Voice Dictation That Never Leaves Your Device
When you're dictating patient notes, legal briefs, or confidential case details, privacy isn't optional. OpenWhispr processes everything locally — no cloud, no data retention, no third-party access. Ever.
View on GitHubThe Privacy Problem with Dictation Software
Most dictation tools — including popular options like Dragon Medical, Google Speech-to-Text, and Apple Dictation — send your audio to cloud servers for processing. For general use, that's fine. But when you're dictating patient diagnoses, medication details, privileged attorney-client communications, or witness statements, that audio contains some of the most sensitive information in existence.
Cloud-based dictation creates a chain of custody problem. Your audio travels across networks, gets processed on someone else's servers, and is subject to their data retention and security policies. Even with HIPAA Business Associate Agreements or legal confidentiality provisions, you're trusting a third party with data you're ethically and legally obligated to protect.
OpenWhispr eliminates this entirely. Every word of audio is processed locally on your device using OpenAI Whisper and NVIDIA Parakeet. Nothing leaves the machine. No network requests, no cloud processing, no data retention by third parties.
How OpenWhispr Protects Sensitive Information
100% Local Processing
All transcription runs on your machine through OpenAI Whisper and NVIDIA Parakeet. Audio data never touches a network. You can verify this yourself — disconnect from the internet and OpenWhispr works exactly the same. No cloud fallback, no "optional" data sharing.
Multiple Speech Models for Accuracy
Choose from OpenAI Whisper and NVIDIA Parakeet models in various sizes. For clinical notes and legal documents where accuracy is critical, the larger models deliver high-quality transcription. For quick notes, smaller models offer faster processing. All running locally.
Custom System Prompts for Formatting
Agent mode transforms raw dictation into structured output. Create system prompts for SOAP notes, intake summaries, deposition outlines, contract clause drafts, or any format your practice requires. Speak naturally, get formatted professional documents.
Open Source — Fully Auditable
The entire codebase is public on GitHub. Your IT security team, compliance officer, or any third-party auditor can inspect exactly what the software does. No black boxes, no "trust us" — verify it yourself.
Real-World Professional Scenarios
Clinical Documentation
After a patient visit, you need to document the encounter in your EHR. Hold the OpenWhispr hotkey and dictate: "45-year- old male presenting with acute lower back pain, onset two days ago, no radiating symptoms, negative straight leg raise..." Agent mode can format this into a structured SOAP note. All processed locally — the patient's PHI never leaves your workstation.
Legal Documentation
Reviewing case files late at night, you need to draft a motion summary. Dictate your analysis of the key arguments, relevant precedents, and proposed remedies. Agent mode structures it into a formal legal document format. Attorney- client privileged information stays entirely on your machine — no cloud processing, no third-party data access.
Field Notes and Intake Documentation
Social workers, paralegals, and intake coordinators often need to document conversations immediately after they happen. With OpenWhispr, dictate detailed notes right after a client meeting while everything is fresh. The push-to-talk hotkey works in any app — your case management system, Word, or wherever you keep records.
Privacy Architecture — How It Works
What OpenWhispr Does
- Processes audio entirely on your local device
- Uses OpenAI Whisper and NVIDIA Parakeet — no cloud API calls
- Stores transcription history locally only
- Works completely offline
- Open source — auditable by anyone
What OpenWhispr Never Does
- Send audio to any external server
- Transmit transcriptions over the network
- Store data in any cloud service
- Share data with third parties
- Require an internet connection to function
Important note: OpenWhispr's agent mode (AI text cleanup) optionally uses cloud LLM APIs if configured. For maximum privacy, use OpenWhispr in local-only mode — all core transcription works without any internet connection. The agent mode is entirely optional and can be configured with your own API keys.
Dictation You Can Trust with Sensitive Data
100% local processing. Open source. Auditable. No cloud, no data retention, no compromises on confidentiality.
No account required · Works offline · Open source forever