Buzz is a free, open-source audio transcription tool that brings OpenAI’s Whisper models to your Linux desktop. It runs completely offline, ensuring your files never leave your machine—crucial for privacy. Buzz processes everything locally using AI models trained on diverse languages and accents, supporting batch transcription, real-time recording, and multiple output formats.
Why Install Buzz on Linux?
Linux users value control, privacy, and efficiency—Buzz delivers all three. Key benefits:
- GPU acceleration for speed or CPU-only processing for compatibility
- Seamless integration with PulseAudio and PipeWire
- Live transcription of system audio or microphone input
- No cloud dependencies or monthly fees
- Offline processing for complete privacy
System Requirements
Before installation, ensure your system meets these requirements:
- Python: 3.8 or higher
- FFmpeg: For audio processing
- RAM: 1GB (tiny model) to 10GB+ (large model)
- GPU: CUDA (NVIDIA) or ROCm (AMD) for acceleration (optional)
- Storage: ~500MB for application and models
- Recommended: 8GB RAM, dual-core processor for medium models
Install Dependencies on Ubuntu
Ubuntu and Debian-based distributions use APT for dependency installation:
sudo apt update
sudo apt install python3 python3-pip ffmpeg portaudio19-dev
Verification: Run python3 --version and ffmpeg -version to confirm installation.
Install Dependencies on Fedora
Fedora uses DNF with slightly different package names:
sudo dnf install python3 python3-pip ffmpeg portaudio-devel
Verification: Check versions with python3 --version and ffmpeg -version.
Install Dependencies on Arch Linux
Arch Linux users benefit from current packages through pacman:
sudo pacman -S python python-pip ffmpeg portaudio
Note: Arch uses python instead of python3 as the default command.
Install Whisper and Buzz via Pip
The universal installation method across all distributions uses pip:
pip3 install openai-whisper
pip3 install buzz-captions
Configure PATH
If pip shows a PATH warning, add ~/.local/bin to your PATH:
export PATH="$HOME/.local/bin:$PATH"
source ~/.bashrc
Launch Buzz: Run buzz in your terminal.
Install Buzz Using Flatpak
Flatpak provides distribution-independent installation with automatic dependency management:
flatpak remote-add --if-not-exists flathub https://flathub.org/repo/flathub.flatpakrepo
flatpak install flathub com.github.chidiwilliams.buzz
Launch: Run flatpak run com.github.chidiwilliams.buzz or use your application menu.
Configure Buzz for Live Transcription
After launching Buzz, configure these options:
- Select your microphone from the input device dropdown
- Choose a Whisper model based on hardware:
- Tiny/Base: Older machines
- Small/Medium: Balance accuracy with performance
- Large: Best accuracy, requires substantial RAM
Command-Line Usage
buzz --model medium --device cuda
Options:
--device cuda: GPU acceleration (NVIDIA)--device cpu: CPU-only processing--language en: Set language explicitly
Output formats: TXT, SRT, VTT for subtitling.
Troubleshooting Common Errors
Dependency Conflicts
If pip reports version incompatibilities, create a virtual environment:
python3 -m venv buzz-env
source buzz-env/bin/activate
pip install openai-whisper buzz-captions
Microphone Access Errors
Check audio group membership:
groups | grep audio
sudo usermod -aG audio $USER
For PipeWire systems:
systemctl --user status pipewire-pulse
GPU Acceleration Issues
NVIDIA GPU users:
- Install CUDA toolkit and cuDNN libraries
- Verify with
nvidia-smi - Install PyTorch with CUDA support:
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
AMD GPU users: Install rocm-hip-sdk, then reinstall PyTorch with ROCm support.
Verify GPU recognition:
python3 -c "import torch; print(torch.cuda.is_available())"
Should return True. If issues persist, fall back to CPU processing.
Conclusion
You now have a powerful offline transcription system on your Linux machine. Buzz converts audio into accurate text without compromising privacy or requiring subscription fees.