Meta AudioCraft Launches – AI Generates Audio And Music Using Text


 Meta launched their open source artificial intelligence (AI) AudioCraft yesterday. It can generate text and music only through user input. According to Meta, AudioCraft content producers who want to find audio or music can generate it themselves without worrying about intellectual rights.



Audiocraft combines the previously launched MusicGen, AudioGen and EnCodec models. All models are trained using audio that has been licensed or produced by Meta itself. A total of 400,000 recordings with metadata equivalent to 20,000 hours were used to train the model.


It can generate background music for video backgrounds and even audio special effects such as traffic, barking dogs or people walking. Audiocraft was launched as open source to allow researchers to generate audio and music using their own data.

Previous Post Next Post

Contact Form