Buzz is a desktop application that transcribes and translates audio offline on your personal computer using OpenAI's Whisper model.
The tool solves the problem of transcribing audio without sending data to cloud services, keeping your recordings private and avoiding network dependencies. It works by running Whisper locally on your machine, with support for multiple backend implementations and hardware acceleration options including CUDA for Nvidia GPUs, Vulkan for most GPUs including integrated ones, and native Apple Silicon support. The application handles both file-based transcription of audio and video files as well as live real-time transcription from a microphone.
Buzz suits developers and users who need offline transcription with privacy guarantees and want to avoid API costs or latency. The tool is particularly valuable for processing noisy audio through speech separation, identifying speakers in transcribed media, and automating workflows via its command-line interface and plugin system. It exports transcripts in multiple formats including TXT, SRT, and VTT, and includes a presentation window for accessibility during live events. The application is available across macOS, Windows, and Linux through various distribution channels including Flatpak, Snap, and Appimage, as well as through PyPI for Python environments.
The maintainers respond to issues and pull requests within a day. Work in the issue tracker centers on enhancement requests, bug reports, and information-gathering on reported issues.
