Stable Audio 2.0 AI music generator
Stable Audio 2.0 generates instrumental music up to 47 seconds, as WAV or MP3, with 15, 30, and 45 second lengths. See the credit table below.
Best for
Stable Audio 2.0 is an instrumental AI music generator for beds, stings, and short cues. Lengths are 15 seconds, 30 seconds, and 45 seconds, with a hard ceiling of 47 seconds. Exports are WAV or MP3. It does not sing lyrics. Vocal songs are MiniMax Music.
Where it fits
Use it under a montage, a logo, or a product loop. Ask for tempo, instruments, and mood in one sentence: a lo-fi beat with vinyl crackle, or a five-second trailer hit with a sub boom. Stay inside the 47 second cap. A longer request is clamped.
How to use it
Open the music generator and choose Stable Audio 2.0. Pick the length closest to the picture, choose WAV if you will mix it or MP3 if you will drop it straight into an edit, and generate. The credit table shows 30 and 47 seconds on the Free plan.
Controls and pricing
Credits follow the length, up to that 47 second ceiling. MiniMax Music is a different purchase: one flat price for a full vocal track of about two to four minutes. If you only need a sting, stay on Stable Audio.
Good to know
Name the instruments and the tempo before you name the mood. A 90 BPM Rhodes loop is easier to cut than a request for something cinematic. Fifteen seconds is a sting, thirty seconds covers a short scene, and forty-five seconds is the longest bed this model will give you.
Also worth knowing
If the cue must hit a picture cut, say so. A swell into a final chord is a different 15 seconds than a loop that could run forever. Stable Audio will not extend past 47 seconds, so design the ending inside that window.
Credits and controls
- 30 seconds
- 30 credits
- 47 seconds
- 47 credits
- Length
- 15s, 30s, 45s (Max 47s)
Example prompts

Example prompt
Lo-fi hip hop beat, dusty vinyl crackle, mellow Rhodes chords, 80 BPM, rainy night study session

Example prompt
Epic cinematic trailer hit with deep sub boom, rising risers and brass stabs, 5 seconds
Example prompt
Cyberpunk synthwave bassline with punchy 80s drums and driving retro lead arpeggios
Example prompt
Dark synth-orchestral luxury ambience, slow pulse, warm strings swell, 70 BPM, 15 seconds
Questions
Can Stable Audio sing lyrics?
No. It is instrumental. Vocal tracks are MiniMax Music.
What is the longest track?
47 seconds. The length list shows 15s, 30s, and 45s, marked as a 47 second maximum.
Which formats can I export?
WAV and MP3.
Does a longer cue cost more?
Yes, up to the 47 second ceiling. The table lists 30 and 47 seconds.
Can I make sound effects here?
Use the sound-effects tool for one-shot effects and ambience.