Push-to-talk dictation for macOS that runs entirely on your Mac. Hold a key, speak, release — transcribed on-device and inserted wherever your cursor is.
Download for macOSNo accounts, no API keys, no network at inference. Your voice never leaves the Mac.
Hold a modifier key, speak, release. Text pastes at your cursor in any app.
WhisperKit (Core ML) or Apple's built-in on-device speech engine — your choice.
Tidy filler words and punctuation with Apple Foundation Models, Ollama, or MLX — with a fully editable prompt.
A floating waveform shows your words appearing as you speak, then “Transcribing…”.
MIT-licensed and open source. Build it, fork it, tweak every prompt and default.
Everything — recording, transcription, and cleanup — happens on your Mac. There are no analytics and no servers. The only network use is a one-time model download, or a local LLM you run yourself.
Requires macOS 26 (Tahoe) or later on Apple Silicon. First dictation downloads the speech model (~600 MB), cached for next time.