Stability AI has introduced Stable Audio Open, a new AI-driven model designed to create music and sound effects from text prompts. This open-source model can generate audio clips up to 47 seconds long, utilizing a dataset of around 486,000 royalty-free samples from FreeSound and the Free Music Archive.
Capabilities and Applications
The model can produce various audio elements including drum patterns, instrumental riffs, ambient sounds, and production features suitable for multimedia projects. Users have the ability to customize the model with their own audio samples to craft personalized soundscapes. Musicians, for instance, can input their recordings to generate distinctive beats or riffs.
Although versatile, Stable Audio Open has certain constraints. It isn't designed to create entire songs, melodies, or vocal tracks. For these requirements, users are advised to use Stability AI's premium service, Stable Audio. Furthermore, the model's terms limit commercial use and its performance varies across different music genres and languages, reflecting the lack of diversity in its training data.
Corporate Context and Industry Implications
Stability AI has been in the spotlight due to controversies, notably after the departure of CEO Emad Mostque. He resigned over disputes concerning the use of copyrighted works for AI training. The company has also reportedly been struggling financially nd is seeking a buyer.
Stable Audio Open appears to be part of an effort to shift attention and underscore Stability AI's subscription services. The rise of AI-generated music is raising copyright questions. Sony Music has warned AI firms about unauthorized use of its content, and Tennessee has enacted the first U.S. law targeting AI misuse in music.
Stability AI claims training its models on copyrighted music is permissible under “fair use,” a topic of ongoing debate. In April, over 200 artists issued an open letter condemning AI's role in music. Signed by names like Stevie Wonder, Billie Eilish, Perl Jam, and Jon Bon Jovi, the letter labeled AI as a threat to human creativity, predicting dire consequences for real artists.
