Careless Whisper turns your voice into text anywhere you can paste. Powered by OpenAI Whisper, running entirely on your machine. No accounts. No subscriptions. No data leaves your computer.
Free and open source. Latest · All releases
How it works
Default is Ctrl+Shift+Space. Works in any app, anywhere, always.
A compact overlay shows the waveform and timer. Press the hotkey again when you're done.
The transcription is pasted wherever your cursor is. Whisper handles 18 languages out of the box.
Why it's different
Every sample stays on your device. No cloud API, no telemetry, no accounts.
Metal GPU acceleration on Apple Silicon. Seconds for short clips.
English, Hebrew, Spanish, French, German, Japanese, Chinese, Arabic, and more — switch from the tray.
Lives in your macOS menu bar or Windows/Linux system tray. No dock clutter, no extra window.
Drop in an MP3, WAV, M4A, MP4, FLAC, or OGG — get back a clean transcript.
Free forever. Fork it, ship it, sell it — we only ask that you keep it open.
Install
Most people want the installer. Developers can build from source in under a minute.
Grab the installer for your OS, run it, grant Microphone + Accessibility (macOS) when prompted, press the hotkey.
Latest release →Requires Rust, Node 18+, and pnpm.
$ git clone https://github.com/YarivGilad/careless-whisper $ cd careless-whisper $ pnpm install $ pnpm tauri dev
Built together
Every commit, bug report, and dumb question made this better. You're welcome here.
