Skip to content
Get packs

AudioCraft Audio Generation

Generate music and sound effects from text descriptions using Meta's AudioCraft models

Source Download skill
Repository
Orchestra-Research/AI-Research-SKILLs
Stars
14,195
Licence
MIT
Files
3 files

What it does

Once installed, the agent can write and run Python code that uses Meta's AudioCraft library (MusicGen for music, AudioGen for sound effects) to turn a text description such as 'chill lo-fi hip hop beat' into an audio file. It also covers melody-conditioned music, stereo output and using the same models through HuggingFace Transformers. The skill is a reference guide that teaches the agent the correct setup, model choices and generation settings.

Great for

  • Musicians or content creators who want placeholder or background music generated from a written description
  • Developers building a music or sound-effect generation feature in an app
  • Game or video makers who need quick ambient sound effects such as a dog barking in a park

Skip it if

  • Commercial releases needing clearly licensed output: the skill points to Stable Audio for longer commercial music and does not address usage rights for generated audio
  • Spoken-word or lyric-based vocals: the skill positions Bark and Jukebox as alternatives for those needs
  • Users without a GPU or Python environment, since the models are large (300M to 3.3B parameters) and the setup requires installing PyTorch and related packages

Try asking

  • “Generate a 30-second epic orchestral soundtrack with strings and brass using MusicGen”
  • “Create a short sound effect of rain on a tin roof with AudioGen”
  • “Make a folk guitar track that follows the melody of my melody.wav file”
  • “Set up a HuggingFace Transformers pipeline that generates 80s pop music in stereo”
Step-by-step guide: how to use AudioCraft Audio GenerationInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 3 original files, the MIT licence and an attribution note.

    audiocraft.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip audiocraft.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip audiocraft.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Be specific in descriptions: name genre, instruments, tempo and mood, since the examples show more detailed prompts produce more targeted results
  • Start with musicgen-small or audiogen-medium to test quickly, then move to larger models for quality
  • Check the bundled troubleshooting reference for setup and GPU memory problems before debugging from scratch

What's inside

  • references/advanced-usage.md
  • references/troubleshooting.md
  • SKILL.md

Redistributed unmodified from Orchestra-Research/AI-Research-SKILLs under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.