Soundbox

•
218
•

Generate AI sound effects, music with lyrics, text-to-speech, and cloned voices locally on your GPU without subscriptions or cloud dependencies.

Soundbox is an open-source, local audio generation workstation and MCP server designed as a self-hosted alternative to proprietary platforms like ElevenLabs and Suno. Powered by open-weight models such as Stable Audio and ACE-Step running natively via audio.cpp without requiring Python, it handles text-to-speech, stem separation, music composition, and voice cloning directly on your hardware. With a multi-track studio timeline and an integrated 24-tool MCP server, developers and AI agents in Claude Code or Cursor can autonomously craft game audio passes, score background tracks, and cast multi-character voiceovers entirely offline.

Key Features

01Integrated multi-track timeline studio to arrange, mix, extend, repaint, and separate audio into stems.
02Comprehensive MCP server providing 24 specialized tools and automated sound skills for Claude Code and Cursor.
03218 GitHub stars
04Local text-to-audio, music generation, TTS, and voice cloning powered by native open-weight models without Python requirements.
05Extensive catalog featuring Stable Audio 3.0, ACE-Step 1.5, MiniMax Music 3, and over 60 community audio models.
06Completely offline and private operation with local SQLite storage, cross-platform desktop installers, and a REST HTTP API.

Use Cases

01Automating in-game sound effect generation and audio passes using autonomous AI coding agents.
02Creating custom instrumental music, looped ambient tracks, and full songs with lyrics for creative projects without subscription fees.
03Producing multi-speaker dialogue and character voice lines for games and storytelling using custom cloned voices.