What it does
Stable Audio 3.0 is a family of audio generation models from Stability AI, trained on fully licensed data. It covers sound effects through to full musical compositions, and the page says tracks can run up to six minutes with complex, changing structure. Generation is meant as the start of the process. The models let you modify a segment of a track, rework part of a song or extend a composition, and they can be customised on your own library. The family has Large, Medium and Small versions, with Small and a Small SFX variant aimed at mobile devices. Medium and Small come with downloadable open weights. You can try it in a web app, call it through the Stability AI API, deploy it on your own infrastructure under an Enterprise license, or download the open weights. The page says the Enterprise license includes legal indemnification.
Capabilities
- Models for sound effects and full musical compositions
- Full tracks up to six minutes long
- Segment edits, rework and extend options
- Customisation on your own audio library
- Open weights for the Small and Medium models
- Web app, API and enterprise licence options
When to use it
- You want to generate music or sound effects from text prompts
- You want to rework or extend part of an existing track
- You are a developer who wants open weights to experiment with or build on
- Your organisation needs models trained on licensed data that can run on its own infrastructure
