Stable Audio

Stable Audio is Stability AI's text-to-audio tool for making original music and sound effects. You describe a sound in plain words — a mood, a genre, instruments, a scene — and it creates a stereo track in seconds. It suits video makers, game developers, and podcasters who need music they can trust in their projects. All models train on fully licensed data, so the output is safe to publish in client work. Pricing details change often, so check the official site for current plans.

Key Features

1. Text-to-audio generation

Type a plain-language prompt about genre, mood, instruments, or tempo, and get a finished 44.1 kHz stereo track within seconds.

2. Audio-to-audio transformation

Upload your own recording and let the model restyle, extend, or reimagine it while keeping the feel of your original clip.

3. Inpainting

Select part of a track and have the model rebuild it. The new section blends with the rest of your audio.

4. Long-form compositions

The current 3.0 model writes instrumental tracks up to 6 minutes 20 seconds long, with clear sections like intros and outros.

5. Experimental stem export

Split a generated track into separate stems for proper mixing and arrangement in your own studio software.

6. DAW plugin

Use Stable Audio 3.0 inside Ableton, Logic, and Pro Tools. Make samples and song starters without leaving your session.

7. Commercially safe licensing

Training data is fully licensed, and paid plans carry a Creator license for commercial projects and music releases.

Frequently Asked Questions (FAQ)

What is stem export in Stable Audio?

It is an experimental feature that splits a generated track into separate layers, or stems. Import these layers into a DAW and mix each one like a real instrument.

Can I use Stable Audio tracks in commercial projects?

Paid plans include a Creator license that covers commercial use and music releases. The free tier's Personal license covers non-commercial projects only.

Does Stable Audio generate vocals or lyrics?

No. Stable Audio 3.0 is instrumental only. It creates music and sound effects, not sung vocals or spoken lyrics, which avoids voice-cloning concerns.

How long can the generated tracks be?

The 3.0 model supports tracks up to 6 minutes 20 seconds. These are structured compositions with distinct sections, not simple loops.

Can I run Stable Audio inside my own software?

Yes. There is a plugin for Ableton, Logic, and Pro Tools, plus an API for developers and open weights for self-hosting under a Community License.

How does audio-to-audio work?

You upload a clip with a text prompt, then adjust sliders for how much the source audio shapes the result. Uploads are scanned for copyright matches first.

Conclusion

Stable Audio is one of the most practical AI music tools for published work. It makes full-length instrumental tracks in seconds, plugs into major DAWs like Ableton and Logic, and trains only on licensed data you can trust. The free tier covers non-commercial tests, while paid plans from about $12 a month add commercial licensing for releases.