
Audio Studio MCP Server
Runs Stable Audio 3 locally on Apple Silicon for music and sound effects, then loops, normalizes and exports the audio.
Published by kanaworksai-huekanaworksai-hue/audio-studio-mcp
What you are opening
This MCP server handles generation, editing and audio processing on a Mac. It loads a separate Stable Audio runtime and model weights rather than supplying an unrestricted MIT model.
How to set it up
- Review the pinned install.sh and the Stable Audio model license; prepare uv and FFmpeg yourself if preferred.
- Run ./install.sh on an Apple Silicon Mac; it creates .venv and downloads the runtime and weights.
- Use the absolute .venv/bin/audio-studio-mcp path in your MCP client, then generate a short test clip.
Requirements and third-party terms
Apple Silicon M1+, Python 3.10–3.13, uv, Git and FFmpeg; 16 GB RAM recommended and roughly 6 GB model storage. Server code and Stable Audio runtime are MIT; weights use the separate Stability AI Community License. Commercial use requires checking the model card and current Stability licensing terms; T5Gemma conditioning also has Gemma terms. The installer offers the official astral.sh uv installer if uv is absent. Generation is local after downloads.
Before you add it
- Read the instructions and review any scripts before running them.
- Check tool access, required credentials and third-party network requests.
- The source license and any resource-specific terms take precedence.


