Stability AI is one of the AI companies that has steered firmly away from using commercial music without permission to train its music models.
It used production music for the first version of its Stable Audio tool, which was focused on creating short samples rather than full songs. However, this week’s Stable Audio 3.0 can generate tracks up to six minutes long.
In fact, this is a family of four models rather than just one: 3.0 Small SFX can generate sound effects ‘on-device’ – meaning it runs on smartphones and laptops. 3.0 Small can generate full-music clips up to two minutes long, on devices too.
It’s the 3.0 Medium and 3.0 Large models that can do more, boosting quality as well as the maximum track length. All bar the Large model are ‘open-weight’, which means they can be downloaded and built on by developers.
“We want to foster the same kind of community-driven innovation in audio that we sparked in image generation with the launch of Stable Diffusion,” explained the company.
Stability AI has recently signed licensing deals with Universal Music Group and Warner Music Group, and this week brings more news on an industry-relations tip. The company has appointed Ethan Kaplan as its head of audio product.
Kaplan has a long history in the digital-music space, having held executive roles at WMG, Fender and Live Nation Labs.
‘ The preceding article may include information circulated by third parties ’
‘ Some details of this article were extracted from the following source musically.com ’
