Meta just announced the launch of AudioCraft, an AI tool set to revolutionize audio and music generation by harnessing the power of text prompts. And guess what? It's open-source too (i.e. publicly accessible)!
AudioCraft seeks to simplify the complexities behind generative frameworks of sound creation, providing a streamlined foundation for high-fidelity audio and music track production.
It is based on three AI models, as detailed below:
- MusicGen: Designed to produce music from text prompts. It can create short pieces or even complete a melody it hears.
- AudioGen: Generates audio samples also based on descriptive text captions. using public sound effects.
- EnCodec: Allows for even higher quality music generation via extracting or compressing audio samples.
With Meta sharing the code behind these models and releasing it to the public, we expect further refinement in the long run so that AudioCraft can successfully cater to both music production newbies and seasoned professionals in the industry.
Guess Meta's going to be your new fave synthesizer for some groovy machine-made music!
Let the creative frenzy begin and delve deeper for more details over here.