations."},{"title":"The network is not always part of the job.","body":"Choose a local transcription model when you need to work offline or keep processing on the PC. Choose Cloud when speed and no model download matter more."}]},"features":{"label":"Speech-to-text that fits the PC","title":"Speak where the work already is.","subtitle":"Review the words in the field before you send or save.","items":[{"image":{"label":"Woman gesturing toward a desktop monitor with a wall calendar behind her"},"title":"Use a hotkey at the current cursor","body":"Open the app you need, put the cursor in the text field, and use the global dictation hotkey. With auto-paste on, the transcription goes back to that field."},{"image":{"label":"Woman holding a small microphone beside a desktop monitor at a home office desk"},"title":"Cloud when you want to skip the model download","body":"OpenWhispr Cloud is the fast, high-accuracy path with no local model to download. It sends audio only to transcribe it, then immediately discards that audio."},{"image":{"label":"Man holding handwritten notes and a pencil while looking at a desktop monitor"},"title":"Review text where it will be used","body":"With auto-paste enabled, the transcription returns to the field where it will be used. Check names, wording, and punctuation before you send or save."}]},"stats":{"ariaLabel":"Key facts for PC speech-to-text","items":[{"label":"Extra app between your voice and the text field","value":"None","footnote":"with auto-paste enabled, dictation returns to the cursor in the app you are using"},{"label":"Unlimited local dictation","value":"$$0","footnote":"local transcription models are included on every plan, including Free"},{"label":"Free Cloud transcription","value":"2,000/week","footnote":"words included each week, separate from unlimited local transcription"}]},"howItWorks":{"label":"From download to first draft","title":"Three steps from a hotkey to a finished field.","subtitle":"Choose Cloud for no-download transcription or install a local model for on-device, offline work.","image":{"label":"Man stretching his shoulder while seated at a desk beside a desktop monitor"},"steps":[{"title":"Download it free.","body":"Install OpenWhispr on your Windows PC. Start with OpenWhispr Cloud or download a local transcription model."},{"title":"Put the cursor where you write.","body":"Open an email, document, note, presentation, or browser field. Use the global hotkey and say what you want to write."},{"title":"Review the text, then finish the job.","body":"With auto-paste enabled, the transcription returns to that field. Check names and wording, then send, save, or keep writing."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Why PC users choose it","title":"10 reasons PC users choose OpenWhispr","subtitle":"It is not another document editor. It is speech-to-text that starts with the cursor you already have and gives you a real choice about how the audio is processed.","items":[{"title":"Dictate into the field you opened","body":"Use the global hotkey from the email, document, note, or browser field you are already working in. With auto-paste enabled, the transcription returns to that cursor."},{"title":"Keep auto-paste under your control","body":"Auto-paste and keeping a transcription in the clipboard are separate Settings choices. Turn auto-paste off when you want to decide where text goes, and enable clipboard retention when you want to keep a copy there."},{"title":"Start with useful Free access","body":"Free includes unlimited local AI models and 2,000 OpenWhispr Cloud words each week. Test real PC writing work without a card or a local word cap."},{"title":"Skip the local download when speed matters","body":"OpenWhispr Cloud gives you the high-accuracy, fast path without downloading a local transcription model. The audio is sent only to transcribe it, then immediately discarded."},{"title":"Keep offline work on the PC","body":"Choose a local transcription model for work without an internet connection or work that needs device-only processing. Local models are included on every plan."},{"title":"Give names and jargon a better chance","body":"Add client names, product names, acronyms, and specialist terms to the custom dictionary. OpenWhispr uses those spellings to guide transcription and cleanup."},{"title":"Do not confuse a transcript with finished writing","body":"Speech-to-text gets the first draft into the correct field. You still see the text where it will be used, so checking it is part of the same writing pass."},{"title":"Use the Windows path without a permission detour","body":"Windows auto-paste does not need the accessibility permission step required on macOS. You can decide whether auto-paste is on in Settings."},{"title":"Work in more than one language","body":"OpenWhispr supports 100+ languages on every plan. Keep the writing destination on your PC while the spoken language changes with the work."},{"title":"Upgrade when your Cloud writing grows","body":"Pro adds unlimited OpenWhispr Cloud transcription, Agent mode, sync, mobile, and API or MCP access. Business keeps those PC writing features and adds priority support."}]},"faq":{"label":"Before you install it","title":"The questions PC users ask before they install speech-to-text software.","subtitle":"Straight answers on Windows, app coverage, privacy, correction, offline work, and paid plans.","items":[{"question":"Does OpenWhispr work on a Windows PC?","answer":"Yes. OpenWhispr is available for Windows, macOS, and Linux. The Free plan is available on desktop, including Windows."},{"question":"Can I dictate into the apps I already use?","answer":"Open the app and put the cursor in the text field you want to fill. OpenWhispr uses a global hotkey. With auto-paste enabled, it puts the transcription at that cursor. It is not an add-in for Word, Outlook, Google Docs, or any other editor."},{"question":"Should I choose Cloud or local speech-to-text?","answer":"Choose OpenWhispr Cloud when you want fast, high-accuracy transcription with no local model download. Choose a local model for offline or device-only processing. Cloud sends audio only to transcribe it, then immediately discards it."},{"question":"What if it gets a name or acronym wrong?","answer":"Add the spelling to the custom dictionary. It guides transcription and cleanup for names, jargon, and acronyms. Review the text before you send or save it, especially where wording matters."},{"question":"Can I use it without an internet connection?","answer":"Yes, after you download a local transcription model. Local models run on your device without internet and are unlimited on every plan, including Free."},{"question":"What do I get by upgrading from Free?","answer":"Free includes unlimited local models plus 2,000 OpenWhispr Cloud words each week. Pro adds unlimited Cloud transcription, Agent mode, 20 hours of meeting recordings each month, sync, mobile, and API or MCP access. Business adds unlimited meeting recordings and chat over your data."}]},"reviews":{"label":"What people tell us","title":"Better than typing. Still your words to review.","subtitle":"From our Discord, GitHub and inboxes. Read these for the PC dictation workflow, not as a promise that every spoken word needs no correction.","score":{"value":"At the cursor","label":"With auto-paste enabled","body":"Start in the text field that needs the words. With auto-paste on, the transcription returns there for review instead of living in a separate editor."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free to start.","Upgrade when the workflow grows."],"caption":"Compare OpenWhispr plans for PC speech-to-text","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"PC users who want unlimited local dictation and 2,000 Cloud words each week"}]},{"parts":[{"text":"PC users who want unlimited Cloud transcription plus Agent mode, sync, mobile, and API or MCP access"}]},{"parts":[{"text":"Teams that also need unlimited meeting recordings and chat over their data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local dictation plus 2,000 OpenWhispr Cloud words each week"}]},{"parts":[{"text":"Unlimited OpenWhispr Cloud transcription plus Agent mode and connected workflow features"}]},{"parts":[{"text":"Everything in Pro plus unlimited meeting recordings and chat over your data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Choose local models, or use the Free Cloud allowance. You can also bring your own API keys."}]},{"parts":[{"text":"Use unlimited OpenWhispr Cloud transcription without a local model download."}]},{"parts":[{"text":"Everything in Pro, plus unlimited meeting recordings and chat over your data."}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" guides names and terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" guides names and terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Agent mode","body":" use AI actions on transcripts"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools and workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" guides names and terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Agent mode","body":" use AI actions on transcripts"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools and workflows"},{"title":"Chat over your data","body":" ask about files and transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Enterprise adds SSO, audit logs, retention controls, and dedicated support."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"Your PC already has the place to write. We built for getting words there.","paragraphs":["OpenWhispr started with a simple idea: people writing all day should not have to choose between typing every first draft and moving speech through a separate app.","The app uses a hotkey and optional auto-paste so you can dictate into the PC apps you already use, then decide what needs a correction.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"One of OpenWhispr’s founders"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Open the app. Put the cursor where you write.","body":"Download OpenWhispr on your Windows PC, choose Cloud or a local model, and dictate the next email, document, note, or browser field without changing where the work lives.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free to download on Windows, macOS and Linux."}}},"useCaseWindows11":{"breadcrumb":{"home":"Home","useCases":"Use Cases","windows11":"For Windows 11"},"metadata":{"title":"Free Speech-to-Text Software for Windows 11 - OpenWhispr","description":"OpenWhispr is free, open-source speech-to-text for Windows 11 that fixes punctuation automatically and works offline, no Copilot+ PC required.","keywords":["free speech to text software for windows 11","windows 11 dictation software","windows 11 voice typing alternative","offline dictation software windows","voice access alternative","speech to text without internet windows 11","free dictation software windows"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text software for Windows 11","operatingSystem":"macOS, Windows, Linux, iOS","description":"Open-source, free speech-to-text software for Windows 11. Local transcription that runs with no internet connection and no Copilot+ hardware required.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Free speech-to-text software for Windows 11","title":"Speak into any app. Get it right, offline.","body":"A local model runs on the Windows 11 PC you already own, no internet or Copilot+ hardware required. Teach it your names and terms once, and stop correcting the same word twice.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman speaking and gesturing toward a laptop at a table with bookshelves behind her."}},"trustBar":{"label":"Dictate into any app on Windows 11, including:","names":["Microsoft Word","Outlook","Gmail","Slack","Google Docs","Chrome"]},"pain":{"label":"What Windows 11 already gave you, and where it stops","title":"Win+H got you this far. It won't get you the rest of the way.","subtitle":"You already tried the built-in tools. Here's exactly where each one runs out.","cards":[{"title":"Say \"question mark\" and get the words \"question mark\", typed out.","body":"Voice Access takes spoken punctuation literally instead of inserting the symbol, and a thinking pause can drop a stray period mid-sentence. OpenWhispr's cleanup converts spoken punctuation to the actual mark and treats a pause as a pause, not a full stop."},{"title":"You corrected the same name a hundred times. It's still wrong.","body":"Voice Access's custom vocabulary only improves the correction list, not what gets typed first. With OpenWhispr Cloud or the Whisper local model, your custom dictionary is sent to transcription itself, so your names, jargon, and terms come out right on the first pass."},{"title":"The good tier needs a new PC. The free tier needs Wi-Fi.","body":"Windows' voice typing runs on Microsoft's cloud and needs a connection, and the auto-punctuating fluid dictation tier is locked to Copilot+ hardware. OpenWhispr's local model runs entirely on the Windows 11 PC you already have, online or off."},{"title":"Win+H for this, Voice Access for that, Word's Dictate for the rest.","body":"Microsoft's own support answer is to switch tools depending on what you're doing. OpenWhispr is one hotkey that works the same way in Word, email, chat, or any app with a cursor."}]},"features":{"label":"From your voice to typed text","title":"Built for the moment Win+H gives up.","subtitle":"Choose the local model for offline, private dictation, or Cloud for the fastest transcription.","items":[{"image":{"label":"Man with glasses speaking toward a desktop monitor at a desk beside a bright window."},"title":"A local model, no internet or new PC required","body":"Download a local model once and OpenWhispr transcribes entirely on your PC, the one you already own. No round-trip to a server, no Copilot+ requirement, no signal needed for transcription."},{"image":{"label":"Person in a green sweater speaking toward a desktop monitor in a kitchen."},"title":"A dictionary that changes what gets typed, not just the correction list","body":"Add your name, your field's jargon, and the proper nouns Windows keeps missing. With OpenWhispr Cloud or the Whisper local model, that list is passed to transcription itself, so the right word shows up the first time."},{"image":{"label":"Man gesturing toward a desktop monitor with shelves of records behind him."},"title":"One hotkey, every app","body":"Word, Outlook, Slack, a web form: press the hotkey and clean, punctuated text pastes in wherever your cursor already is. Nothing to install per app, no add-in, no plugin to configure."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Internet needed to dictate","value":"None","footnote":"the local model transcribes entirely on your Windows 11 PC"},{"label":"Cost to dictate offline","value":"$$0","footnote":"unlimited local transcription in the Free plan"},{"label":"Punctuation cleanup","value":"Automatic","footnote":"converts spoken \"comma\" and \"question mark\" into the mark, by default"}]},"howItWorks":{"label":"Download to first sentence","title":"Three steps from download to typed text.","subtitle":"Choose Cloud transcription, or download a local model that needs no internet.","image":{"label":"Man speaking toward a desktop monitor at a wooden desk in a brick-walled room."},"steps":[{"title":"Download it free.","body":"Setup offers OpenWhispr Cloud or a local model. Pick local and it downloads once, then works with no internet."},{"title":"Add your names and terms.","body":"Put the proper nouns and jargon Windows keeps missing into the custom dictionary, once."},{"title":"Press the hotkey and speak.","body":"Clean, punctuated text pastes into whatever you're typing into: Word, email, chat, any app. Want the cleanup pass offline too? Set a local model for that in Settings; transcription already is."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Why Windows 11 users switch","title":"10 reasons Windows 11 users choose OpenWhispr over the built-in tools","subtitle":"Every one of these is a specific gap in Win+H or Voice Access, checkable in the app itself.","items":[{"title":"Works with the Wi-Fi off","body":"A local model needs no connection to transcribe, unlike Windows' voice typing, which runs on Microsoft's cloud and stops working the moment your connection does."},{"title":"No Copilot+ PC required","body":"Windows' best free auto-punctuation tier is locked to Copilot+ hardware. OpenWhispr's local dictation runs on the ordinary Windows 11 PC you already own, no hardware upgrade required."},{"title":"One hotkey, every app, not just Word","body":"Word's Dictate only works in Word, and Win+H behaves differently app to app. OpenWhispr's hotkey works the same way everywhere, so there's one habit to build instead of three."},{"title":"A dictionary that changes the first pass","body":"Voice Access's custom vocabulary only reorders the correction suggestions. With OpenWhispr Cloud or the Whisper local model, your custom dictionary is sent to transcription itself, so a term you've taught it shows up correctly the first time, not the fifth."},{"title":"Punctuation from how you speak, not a command you say right","body":"Say \"question mark\" in Windows and you may get the literal words typed out. OpenWhispr's cleanup converts spoken punctuation into the mark and treats a thinking pause as a pause, not a period."},{"title":"Self-corrections dropped automatically","body":"Say \"send it by Thursday, no wait, Friday\" and OpenWhispr keeps only \"Send it by Friday.\" Windows' tools transcribe the retry verbatim and leave you to fix it by hand. Need to fix something after it's already typed? Agent mode (Pro) edits it in place, no re-highlighting."},{"title":"Free for the whole machine, no per-minute meter","body":"Unlimited local dictation at $0, plus 2,000 Cloud words a week. No trial clock, no per-minute charge, no forced upgrade to keep dictating."},{"title":"Cloud audio is transcribed, then gone","body":"Choose OpenWhispr Cloud for the highest accuracy and speed, with no local model to download. Audio is sent only to transcribe it, then immediately discarded, never stored."},{"title":"Open source, so you can check it yourself","body":"How OpenWhispr's local dictation works isn't a promise, it's code. Read exactly what happens to your audio on GitHub before you trust it with anything."},{"title":"More languages than Voice Access supports","body":"Voice Access covers 15 languages and dialects, and Windows' best free auto-punctuation only works in English. OpenWhispr supports 100+ languages on every plan."}]},"faq":{"label":"Before you install it","title":"Six questions Windows 11 users ask before switching from Win+H.","subtitle":"Straight answers on internet, hardware, accuracy, and cost. More questions? Ask in Discord.","items":[{"question":"How is this different from Windows' own Win+H voice typing?","answer":"Win+H runs on Microsoft's cloud speech service and needs an internet connection, and the auto-punctuating fluid dictation tier is exclusive to Copilot+ PCs. OpenWhispr's local model transcribes entirely on the Windows 11 PC you already have, works offline, and adds a custom dictionary for your own vocabulary."},{"question":"Do I need a Copilot+ PC?","answer":"No. OpenWhispr's local dictation is unlimited and free on an ordinary Windows 11 PC. Copilot+ is Microsoft's own hardware requirement for their fluid dictation tier, not something OpenWhispr needs."},{"question":"Will it get my name and technical terms right?","answer":"Add them to the custom dictionary once. With OpenWhispr Cloud or the Whisper local model, that list is passed to transcription itself, so the right word shows up on the first pass instead of only appearing in a correction menu."},{"question":"Does it work without an internet connection?","answer":"Yes, for transcription: download a local model once during setup and speech-to-text runs entirely on your device afterward, no connection required. The punctuation cleanup pass uses a quick cloud call by default; set a local model for that too in Settings if you want the whole pipeline offline."},{"question":"Is it actually free, or a trial?","answer":"Free includes unlimited local dictation and local AI models, plus 2,000 Cloud words a week, with no time limit. Pro adds unlimited Cloud transcription, Agent mode, 20 hours of meeting recordings, sync, mobile, and API/MCP access."},{"question":"Does it work in every app, or just Word?","answer":"Every app. One system-wide hotkey pastes clean text wherever your cursor already is: Word, Outlook, a browser, Slack, a chat box. There's no add-in to install into any of them."}]},"reviews":{"label":"What people tell us","title":"Built by people who wanted their own machine to do this.","subtitle":"From our Discord, GitHub and inboxes. They're not Windows 11 testimonials, but the no-WiFi, local-model story below will sound familiar.","score":{"value":"0","label":"Audio stored on our servers with local transcription","body":"With a local transcription model the audio never leaves your PC, so there's nothing on our side to lose, breach or hand over."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free to start.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for Windows 11","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Windows 11 users who want offline local dictation and 2,000 Cloud words a week"}]},{"parts":[{"text":"Windows 11 users who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Individuals and teams documenting at volume with unlimited meeting recordings and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited offline local dictation plus 2,000 Cloud words a week"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings plus chat over data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Enterprise adds SSO, audit logs, retention controls, and dedicated support."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"The built-in tool should have just worked.","paragraphs":["Free dictation software shouldn't need an internet connection or new hardware to be any good. That's the position OpenWhispr builds from.","We build in public. The app is open source, so you can inspect how it works before you install it, instead of taking our word for it.","Community support is available in Discord. Paid plans add email or priority support."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"$$0 · open source · punctuation fixed automatically","title":"Stop fighting Win+H. Start dictating.","body":"Download OpenWhispr, add your names and terms to the custom dictionary, and dictate into whatever you're already typing into.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free on Windows, macOS, and Linux. Paid plans add Cloud transcription, mobile, and workflow tools."}}},"useCaseWindows10":{"breadcrumb":{"home":"Home","useCases":"Use Cases","windows10":"For Windows 10"},"metadata":{"title":"Speech to Text Software for Windows 10 - OpenWhispr","description":"OpenWhispr is free, open-source speech to text software built for Windows 10, no Windows 11 or internet connection required.","keywords":["speech to text software for windows 10","windows 10 dictation software","voice to text windows 10","offline speech to text windows 10","free dictation software windows 10","dictation software no windows 11"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech to text software for Windows 10","operatingSystem":"macOS, Windows, Linux","description":"Open-source speech to text software for Windows 10 or later. Choose Cloud for speed with no model download, or local processing that works with no internet at all.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech to text software for Windows 10 PCs","title":"Press the hotkey. Get the reply, ready to send.","body":"Installs on Windows 10 or later, no Windows 11 required. Local models work fully offline, once downloaded.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman speaking and gesturing at a desktop computer beside plants and large windows."}},"trustBar":{"label":"Dictate into:","names":["Microsoft Word","Microsoft Outlook","Microsoft Edge","Microsoft Teams","Notepad"]},"pain":{"label":"What stalls dictation on Windows 10","title":"Windows 10 dictation stalled. Yours doesn't have to.","subtitle":"You need something that installs on the OS you actually have, keeps working without an internet connection, and gets your words right the first time.","cards":[{"title":"Win+H needs the internet, every time.","body":"Windows 10's built-in dictation requires a live connection to work at all, and it stops responding for some people after a Windows update, with no fix in sight. OpenWhispr's local models run entirely on your device, so a dropped connection never stops you mid-sentence."},{"title":"Voice access never came to your PC.","body":"Microsoft's newer voice tools are gated to Windows 11 version 22H2 and later; Windows 10 keeps the deprecated Windows Speech Recognition, and it isn't getting better. OpenWhispr installs on Windows 10 or later. No upgrade required."},{"title":"It typed \"background noise\" into an email.","body":"Generic recognition mishears background sound as words and turns punctuation into a spoken chore. OpenWhispr's custom dictionary learns your names and terms, and local speech models built for accuracy do the listening."},{"title":"The professional option you remember is gone.","body":"The consumer edition of Dragon was discontinued in 2023, with no replacement at that price. OpenWhispr is free, with unlimited local dictation for as long as you use it. No card required."}]},"features":{"label":"What you get","title":"Dictation that doesn't care which Windows you're on.","subtitle":"Cloud is the easy choice for an older CPU; local is the choice when you want zero reliance on a connection. Both run on the Windows 10 you have today.","items":[{"image":{"label":"Man speaking and gesturing at a desktop computer in a workshop."},"title":"Local processing that never needs a connection","body":"Download a Whisper or Parakeet model once, and every dictation after that runs on your own device. No internet, no server, and it keeps working through the day the Wi-Fi drops."},{"image":{"label":"Person speaking and gesturing at a desktop monitor at a kitchen table."},"title":"Cloud transcription with nothing to download","body":"Sign in and dictate. OpenWhispr Cloud transcribes on our servers for the fastest, most consistent results, with no model to download and no strain on an older machine."},{"image":{"label":"Man gesturing toward a laptop in a room with acoustic panels and a guitar."},"title":"Dictate wherever your cursor already is","body":"Press a hotkey and speak. The text is typed into Word, Outlook, Edge, Teams, Notepad or any app that accepts typed text. No per-app setup needed."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Internet needed in local mode","value":"None","footnote":"once the model is downloaded, dictation works fully offline"},{"label":"Cost for unlimited local dictation","value":"$$0","footnote":"free forever, no Windows 11 upgrade needed"},{"label":"Setup takes","value":"5 min","footnote":"from download to your first dictated line on Windows 10"}]},"howItWorks":{"label":"How it works","title":"Installed on the Windows 10 you already have.","subtitle":"Start free. Three simple steps and you're dictating today.","image":{"label":"Man speaking and gesturing at a laptop beside a window and a mug."},"steps":[{"title":"Download for free.","body":"Windows 10 or later, 64-bit. No Windows 11, and nothing extra to grant beyond your microphone."},{"title":"Dictate the note.","body":"Press the hotkey and speak. Choose Cloud for speed, or a local model if you'd rather skip the connection entirely."},{"title":"Review, paste, done.","body":"Dictation cleanup tidies filler words, grammar and punctuation before it lands where your cursor is, so a quick reply is ready to send. Edit if you need to, and you're done."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"No convincing needed","title":"10 reasons Windows 10 users choose OpenWhispr","subtitle":"Not a workaround for an OS Microsoft moved past. A dictation tool built to run on it.","items":[{"title":"Runs on Windows 10, not just Windows 11","body":"The minimum requirement is Windows 10 or later, 64-bit. No Windows 11 gate, no new hardware to buy."},{"title":"Works with no internet connection at all","body":"Local models process entirely on your device once downloaded, unlike Windows' built-in dictation, which needs a live connection every time."},{"title":"The published requirement has no expiry date","body":"Microsoft's own Office security updates on Windows 10 are set to run through 10 October 2028; Edge and WebView2 keep updating on Windows 10 22H2 until at least that same date. OpenWhispr's requirement is simply Windows 10 or later, with no end date attached to it."},{"title":"Free, with no card and no expiry","body":"Unlimited local dictation on the Free plan, forever. No trial clock, no card required to start."},{"title":"Dictate into whatever you're already using","body":"One hotkey types clean text into Word, Outlook, Edge, Teams, Notepad, or any app with a text field. Nothing to configure per app."},{"title":"Hold to talk or tap to talk, your choice","body":"Windows supports both activation modes for dictation: hold the key while you speak, or press once to start and again to stop. Built-in dictation forces an all-or-nothing switch between speaking and typing that breaks your rhythm; picking the mode that fits you means it doesn't."},{"title":"Teach it the words it keeps getting wrong","body":"A custom dictionary learns names, product codes and jargon, so the terms you use every day stop coming out mangled."},{"title":"GPU acceleration for local models, one click away","body":"NVIDIA, AMD and Intel GPUs get a one-click acceleration download from Settings. On older hardware, pick a smaller local model, or skip local entirely and use Cloud, which asks very little of your computer."},{"title":"Say what you want changed, not just what you said","body":"Select the text, press the voice agent hotkey (set it once in Settings), and say what to change: \"make this more formal\", \"fix the typos\". The rewritten version lands in place, not your raw words."},{"title":"Open source, so IT can check for themselves","body":"Every line is public on GitHub. Before it goes on a work PC, someone can read exactly what happens to the audio."}]},"faq":{"label":"Before you download","title":"The questions Windows 10 users ask first.","subtitle":"Straight answers on compatibility, cost and accuracy. If yours isn't here, ask us in Discord. We reply within 24 hours.","items":[{"question":"Does OpenWhispr actually run on Windows 10?","answer":"Yes. It installs on Windows 10 or later, 64-bit, no Windows 11 required."},{"question":"Do I need an internet connection to dictate?","answer":"Not if you choose a local model. Local transcription runs entirely on your device and works with no internet connection at all once the model is downloaded. OpenWhispr Cloud needs a connection, since your audio is sent to our servers to transcribe it."},{"question":"Will it run well on an older Windows 10 PC?","answer":"Yes. Pick a smaller local model for older hardware, or use the one-click GPU acceleration in Settings. Cloud is the alternative if you'd rather skip local models altogether."},{"question":"How is this different from Windows' built-in dictation?","answer":"Win+H needs an internet connection even for plain dictation, and the older Windows Speech Recognition's per-user training option isn't reliably there anymore either. OpenWhispr's custom dictionary learns the names and terms that keep coming out wrong, with no training menu to go hunting for."},{"question":"What does it cost?","answer":"Free, with unlimited local dictation and no card needed. Pro is $8/month for unlimited Cloud transcription, sync and mobile. Business is $16/user/month and adds unlimited meeting recordings and chat over your data."},{"question":"Can it learn my own words and terms?","answer":"Yes. The custom dictionary learns names, product codes and jargon you use often, on every plan."}]},"reviews":{"label":"What users say","title":"What people actually said, unedited.","subtitle":"From our Discord, GitHub and inboxes: people who kept an older machine going and wanted the dictation to keep up.","score":{"value":"0","label":"Files stored on a server","body":"In local mode the audio never leaves your device. Nothing is stored on a server, and nothing is sent to a transcriber."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for Windows 10 users","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Windows 10 users who want free, unlimited local dictation, no Windows 11 or internet required"}]},{"parts":[{"text":"Windows 10 users who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Teams keeping older Windows 10 machines productive, with unlimited meeting recordings and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Dragon Home, the affordable paid option, is gone from sale. OpenWhispr is free, and stays free"}]},{"parts":[{"text":"Cloud transcription that doesn't ask anything of an older CPU"}]},{"parts":[{"text":"Unlimited meeting recordings plus chat over your data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Some machines aren't getting upgraded. We built for those too.","paragraphs":["OpenWhispr started with a simple objection: dictation software shouldn't stop working the day your OS falls out of support. Windows 10 is still Windows, so dictation, Cloud and local transcription work the same as they do everywhere else we ship.","We build in public. Every release and every line of code is on GitHub, where 5,000+ engineers have starred, forked and audited the project. When we say local dictation runs on your device, that's not a promise in a privacy policy. It's a property of the code, and anyone still running Windows 10 can verify it themselves.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"Team / community photo"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Your words, typed on the Windows 10 PC you already have.","body":"Download OpenWhispr and your next dictation lands as clean text, offline if you want it. No Windows 11 required.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Takes 5 minutes to set up. If it doesn't work, we'll fix it with you in Discord. Free forever on Mac, Windows & Linux."}}},"useCaseUbuntu":{"breadcrumb":{"home":"Home","useCases":"Use Cases","ubuntu":"For Ubuntu Users"},"metadata":{"title":"Speech to Text Software for Ubuntu - OpenWhispr","description":"OpenWhispr turns speech into editable text at your Ubuntu cursor. Use Cloud for fast setup or local models for offline and sensitive work.","keywords":["speech to text software for ubuntu","ubuntu speech to text","ubuntu dictation software","voice typing ubuntu","linux dictation app","offline speech to text ubuntu"]},"schema":{"name":"OpenWhispr","applicationCategory":"Ubuntu speech-to-text software","operatingSystem":"Linux","description":"Open-source speech-to-text software for Ubuntu users. Use a global hotkey to turn speech into editable text for the field you have focused. Choose Cloud without a local model download, or local for offline and device-only transcription.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text software for Ubuntu users","title":"Press the hotkey. Get text at your cursor.","body":"Speak into the document, message or prompt you already have open. Choose Cloud for a fast start without a local model download, or local for offline and sensitive work.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman speaking beside a desktop monitor with one hand on a keyboard in a bright room."}},"trustBar":{"label":"Dictate into:","names":["Documents","Messages","Prompts","Terminals","Browser forms","Text editors"]},"pain":{"label":"What makes voice typing feel unfinished","title":"You need text in this field, not a transcript in another tab.","subtitle":"A hotkey should turn one spoken paragraph into an editable draft where you are already writing. It should not turn that paragraph into a browser, script, or copy-paste detour.","cards":[{"title":"A browser tab leaves you copying.","body":"You are writing a reply, document, or prompt now. Browser transcription makes the words arrive somewhere else first. OpenWhispr is a desktop app for the moment the field in front of you needs a first draft."},{"title":"Source-code setup is not a shortcut.","body":"A usable Ubuntu dictation tool starts as a Linux desktop app. You should be choosing how to transcribe, not assembling a voice-typing workflow before you can write one sentence."},{"title":"Wayland auto-paste is conditional.","body":"OpenWhispr chooses a compositor-specific paste method instead of promising the same behavior on every desktop session. KDE with XWayland tries the RemoteDesktop portal first; GNOME and other Wayland compositors try Linux input access first, then use later fallbacks."},{"title":"A missed paste should not erase the sentence.","body":"When automatic paste is unavailable, OpenWhispr copies the completed transcription to the clipboard. Paste it with Ctrl+V, edit it, and carry on without dictating the paragraph again."}]},"features":{"label":"The whole voice-to-text loop","title":"Speak. See editable text. Keep writing.","subtitle":"The job is typing by voice in the field you already opened. Cloud and local models are two deliberate ways to do it, not separate products.","items":[{"image":{"label":"Woman speaking and gesturing beside an open laptop at a desk."},"title":"One hotkey for the field you are in","body":"Put the cursor where the sentence belongs, press the global hotkey, speak, and release. With auto-paste available, OpenWhispr puts the transcription in the active app. Otherwise it is ready in the clipboard."},{"image":{"label":"Woman speaking toward a desktop monitor while using a mouse at a wooden desk."},"title":"Set up automatic paste for your desktop session","body":"On Wayland, OpenWhispr chooses a compositor-specific paste method. On GNOME and KDE, approve the RemoteDesktop permission when asked. If automatic paste cannot reach the field, the completed transcription stays in the clipboard."},{"image":{"label":"Man holding a sheet of paper and speaking beside a laptop near a rain-streaked window."},"title":"Choose Cloud or local for the conditions","body":"Choose OpenWhispr Cloud for top accuracy and speed without a local model download, with audio discarded after transcription. Download Whisper or Parakeet when offline or device-only transcription is the better fit."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Local-mode audio uploaded","value":"None","footnote":"a downloaded model transcribes on your Ubuntu device"},{"label":"Cost for local dictation","value":"$$0","footnote":"unlimited local AI models on every plan"},{"label":"When automatic paste fails","value":"Ctrl+V","footnote":"the transcription is copied for a manual paste"}]},"howItWorks":{"label":"How it works on Ubuntu","title":"Start in the field you want to fill.","subtitle":"Install the Linux desktop app, choose Cloud or a local model, then use the hotkey in the document, reply, or prompt already in front of you.","image":{"label":"Man speaking toward a desktop monitor at a desk beside a window overlooking the sea."},"steps":[{"title":"Install and choose a transcription path.","body":"Download the Linux desktop app. Choose OpenWhispr Cloud to start without downloading a model, or download a local Whisper or Parakeet model for device-only transcription."},{"title":"Put the cursor where the sentence belongs.","body":"Open your document, message, or prompt. Put the cursor where the sentence belongs, press the global hotkey, speak, then release it."},{"title":"Review the text where you work.","body":"With auto-paste set up, OpenWhispr sends the text to the active app. Otherwise paste the completed transcription from the clipboard, then edit, send, or save it."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"The parts that make it usable","title":"10 reasons Ubuntu users choose OpenWhispr","subtitle":"The result should be editable text where you are writing. These are the product choices and Linux details that make that workflow practical.","items":[{"title":"Start at the cursor you already need","body":"Put the cursor in the document, reply, or prompt first. The global hotkey turns the spoken paragraph into text for the active app, so your draft starts where the next sentence belongs."},{"title":"Use a Linux desktop app, not a browser detour","body":"OpenWhispr ships Linux desktop targets, including a Debian package. Install the app, set a hotkey, and dictate from the program where the writing is already happening."},{"title":"Wayland auto-paste has a visible setup path","body":"OpenWhispr does not promise one Wayland paste path. KDE with XWayland tries the RemoteDesktop portal first; GNOME and other Wayland compositors try Linux input access first, then use later fallbacks. The Debian package recommends wl-clipboard for clipboard support."},{"title":"Clipboard mode is still a usable fallback","body":"If automatic paste is not available, the transcription is copied to the clipboard. One Ctrl+V puts the completed sentence in the focused field without repeating the dictation."},{"title":"Cloud skips the local-model download","body":"Choose OpenWhispr Cloud when you want top accuracy and speed without managing a local model. Cloud audio is sent only to transcribe it, then never stored."},{"title":"Local works offline and stays on the device","body":"Download a Whisper or Parakeet model once, then transcribe without an internet connection. The local path keeps the audio and transcription work on your Ubuntu machine."},{"title":"Speak one language, paste another","body":"Use the dedicated translation hotkey to dictate in one language and paste text in another. It still starts from the focused field, not a separate transcription queue."},{"title":"Keep project words out of the correction loop","body":"Add project names, technical terms, and acronyms to the custom dictionary. It helps OpenWhispr put the spelling you use into the field already open, so you edit instead of retype it."},{"title":"Choose the engine for the language you speak","body":"Before you dictate into the next field, choose an engine that supports the language you speak. The website's 100+ refers to the underlying Whisper model family; the app currently offers 60 selectable transcription languages plus auto-detect."},{"title":"The local path starts at $0","body":"Every plan includes unlimited local AI models. Start with device-only dictation at no cost, then use the plan comparison to decide whether Cloud's no-download convenience is worth an upgrade."}]},"faq":{"label":"Before you install","title":"Six Ubuntu questions worth answering first.","subtitle":"Straight answers on typing by voice, Wayland setup, offline work, and where your audio goes.","items":[{"question":"Can this replace the keyboard for a paragraph?","answer":"Put the cursor in the field you want to fill, press the hotkey, speak, and release. With automatic paste available, the transcription goes to the active app. Otherwise paste it from the clipboard and edit it there."},{"question":"Does automatic paste work on Ubuntu Wayland or X11?","answer":"Current releases provide Linux .deb, AppImage, RPM, and tar.gz downloads. On Wayland, OpenWhispr uses a compositor-specific paste path: KDE with XWayland tries the RemoteDesktop portal first; GNOME and other Wayland compositors try Linux input access first, then use later fallbacks. The Debian package recommends wl-clipboard for clipboard support. Test the hotkey-to-field flow on your own session. If automatic paste is unavailable, paste the transcription manually with Ctrl+V."},{"question":"Should I choose Cloud or a local model?","answer":"Choose OpenWhispr Cloud for top accuracy and speed without a local model download. Cloud audio is sent only to transcribe it, then never stored. Choose a local Whisper or Parakeet model when you need offline or device-only transcription."},{"question":"Can I dictate with no internet connection?","answer":"Yes. Download a local model first, then transcription runs on your Ubuntu device without an internet connection. Local models are unlimited on every plan."},{"question":"Can it handle my project names and acronyms?","answer":"Add them to the custom dictionary. It is included on every plan and is designed to improve recognition of names, jargon and acronyms in your transcripts."},{"question":"What does a paid plan add after the free local path?","answer":"Free includes unlimited local models and 2,000 OpenWhispr Cloud words each week. Pro is $8 monthly or $80 yearly for unlimited Cloud transcription, 20 monthly meeting-recording hours, sync, API and MCP access, Agent mode, and the iPhone and iPad app. Business is $16 monthly or $160 yearly and adds unlimited meeting recordings, chat over your data, and priority support."}]},"reviews":{"label":"What users say","title":"Built for people who need text in the field they are using.","subtitle":"Three users on why local models, voice input, and offline work changed their typing workflow.","score":{"value":"At your cursor","label":"Where live dictation belongs","body":"With automatic paste available, OpenWhispr puts the transcription in the active app. Otherwise, it remains in the clipboard for you to paste and review there."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for Ubuntu dictation","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Ubuntu users who want unlimited local dictation with a limited Cloud allowance"}]},{"parts":[{"text":"Ubuntu users who want unlimited Cloud dictation without a local model download"}]},{"parts":[{"text":"People who need paid desktop workflow features beyond unlimited Cloud dictation"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local dictation, plus 2,000 OpenWhispr Cloud words each week"}]},{"parts":[{"text":"Unlimited Cloud dictation, plus sync, API and MCP access"}]},{"parts":[{"text":"Unlimited Cloud dictation, team features, chat over your data and priority support"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with team features and admin, SSO, SAML and SCIM, audit logs, retention controls, and dedicated support."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built for the Ubuntu desktop","title":"The sentence belongs in your app, not in a transcription queue.","paragraphs":["OpenWhispr starts from a simple expectation: a hotkey should turn a spoken thought into editable text for the Ubuntu app where the cursor already is. The rest of the desktop workflow should not become somebody else's project.","We build in public. The desktop app is MIT licensed on GitHub. You can inspect the Linux downloads, local-model choices, and Wayland setup path before choosing how you want to transcribe.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"Free · open source · built for desktop dictation","title":"Put the next sentence at your Ubuntu cursor.","body":"Download OpenWhispr, choose Cloud or a local model, then speak into the document, message, or prompt already open. Review the text where it lands.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free local models on Linux. Cloud starts with 2,000 words each week."},"byline":{"authorLabel":"Written by","author":"Joshua Padoa","aiDisclosure":"Drafted with AI assistance","reviewerLabel":"Reviewed and fact checked by","reviewer":"Wesley van der Hoop","updatedLabel":"Last reviewed","dateModified":"2026-09-10"}}},"useCaseMicrosoftWord":{"breadcrumb":{"home":"Home","useCases":"Use Cases","microsoftWord":"For Microsoft Word Users"},"metadata":{"title":"Free Speech to Text Software for Microsoft Word - OpenWhispr","description":"OpenWhispr turns speech into editable text. With auto-paste, it can paste at an active, writable Word cursor or keep text in clipboard. No Word add-in.","keywords":["free speech to text software for microsoft word","speech to text for Word","free dictation software for Word","voice typing for Microsoft Word","Word dictation alternative","offline speech to text for Word"]},"schema":{"name":"OpenWhispr","applicationCategory":"Speech-to-text software for Microsoft Word users","operatingSystem":"macOS, Windows, Linux","description":"Desktop speech-to-text software that can paste generated text at an active writable cursor in Microsoft Word. It is not a native Word integration.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech-to-text software for Microsoft Word","title":"Set the cursor in Word. Get editable text.","body":"OpenWhispr is desktop dictation software, not a Word add-in. With auto-paste, it can paste generated text at an active, writable Word cursor.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Woman speaking and gesturing beside a laptop at a table in a bright studio."}},"trustBar":{"label":"Compare these dictation and review tools:","names":["Microsoft Word","Word Dictate","Windows voice typing","Voice Access","Microsoft Editor","Track Changes"]},"pain":{"label":"What slows a Word draft","title":"The cursor is waiting. The first draft is still stuck in your head.","subtitle":"The practical question is not whether Word has dictation. It is how spoken text reaches your document, and what still needs a review pass.","cards":[{"title":"Native Word Dictate is only one route.","body":"Word Dictate is Word's own feature. OpenWhispr is separate desktop software that can paste text at the active Word cursor when auto-paste is enabled. The two are not the same kind of tool."},{"title":"Leaving Word to move text back in.","body":"Copying a draft from another window creates another stop. OpenWhispr can paste generated text at the focused, writable Word cursor, so the text arrives in the document you already have open."},{"title":"Recognition errors turn into editing work.","body":"Names, terms and a rough microphone can still need correction. OpenWhispr gives you editable text and a custom dictionary. Word remains where you proofread and revise the result."},{"title":"Free should not hide the tradeoff.","body":"Local speech-to-text is unlimited and free. OpenWhispr Cloud includes 2,000 words a week on Free, then paid plans add unlimited Cloud transcription and other workflow features."}]},"features":{"label":"What OpenWhispr does","title":"Use speech to make the Word draft. Keep Word for Word work.","subtitle":"Choose Cloud for fast transcription without a local model download. Choose local processing for device-only or offline work. Then review the editable text in Word.","items":[{"image":{"label":"Man speaking beside a laptop at a desk surrounded by books and papers."},"title":"Generated text at the active Word cursor","body":"Put the cursor in a writable part of Word, use the desktop hotkey and speak. With auto-paste, OpenWhispr can paste generated text at that focused cursor. Otherwise, use the clipboard."},{"image":{"label":"Woman gesturing beside a laptop at a train window."},"title":"Cloud for speed, local for device-only work","body":"Cloud transcription is the fast option with no local model download and transcription audio discarded after processing. Local models run on your device and can work offline."},{"image":{"label":"Woman speaking beside a laptop and mug at a library table."},"title":"Word still owns formatting and review","body":"OpenWhispr does not control Word formatting, comments, Track Changes or Word commands. It supplies text. Use Word's own tools to format, comment, track and accept edits."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Word add-in required","value":"None","footnote":"OpenWhispr is a desktop app, not a native Word integration"},{"label":"Cost for unlimited local speech-to-text","value":"$$0","footnote":"on every desktop plan, with a local model"},{"label":"Setup estimate","value":"5 min","footnote":"from download to a first local dictation"}]},"howItWorks":{"label":"How it works","title":"A Word draft you can review before the next sentence slips away.","subtitle":"Choose a transcription path, start at the active Word cursor, then make Word's own review tools the final pass.","image":{"label":"Woman speaking and gesturing toward a laptop at a wooden table under a pendant lamp."},"steps":[{"title":"Download for free.","body":"OpenWhispr runs on macOS, Windows and Linux. Choose Cloud for no local model download or local for on-device and offline speech-to-text."},{"title":"Set the active Word cursor and speak.","body":"Click a writable place in Word, press the desktop hotkey and dictate. With auto-paste, generated text can be pasted at the active cursor. Otherwise, use the clipboard."},{"title":"Review the text in Word.","body":"Correct, format, comment and track edits with Word's own tools. OpenWhispr does not operate those Word features for you."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"Before you choose a Word dictation method","title":"10 reasons Microsoft Word users choose OpenWhispr","subtitle":"This is speech-to-text for the document you already use, with clear boundaries around what happens at the cursor and what stays a Word job.","items":[{"title":"No Word add-in to install","body":"OpenWhispr is desktop dictation software. It does not add a ribbon, plugin or API connection to Word. It uses the active writable cursor when automatic pasting is enabled."},{"title":"One hotkey instead of a new drafting surface","body":"Start from the document you already have open. The desktop hotkey records your speech, then generated text can be pasted at the active Word cursor."},{"title":"Editable text, not a locked transcript","body":"With automatic pasting, text can reach your active cursor. Otherwise, use the clipboard. You can revise the draft in Word just as you would after typing it. Review remains important for names, specialist terms and meaning."},{"title":"Word retains its own controls","body":"Formatting, comments, Track Changes and native Word commands stay in Word. OpenWhispr does not control them or claim to edit Word as an integration."},{"title":"A custom dictionary for names and jargon","body":"Add names, terms and acronyms to OpenWhispr's custom dictionary as context hints for the speech model. It can improve recognition, but you still review the final Word text."},{"title":"Local speech-to-text works offline","body":"Choose a local speech model to process dictation on your device without an internet connection. That is useful when online speech input is not the right fit."},{"title":"Cloud when you want the fast setup","body":"OpenWhispr Cloud provides speed and accuracy without downloading a local model. Cloud transcription audio is sent only to transcribe, then immediately discarded."},{"title":"A clipboard option when you prefer it","body":"You can keep dictated text in your clipboard for manual placement. Automatic pasting is optional, which keeps the active-cursor workflow distinct from an app integration."},{"title":"Unlimited local dictation at no cost","body":"Local AI models are unlimited on every desktop plan, including Free. This makes it practical to test a real drafting and correction workflow before paying for Cloud features."},{"title":"Open source, so the mechanism is inspectable","body":"OpenWhispr is open source. You can inspect how the app handles dictation rather than treating a Word compatibility claim as a black box."}]},"faq":{"label":"Before you download","title":"The questions Word users ask first.","subtitle":"Straight answers about active-cursor text insertion, Word's own tools, privacy, local processing and what Free includes.","items":[{"question":"Is OpenWhispr a Microsoft Word integration?","answer":"No. OpenWhispr is a desktop dictation app, not a Word add-in, plugin, API integration or Word-specific feature. With auto-paste, it can paste generated text at the active, writable cursor. That cursor can be in Word. Otherwise, use the clipboard."},{"question":"How is this different from Word Dictate?","answer":"Word Dictate is Microsoft's native Word feature. OpenWhispr runs separately on your desktop. Its generated text can be pasted at an active Word cursor, but it does not become part of Word or gain native Word controls."},{"question":"Can OpenWhispr format Word text or use Track Changes?","answer":"No. It does not control Word formatting, comments, Track Changes or native Word commands. Use Word for those actions after the text arrives at the cursor."},{"question":"What does free speech-to-text include?","answer":"The Free plan includes unlimited local AI models and 2,000 words a week of OpenWhispr Cloud transcription. Local speech-to-text has no internet requirement. Pro adds unlimited OpenWhispr Cloud transcription, sync, mobile and API or MCP access."},{"question":"What happens to my audio?","answer":"With a local transcription model, processing stays on your device. With OpenWhispr Cloud, audio is sent only to transcribe it and is immediately discarded. Choose the path that fits the document and your working conditions."},{"question":"What if a name or specialist term is wrong?","answer":"Review the editable text in Word before you finish the document. You can add names, jargon and acronyms to OpenWhispr's custom dictionary as context hints, then correct anything the finished document still needs."}]},"reviews":{"label":"What users say","title":"Less typing. More control over where the text goes.","subtitle":"Three verbatim notes from people using OpenWhispr. They speak to the tool, not a Microsoft Word integration.","score":{"value":"1","label":"document cursor you choose","body":"With auto-paste, generated text can go to the active writable cursor. Word is one possible destination, not an integration surface. Otherwise, use the clipboard."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for Microsoft Word users","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Word users who want a useful free start with local speech-to-text and limited Cloud transcription"}]},{"parts":[{"text":"Word users who want unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Teams that also need unlimited meeting recordings and chat over data"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Useful Free access, including unlimited local speech-to-text and limited Cloud transcription"}]},{"parts":[{"text":"Unlimited Cloud transcription plus mobile, sync, and API/MCP access"}]},{"parts":[{"text":"Unlimited meeting recordings and chat over your data"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"Moderate: configure users, Agent mode, and data workflows"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","We also offer a custom Enterprise plan with managed onboarding and tailored features."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Built in the open","title":"Writing should not require leaving the document you already use.","paragraphs":["OpenWhispr started with a simple idea: say what you mean, then put the resulting text where you are already writing. In this workflow, that can be the active cursor in a Word document.","We build in public. Every release and every line of code is on GitHub, so anyone can inspect how the app handles dictation. When we say local dictation stays local, that is a property of the code you can verify.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"OpenWhispr team and community photo placeholder"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Keep Word as the workspace. Use speech for the first draft.","body":"Download OpenWhispr, put the cursor in Word and speak. With auto-paste, generated text can arrive at the active writable cursor. Otherwise, use the clipboard before your Word review.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Set up a hotkey, select a transcription path and test it in a Word document. Free local speech-to-text is available on macOS, Windows and Linux."},"byline":{"authorLabel":"Written by","author":"Joshua Padoa","aiDisclosure":"Drafted with AI assistance","reviewerLabel":"Reviewed and fact checked by","reviewer":"Wesley van der Hoop","updatedLabel":"Last reviewed","dateModified":"2026-09-10"}}},"useCaseGoogleDocs":{"breadcrumb":{"home":"Home","useCases":"Use Cases","googleDocs":"For Google Docs Users"},"metadata":{"title":"Google Docs App Speech To Text - OpenWhispr","description":"OpenWhispr is a free desktop speech-to-text app that pastes dictated text at your active Google Docs cursor. No Google Docs add-on required.","keywords":["google docs app speech to text","speech to text for google docs","google docs dictation app","voice to text google docs desktop","google docs speech to text app","dictate into google docs"]},"schema":{"name":"OpenWhispr","applicationCategory":"Desktop speech-to-text software for Google Docs users","operatingSystem":"macOS, Windows, Linux","description":"Open-source desktop speech-to-text software that turns speech into editable text at a focused cursor. It is not a Google Docs add-on or native integration. Free on macOS, Windows and Linux.","price":"0","priceCurrency":"USD"},"landing":{"hero":{"eyebrow":"Speech to text for Google Docs users","title":"Press the hotkey. Get text in your Google Doc.","body":"On desktop, OpenWhispr turns your speech into editable text and pastes it at a focused Google Docs cursor. It is system-wide dictation, not a Google Docs add-on or native integration.","cta":{"label":"Download for free","href":"/download"},"trust":"Open source · 80,000+ downloads","image":{"label":"Man gesturing toward a laptop at an outdoor table overlooking city buildings."}},"trustBar":{"label":"Dictate into on desktop:","names":["Google Docs","Microsoft Word","Gmail","Notion","Slack","Obsidian"]},"pain":{"label":"The writing moment","title":"The cursor is ready. Your hands are not.","subtitle":"You want a paragraph in this document, not a transcript stranded in another app.","cards":[{"title":"A blank Google Doc and no easy way to type.","body":"OpenWhispr starts from a desktop hotkey. Speak the paragraph, then return to editable text at the focused document cursor."},{"title":"A recorder leaves you with a transfer job.","body":"The destination matters. Desktop dictation can paste text into the document in front of you, so you review the draft there instead of moving text from a separate transcript."},{"title":"The browser tool is not the same as mobile.","body":"Google Docs desktop voice typing and phone keyboard dictation are separate paths. This OpenWhispr page describes the desktop app, not a claim about dictating into Docs from mobile."},{"title":"An integration claim creates the wrong expectation.","body":"OpenWhispr does not need a Google Docs plugin to use the active cursor. It is a system-wide desktop app, not a native Google Docs integration."}]},"features":{"label":"Desktop dictation for the document in front of you","title":"Speak the draft. Review it in place.","subtitle":"Use a global hotkey while a writable document cursor is focused. Local transcription is unlimited on every plan.","items":[{"image":{"label":"Man wearing glasses gesturing toward a desktop monitor at a desk with papers."},"title":"A hotkey, then editable text at the cursor","body":"Focus the Google Docs document on desktop, hold your OpenWhispr hotkey, and speak. The app uses the focused editable cursor for its paste path."},{"image":{"label":"Woman wearing a blue headscarf gesturing toward a laptop at a bright office table."},"title":"Choose the processing route that fits the draft","body":"Local models run on your device and work offline. OpenWhispr Cloud offers top accuracy and speed without a local model download, then immediately discards transcription audio."},{"image":{"label":"Woman holding a pen over papers and gesturing beside a desktop monitor in an evening office."},"title":"A usable fallback when automatic paste cannot run","body":"If the desktop app cannot automatically paste, it copies the dictated text to the clipboard for a manual paste. That is a fallback, not a Google Docs integration."}]},"stats":{"ariaLabel":"Key facts","items":[{"label":"Cloud transcription audio retained","value":"None","footnote":"Cloud audio is transcribed then immediately discarded. Local audio stays on your device."},{"label":"Cost for local desktop dictation","value":"$$0","footnote":"local dictation is unlimited on every plan, including Free"},{"label":"Desktop start","value":"Hotkey","footnote":"download, choose a model, focus a writable document"}]},"howItWorks":{"label":"From desktop hotkey to document","title":"Focus the doc. Speak the text. Review the result.","subtitle":"OpenWhispr uses system-wide desktop dictation. It does not install inside Google Docs or claim a native Docs connection.","image":{"label":"Man gesturing toward a desktop monitor at a wooden desk with a keyboard and mug."},"steps":[{"title":"Download for free.","body":"Install on macOS, Windows, or Linux. Local models are unlimited on every plan and can run offline."},{"title":"Focus Google Docs, then hold the hotkey.","body":"Keep a writable Google Docs cursor focused in your desktop browser. Speak the paragraph and OpenWhispr uses its paste path for that active cursor."},{"title":"Review the editable paragraph.","body":"Correct the document where you are already working. If automatic paste is unavailable, paste the copied text manually from the clipboard."}],"cta":{"label":"Download for free","href":"/download"}},"reasons":{"label":"The desktop route, clearly bounded","title":"10 reasons Google Docs users choose OpenWhispr","subtitle":"For people who want speech to become editable text in the active desktop document, without claiming a Google Docs add-on.","items":[{"title":"The cursor stays the destination.","body":"OpenWhispr is built to paste dictated text at a focused editable cursor. In Google Docs on desktop, that means the active document remains where you review the paragraph."},{"title":"No Docs add-on to install.","body":"The workflow relies on the desktop app's system-wide hotkey and paste behavior, not a Google Docs extension, sidebar, or native integration."},{"title":"Desktop and phone are different decisions.","body":"This is the desktop route for macOS, Windows, and Linux. Google Docs mobile voice paths and the paid OpenWhispr iPhone and iPad app are separate product contexts."},{"title":"Keep the document focused while you dictate.","body":"The desktop paste path is designed for a focused editable cursor. If automatic paste is unavailable, OpenWhispr copies the dictated text to the clipboard for a manual paste."},{"title":"Clipboard fallback preserves the text.","body":"When automatic pasting is unavailable, OpenWhispr writes the result to the clipboard so you can paste it manually into the document."},{"title":"Local works when the connection does not.","body":"Choose local models for on-device, offline transcription. You can keep drafting on the desktop even without an internet connection."},{"title":"Local desktop dictation is unlimited and free.","body":"Every plan includes unlimited local AI models. The Free plan's 2,000-word weekly limit is for OpenWhispr Cloud transcription, not local dictation."},{"title":"Cloud is the no-download route.","body":"OpenWhispr Cloud is the choice for top accuracy and speed without downloading local models. Cloud transcription audio is immediately discarded."},{"title":"A custom dictionary can help with document terms.","body":"Add recurring names, jargon, and acronyms to OpenWhispr's custom dictionary for sharper transcripts when your document uses specialized language."},{"title":"Pro adds cloud and cross-device workflow features.","body":"Pro includes unlimited OpenWhispr Cloud transcription, sync across devices, personal API access, MCP integration, and the iPhone and iPad app. None of those features claim a native Google Docs integration."}]},"faq":{"label":"Before you choose a route","title":"Questions about Google Docs speech to text.","subtitle":"Straight answers on desktop, mobile, privacy, cost, and the integration boundary.","items":[{"question":"Can I dictate directly into Google Docs with OpenWhispr?","answer":"On desktop, focus a writable cursor in the active Google Docs document, then use the OpenWhispr hotkey. The app's system-wide paste path is designed to put dictated text at that focused cursor. It is not a Google Docs add-on or native integration."},{"question":"What if automatic paste does not work?","answer":"OpenWhispr writes the dictated text to the clipboard. You can paste it manually into Google Docs. On macOS, automatic pasting requires Accessibility permission."},{"question":"Is this the same as Google Docs Voice typing?","answer":"No. Google Docs has its own desktop Voice typing feature. OpenWhispr is a separate system-wide desktop dictation app that can use a focused editable cursor in Google Docs."},{"question":"Does OpenWhispr dictate into Google Docs on my phone?","answer":"This page makes no claim that it does. The OpenWhispr iPhone and iPad app is included with Pro and above, but mobile dictation into Google Docs is not established by the desktop product sources used here."},{"question":"Can I use local speech to text offline?","answer":"Yes. Local AI models run entirely on your device without an internet connection and are unlimited on every plan. Google Docs itself may still need its usual connection to save and sync your document."},{"question":"What does Pro add for this workflow?","answer":"Free includes unlimited local models and 2,000 OpenWhispr Cloud words each week. Pro adds unlimited Cloud transcription, sync across devices, personal API access, MCP integration, and the iPhone and iPad app."}]},"reviews":{"label":"What people tell us","title":"A desktop app for words at the cursor.","subtitle":"These are general product reviews, not claims of a native Google Docs integration.","score":{"value":"Desktop","label":"System-wide dictation","body":"The relevant product behavior here is system-wide desktop dictation with paste at a focused editable cursor, plus a clipboard fallback."},"items":[{"quote":"Started using OpenWhispr with my local models and hardware. I have to say, this is better than typing.","name":"0xSero","handle":"@0xSero"},{"quote":"I switched to voice input for my Claude Code chats. Even when I stumble and rephrase mid-sentence, the transcription comes out clean and polished.","name":"j3iiifn","handle":"@j3iiifn"},{"quote":"I needed a local model for when I'm travelling or have no WiFi, and this perfectly solves it. The UI and performance are great.","name":"Gabe","handle":"@gabe__perez"}]},"pricing":{"label":"Pricing","titleLines":["Free forever.","Upgrade when you want more."],"caption":"Compare OpenWhispr plans for Google Docs users","tiers":[{"id":"free","name":"Free","badge":"Popular","price":"$$0","priceDetail":"No card details needed","cta":{"label":"Download for free"},"featured":true},{"id":"pro","name":"Pro","cta":{"label":"Start with Pro"}},{"id":"business","name":"Business","cta":{"label":"Start with Business"}}],"rows":[{"label":"Best for","cells":[{"featured":true,"parts":[{"text":"Desktop Google Docs users who want unlimited local dictation"}]},{"parts":[{"text":"People who want unlimited Cloud transcription and cross-device features"}]},{"parts":[{"text":"Teams that need unlimited meeting recordings and priority support"}]}]},{"label":"Main reason to choose","cells":[{"featured":true,"parts":[{"text":"Unlimited local desktop dictation at no cost."}]},{"parts":[{"text":"Unlimited Cloud transcription plus sync, API, MCP, and mobile access"}]},{"parts":[{"text":"Everything in Pro plus unlimited meeting recordings and priority support"}]}]},{"label":"Offline speech-to-text","tooltip":"Offline speech-to-text runs entirely on your own device, it works without internet connection, and it's unlimited and free.","cells":[{"featured":true,"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]},{"parts":[{"text":"Included","emphasis":"accent"}]}]},{"label":"Cloud speech-to-text","tooltip":"Cloud speech-to-text sends audio to OpenWhispr Cloud for the highest accuracy and speed, with no local models to download.","cells":[{"featured":true,"parts":[{"text":"2,000 words/week","emphasis":"strong"},{"text":" via OpenWhispr Cloud, or unlimited using your own API keys"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"},{"text":" OpenWhispr Cloud speech-to-text"}]}]},{"label":"Meeting recordings","cells":[{"featured":true,"parts":[{"text":"5 hours/month","emphasis":"strong"}]},{"parts":[{"text":"20 hours/month","emphasis":"strong"}]},{"parts":[{"text":"Unlimited","emphasis":"strong"}]}]},{"label":"Privacy & data handling","cells":[{"featured":true,"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]},{"parts":[{"text":"Offline audio stays on your device. Cloud audio is sent only to transcribe it, then never stored."}]}]},{"label":"Setup experience","cells":[{"featured":true,"parts":[{"text":"Hands-on: install app, choose local models, optionally add API keys"}]},{"parts":[{"text":"Easy: install app, upgrade, use unlimited cloud speech-to-text"}]},{"parts":[{"text":"More: unlimited meeting recordings and chat over your data"}]}]},{"label":"Advanced features","cells":[{"featured":true,"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"}]},{"bullets":[{"title":"Custom dictionary","body":" improves names/terms"},{"title":"100+ languages","body":" multilingual speech-to-text"},{"title":"Device sync","body":" use across devices"},{"title":"Mobile companion app","body":" dictate on mobile"},{"title":"API / MCP access","body":" connect tools/workflows"},{"title":"Agent mode","body":" automate actions"},{"title":"Chat over your data","body":" ask about files/transcripts"}]}]},{"label":"Team management","cells":[{"featured":true,"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"None","emphasis":"muted"}]},{"parts":[{"text":"Basic team / workflow use"}]}]},{"label":"Support","cells":[{"featured":true,"parts":[{"text":"Community support"}]},{"parts":[{"text":"Email support"}]},{"parts":[{"text":"Priority support"}]}]}],"enterprise":{"paragraphs":["Need SSO, compliance features, admin controls, or dedicated support for a larger team?","Contact us about Enterprise."],"link":{"label":"Talk to us","href":"/contact-sales"}}},"about":{"label":"Why we built it","title":"The sentence belongs in the document you already opened.","paragraphs":["OpenWhispr started with a simple objection: getting your own sentences into the document faster should not mean moving a transcript between apps.","On desktop, its system-wide hotkey uses a focused editable cursor. That is useful in Google Docs without being a Docs add-on.","And if anything breaks, you talk to the people who built it, in Discord."],"image":{"label":"One of OpenWhispr's founders"}},"finalCta":{"tagline":"Free · open source · local-first","title":"Open the doc. Press the hotkey. Start the paragraph.","body":"Download OpenWhispr for desktop and turn speech into editable text at a focused cursor. It is system-wide dictation, not a Google Docs add-on.","cta":{"label":"Download for free","href":"/download"},"finePrint":"Free to download on Mac, Windows and Linux."}}},"offerLanding":{"title":"你的创作者优惠","description":"领取 OpenWhispr Pro 前 3 个月八折优惠。","hero":"领取 3 个月八折优惠","nav":"领取折扣","bottom":"领取优惠","bottomTitle":"领取创作者优惠","offer":"前 3 个月享受八折优惠","inlinePrice":"前 3 个月每月 {discounted}，之后每月 {regular}。","priceUnavailable":"价格暂时不可用。"},"offerAccount":{"title":"创建账号或登录","priceReminder":"前 3 个月每月 {discounted}，<renewal>之后每月 {regular}。</renewal>","verificationMessage":"点击邮件中的验证链接，然后返回此处继续领取创作者优惠。","processing":"正在使用你的账号继续…","backToOffer":"返回优惠"},"offerCheck":{"title":{"checking":"正在核实优惠…","eligible":"你的创作者优惠","retryable":"无法确认你的优惠","ineligible":"此账号不符合条件","unavailable":"此优惠不可用","already_subscribed":"你已拥有 Pro","pending":"正在确认你的购买"},"creator":"你的创作者优惠","priceUnavailable":"价格暂时不可用。","signedIn":"当前登录账号","switchAccount":"切换账号","checkoutFailed":"无法打开结账页面。请重试，我们会再次核实优惠。","checkingBody":"正在核实此账号的优惠和付款安排。","product":"OpenWhispr Pro","plan":"个人套餐 · 按月 · {currency}","discount":"八折优惠","perMonth":"/ 月","terms":"前 3 次月付享优惠，之后为每月 {regular}。","today":"今天","trialEnds":"{date} · 试用结束","firstPayment":"首次付款","atCheckout":"结账时支付 {amount}","nextPayments":"接下来的 2 次月付","each":"每次 {amount}","renewal":"从第 4 次付款起","afterTrial":"折扣将在试用结束后开始。","tax":"价格以 {currency} 计。适用税费将在结账时显示。","opening":"正在打开安全结账页面…","retryCheckout":"重试结账","checkout":"继续结账","free":"继续使用 Free","download":"下载 OpenWhispr","retry":"重试","plans":"查看常规套餐","manage":"管理订阅","body":{"retryable":"优惠核实未完成。请重试以查看折扣价格。","ineligible":"此优惠仅适用于首次个人订阅。此账号不符合条件。","unavailable":"未找到此账号可用的创作者优惠。你仍可免费使用 OpenWhispr 或查看常规套餐。","already_subscribed":"你可以下载 OpenWhispr 并用此账号登录，或管理现有订阅。","pending":"结账已完成，正在确认访问权限。请下载 OpenWhispr 并用此账号登录。请勿重复购买。"},"returnPending":"现在即可下载。请用同一账号登录，订阅访问权限仍在核实中。请勿重复购买。","returnExpired":"此结账已过期，未产生任何扣款。你可以从优惠页面重新开始。","returnNotOwned":"此结账属于另一个账号。请用结账时使用的账号登录。","existingCheckout":"此账号已有另一个结账正在进行。请完成或关闭它，或稍后再试。","manageFailed":"无法打开订阅管理页面。请重试。"}},"now":"$undefined","timeZone":"UTC","children":"$L16"}] 17:I[89554,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"ThemeProvider"] 18:I[50103,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"GitHubStarsProvider"] 19:I[28473,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"ConsentProvider"] 1a:I[44149,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"SiteNavigation"] 1b:I[44149,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"SitePageStatus"] 1c:I[31494,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js","/_next/static/immutable/chunks/01pbao-w8op81.js"],"default"] 1d:I[44149,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"SiteFooter"] 1e:I[95338,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"default"] 1f:I[88667,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"AttributionCapture"] 20:I[24901,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndgirh2-it.js","/_next/static/immutable/chunks/0_nv14y4sgdwj.js","/_next/static/immutable/chunks/3c5c8c7dedp60.js"],"MetaPixel"] 21:I[21307,["/_next/static/immutable/chunks/1ty5ydjjcibls.js","/_next/static/immutable/chunks/0t-su5g1z5lkv.js","/_next/static/immutable/chunks/21bcxxao7anzt.js","/_next/static/immutable/chunks/2w6h9kjzbvct2.js","/_next/static/immutable/chunks/1p2ndg</article></body></html>