MusicGen (AudioCraft)
A text-to-music model in the MIT-licensed AudioCraft library, with melody guidance and code for generating music and training models.
MusicGen is Meta AI's model for generating music from written descriptions. It's part of AudioCraft, a PyTorch library for developers and audio researchers building music generation tools in their own computing environment.
Text prompts describe the music to generate, and melody conditioning lets a musical reference guide the result. AudioCraft includes MusicGen inference and training code, plus reusable components and training pipelines for audio research. MusicGen uses the EnCodec audio codec to turn its generated audio representation into a waveform. AudioCraft's code is open source under the MIT license.
MusicGen is distributed through Meta’s AudioCraft library and downloadable MusicGen checkpoints. AudioCraft code uses MIT, while the released model weights use CC BY-NC 4.0. The library’s software version and the chosen MusicGen checkpoint are separate.
Logotyper
