Audiocraft Audio Generation

Meta's AudioCraft library wrapped for Claude, giving you MusicGen for text-to-music and AudioGen for sound effects. You get models from 300M to 3.3B parameters, melody conditioning if you want to hum a tune and have it generate around that, and proper stereo output. Generation params are straightforward: set duration, temperature, and guidance strength. The HuggingFace integration works well if you're already in that ecosystem. Honest take: this is solid for prototyping music apps or generating background audio, but you're capped at 30 seconds and the quality ceiling is what it is. If you need longer commercial tracks, look at Stable Audio instead.

Install to Claude Code

npx -y skills add orchestra-research/ai-research-skills --skill audiocraft-audio-generation --agent claude-code