Music generation
Stability AI's music and sound model, with a hosted app and open weights you can run yourself.
Stable Audio is Stability AI's audio generation line, strongest at instrumentals, loops, textures and sound design rather than vocal pop songs. The hosted app offers tracks up to three minutes, with stem separation and MIDI on higher tiers. Its distinguishing feature is openness: Stable Audio 3.0 Small, Small SFX and Medium ship downloadable weights on Hugging Face under Stability's Community License, which is free for research and for commercial use by anyone under $1 million in annual revenue. Larger companies need an enterprise license. Stability has also emphasized training on licensed data, which makes it a common pick for teams worried about provenance.
Looking to make music with AI — for free? Here are our favorite platforms that let you generate full songs, vocals, or instrumentals without paying a dime. From the names you know to those you don’t, these tools prove you don’t need a studio to sound pro.
Sound designers and SFX libraries launched the Professional Sound Alliance to push for consent and payment when their work trains AI, even as its founding libraries already license to companies like Sony AI.
