What is Buzz?
Buzz is an open-source application for transcribing and translating audio locally on your personal computer. It leverages OpenAI's Whisper models to provide automatic speech recognition (ASR) and optional translation, while keeping audio processing offline to preserve user privacy.
Key features
- Offline transcription and translation using Whisper models.
- Desktop GUI for macOS, Windows and Linux and a command-line interface via the PyPI package.
- Multiple installation options: native installers (dmg for macOS, installer for Windows via SourceForge), winget (Windows), Flatpak and Snap (Linux), and pip (buzz-captions).
- GPU support via PyTorch and CUDA for faster transcription on compatible Nvidia GPUs (the project documents specific torch/CUDA wheel versions and additional NVIDIA packages).
- MIT open-source license, active CI and test coverage badges, and regular releases on GitHub.
- Focus on privacy: audio is processed locally, not uploaded to a cloud service by default.
Installation & usage
- macOS: download the .dmg from the project releases (SourceForge link provided in docs).
- Windows: installer available via SourceForge; also installable with winget (package ChidiWilliams.Buzz).
- Linux: available as Flatpak (Flathub) or Snap; Snap requires a few system packages and snap connections as documented.
- PyPI: install with pip (requires ffmpeg and Python 3.12), then run via
python -m buzz. - GPU: instructions in the repository show how to install matching torch/torchaudio wheels and NVIDIA runtime/cu libraries for CUDA acceleration.
Use cases
- Transcribing interviews, podcasts, lectures, or meeting recordings locally.
- Real-time or batch subtitle generation and translation for videos.
- Privacy-sensitive transcription where cloud upload is undesirable.
Community & project status
- The GitHub project is popular (noted star count in repository metadata) and includes documentation site, screenshots, CI, and code coverage integrations.
- Open-source contributions are welcome; the project follows an MIT license and publishes releases.
Links
See the project's documentation site (official_url) and the GitHub repository page for installation details, usage examples, and contribution instructions.
