Meta Open-Sourced AudioCraft, a Generative AI Software for Sound and Music

IBL News | New York

Meta this week shared details on AudioCraft, a single open-source code base for music, sound, compression & generation of high-quality, realistic audio from text.

AudioCraft consists of three models: MusicGen, AudioGen, and EnCodec.

• MusicGen, trained with Meta-owned and specifically licensed music, generates music from text-based user inputs. [See Demo]

• AudioGen, which was trained on public sound effects, generates audio from text-based user inputs.

• EnCodec decoder allows for higher quality music generation with fewer artifacts.

Meta’s pre-trained AudioGen model lets users generate environmental sounds and sound effects like a dog barking, cars honking, or footsteps on a wooden floor.

The models are available for research purposes and to further people’s understanding of the technology.

“We’re excited to give researchers and practitioners access so they can train their own models with their own datasets for the first time and help advance the state of the art,” said Meta in a blog post.

Samples of generated audio.