Stable Audio 2.0
Stable Audio 2.0 is an earlier hosted generation model with text-to-audio and audio-to-audio workflows. Its API can transform an uploaded sample into a new composition guided by a prompt.
Model design and access
An audio-to-audio request needs an input recording and a text instruction. This is a generative transformation workflow: the input conditions a new result. It should not be treated as lossless restoration or an ordinary audio filter.Stable Audio 2.0 official model documentation
The text-generation API still lists stable-audio-2 as an available model value. It exposes sampling steps and a guidance control specific to 2.0. These controls differ from the 2.5 option on the same route.Stable Audio 2.0 and 2.5 text-to-audio API
Practical use
Practical guidance. Keep 2.0 in view for projects that already use its transformations or depend on its sampling controls. Compare generated timbre and structure with the input before replacing an original recording. Version labels help preserve an intended sound, but exact reproduction also depends on the complete generation settings.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API |
| Audio format | 44.1 kHz stereo.Stable Audio 2.0 official model documentation |
| Maximum duration | Up to three minutes.Stable Audio 2.0 official model documentation |
Related models
- Stable Audio 2.5
Stability AI · Stable Audio
A hosted music and sound-generation model in the Stable Audio family.
Launch date not confirmedAPI - Stable Audio 3.0 Large
Stability AI · Stable Audio
A hosted audio-generation model that produces stereo music and sound at 44.1 kHz.
Launch date not confirmedAPI · Provider product