The open-source models behind text-to-music and text-to-sound. Run MusicGen locally or in your pipeline.
AudioCraft bundles MusicGen (text-to-music), AudioGen (text-to-sound), and EnCodec. It’s a research release, not a product — you run it in Python or via a UI like MusicGen-web. For builders who want royalty-free, local music generation, it’s a cornerstone.
Who it's for: Developers and researchers who want to generate music/sound programmatically without API fees.
Prompt a genre and mood, get a short musical clip.
Generate sound effects and ambient beds from text.
The compression model underpinning the generators.
Run on your GPU, no per-generation cost.
For developers who want open, local music generation, AudioCraft (via MusicGen) is the reference toolkit in 2026. Non-technical users should use Suno or Udio instead.