OpenWhispr

OpenWhispr is an open source voice-to-text assistant that turns speech into text in any app, including when offline. Its site describes it as private by design and highlights dictation, AI meeting notes, AI chat that knows meetings, and local transcription of audio files.

The product is built around private, local use. It says transcription history lives in a local database on the user’s machine, that local AI models run directly on the user’s hardware, and that no audio is sent anywhere when using local models. For cloud processing, it states zero data retention: audio passes through for transcription and is not stored, logged, or copied. The page also says it never trains on user transcriptions without explicit consent. Other listed features include AI cleanup, voice instructions such as drafting an email or cleaning up notes, auto-learning from corrections, and support for 100+ languages with auto-detection and switching mid-conversation.

OpenWhispr says it works across all apps that accept text input, and the page names ChatGPT, Claude, Cursor, Google Docs, Gmail, Slack, Teams, Superhuman, Notes, Grammarly, iMessage, Mail, and X as examples. It is available on macOS, Windows, and Linux. The page also shows an app interface with sections for notes, upload, dictionary, integrations, and settings.

Use cases named on the site include medical dictation for clinicians and legal dictation for lawyers, along with Mac dictation and Windows voice-to-text. Pricing is shown as free, with the page stating that dictation, AI cleanup, offline mode, and 100+ languages are all free. The code is public on GitHub for inspection, audit, and verification.