A real-time voice transcription desktop application powered by faster-whisper (large-v3 model). Record audio with a hotkey and have it automatically transcribed and pasted at your cursor position.
- Real-time voice recording — toggle recording with a keyboard shortcut
- Automatic transcription — uses the Whisper large-v3 model via faster-whisper
- Auto-paste — transcribed text is automatically inserted at your cursor position
- GPU acceleration — automatically detects and uses CUDA when available
- Visual indicators — on-screen overlays show recording and transcription status
- Accent support — full support for accented characters (French, etc.)
- Re-inject — re-paste the last transcription with a shortcut
- Python 3.9+
- Windows (uses Windows-specific GUI overlays and hotkeys)
- A microphone
- (Optional) NVIDIA GPU with CUDA for faster transcription
-
Clone the repository:
git clone https://github.com/BerthalonLucas/whisper_app.git cd whisper_app -
Create and activate a virtual environment:
python -m venv venv venv\Scripts\activate
-
Install dependencies:
pip install -r requirements.txt
-
(Optional) Install CUDA support for GPU acceleration — install PyTorch with CUDA or ensure
ctranslate2can detect your GPU.
python app.pyOr use the provided batch file on Windows:
launch_whisper.bat| Shortcut | Action |
|---|---|
Ctrl + F9 |
Start / Stop recording |
Ctrl + Shift + F9 |
Re-paste the last transcription |
Ctrl + F10 |
Quit the application |
You can also transcribe an existing audio file from the command line:
python transcribe.py path/to/audio.wav- Press
Ctrl + F9to start recording — a red ● REC indicator appears on screen. - Press
Ctrl + F9again to stop — the audio is sent to the Whisper model for transcription. - A blue progress overlay shows transcription status.
- Once complete, the transcribed text is automatically pasted at your current cursor position.
This project is licensed under the MIT License. See the LICENSE file for details.