Features
100% local, on-device transcription with whisper.cpp and Metal acceleration
Push-to-talk workflow: hold to record, release to transcribe
Works in any macOS app (cursor insertion)
No network requests, accounts, or telemetry
MIT licensed, free and open source forever
Supports 99 languages via Whisper multilingual models
Customizable hotkey (default Globe key), model size, and language
Configurable spoken punctuation toggle
Lightweight menu bar app, runs as background service
Copy last dictation from menu bar
Homebrew installation with guided setup
Configurable toggle mode and max recording storage
Quantized model variants available for reduced size
Real-time or faster transcription on Apple Silicon M1 and above
Personalized voice training with 50 phrase cards
Real-time speech-to-text via web app
Chrome extension for dictation in web forms
Webex integration with AI captioning
Microsoft Teams integration with live captions (add-on)
Zoom integration with live captions (add-on)
Amazon Alexa integration for smart home control
Proprietary database of atypical speech patterns
Continuous learning as user speaks
Stand-alone web app (no install required)
Mobile app for Alexa integration
API for custom integrations (contact sales)
