GitHub - YarivGilad/careless-whisper

A lightweight, always-on desktop app for local voice-to-text transcription. Lives in the system tray / menu bar, records on a global hotkey, transcribes locally with Whisper, and pastes the result into your focused app. No cloud. No data leaves your machine.

Supports macOS and Windows.

Website: yarivgilad.github.io/careless-whisper

Download

Get the latest version from the Releases page:

Install

macOS

1. Download the .dmg file above.
2. Open it and drag Careless Whisper to your Applications folder.
3. Launch from Applications (or Spotlight).

The app has no Dock icon — it lives in the menu bar (top-right of your screen).

"Careless Whisper is damaged and can't be opened"

Don't worry — the app is perfectly fine! macOS shows this warning for apps that aren't code-signed with Apple's $99/year Developer certificate. This is standard for open-source projects that are trying to be given away for free and avoid the Apple penalty for creative generosities. Until this project gets funded (don't hold your breath — it's a weekend side project), macOS users are welcome to run this one-time fix in Terminal:

If you dragged the app to Applications:

xattr -cr "/Applications/Careless Whisper.app"

If you're running it straight from the DMG:

xattr -cr "/Volumes/Careless Whisper/Careless Whisper.app"

After that, the app will open normally.

Windows

1. Download the installer from the Releases page.
2. Run the installer and follow the prompts.

The app lives in the system tray (bottom-right of your screen).

First launch

The Settings window will open automatically because no model is downloaded yet.

1. Pick a model and click Download (the base model is a good starting point — ~142 MB, fast).
2. Wait for the download to finish.
3. Your OS will ask for Microphone access the first time you record — allow it.
4. macOS only: Go to System Settings → Privacy & Security → Accessibility and enable Careless Whisper so it can paste text into other apps.

Record and transcribe

1. Click into any text field in any app (your target).
2. Press the hotkey (default: Cmd+Shift+Space on macOS, Ctrl+Shift+Space on Windows) — a small recording indicator appears.
3. Speak.
4. Press the hotkey again to stop — the transcribed text is pasted directly where your cursor was.

The hotkey, language, and other options can be changed from Settings in the tray menu.

Default Hotkey

Cmd+Shift+Space (macOS) / Ctrl+Shift+Space (Windows) — press to start recording, press again to stop, transcribe, and paste.

Whisper Models

On first launch the app will prompt you to download a model. Models are stored locally on your machine.

Permissions

macOS

• Microphone — to record your voice
• Accessibility — to paste transcribed text into other apps (System Settings → Privacy & Security → Accessibility)

Windows

• No special permissions needed.

Building from Source

pnpm install
pnpm tauri dev

pnpm tauri dev -- --no-default-features

src-tauri/src/
├── audio/          # Mic capture + resampling
├── transcribe/     # whisper-rs wrapper
├── hotkey/         # Global hotkey registration
├── output/         # Clipboard + paste simulation
├── models/         # Model download & management
├── config/         # Settings persistence
├── tray.rs         # System tray setup
└── commands.rs     # Tauri IPC handlers

src/
├── components/     # Overlay, Settings, ModelManager
├── hooks/          # Tauri event subscriptions
└── styles/         # CSS