According to a blog post by Meta, the parent company of Facebook and Instagram, on August 2nd, it has launched a series of productive artificial intelligence (AI) models called AudioCraft to create music using various sources. The package of productive AI tools includes MusicGen and AudioGen, which work with text-based inputs to create new sounds, as well as another tool called EnCodec, which enables “higher quality music production with less artificiality.”
In the announcement, Meta stated that the MusicGen model is trained with its own or “specially licensed” music.
This comes amidst significant debates surrounding the training of artificial intelligence with copyrighted works in various artistic fields, including a lawsuit filed against Meta due to copyright infringement during AI training.
Meta has made MusicGen and AudioGen available for use by the research community and developers in various dimensions. As these projects evolve, Meta anticipates that the models will be useful for both amateurs and professionals in the music industry.
Harvey Mason Jr., CEO of the Recording Academy, recently likened the emergence of music generated by artificial intelligence to the early days of studios entering the music scene in an interview.
Meta’s release of productive AI music tools comes shortly after Google launched similar tools called MusicLM.
In May, the company announced that it was accepting “beta users” for its products through the AI Test Kitchen platform.
Meta, along with many other technology giants including Google and Microsoft, is actively releasing new AI tools as part of the race to develop and deploy the most powerful models.
On August 1st, Meta announced the launch of new AI chatbots with personalities that users on their platforms can interact with as search assistants and fun products.