Anatomy of a Stable Audio prompt
Stable Audio weights tokens toward the start of the prompt. That single fact shapes the whole format: front-load what must be true, push nuance toward the end.
Layer by layer
- Genre first, one primary. "melodic house" beats "melodic house with trance and UK garage elements". Secondary influences belong later, if at all.
- BPM immediately after. Stable Audio respects explicit tempo unusually well — "126 BPM" is a coordinate, not a suggestion.
- Energy / mood words. "driving", "rain-soaked", "statuesque and slow" — concrete sensory vocabulary is strongest.
- Instrumentation in an order you mean. First listed ≈ most prominent.
- Production character. "tape-saturated", "wide stereo pads", "dry punchy drums" — the layer most prompts miss.
- Exclusions. In text form ("no vocals") or in Stable Audio's dedicated negative prompt field where available.
Stable Audio prompt examples (copy-paste)
Lo-fi hip hop · 84 BPM
Melodic techno · 126 BPM
Cinematic trailer · 140 BPM
Ambient · 72 BPM
More ready-made prompts across moods and engines: AI music prompt examples.
Prompt length and what to leave out
- Length: 2–4 dense lines beat either extreme — one-line prompts underdetermine the output; essay-length prompts dilute the front-load.
- Leave out artist names. They trigger content filters and describe less than structural vocabulary does.
- Leave out contradictory tags. "Drum and bass ambient" pulls two tempos; pick a base and borrow one texture instead.
- Iterate on one layer. If the output is close but dull, fix the production clause — not the first reroll of an unchanged prompt.
Generate a Stable Audio prompt from a reference track
The hardest part of a Stable Audio prompt is the honest numbers: tempo, rhythm density, brightness, dynamics, stereo width. Music DNA Studio measures them from any MP3, WAV, FLAC or MP4 — locally in your browser, nothing uploaded — then outputs a Stable Audio-shaped prompt (and versions for Eleven Music, Soundraw, Mubert, Beatoven and AIVA) with exclusions set. You keep the creative layer; the spec underneath is measured.
Try It On A Track You Love
Drop it in, pick Stable Audio, copy the prompt. Free, no signup, nothing leaves your device.
ANALYSE YOUR AUDIO — FREE →FAQ
What is the best format for a Stable Audio prompt? ▾
Front-loaded description: primary genre and BPM first, then energy and instrumentation, then production character. Tokens early in the prompt carry more weight.
Does Stable Audio respect BPM in prompts? ▾
Yes. Explicit tempo like "126 BPM" is one of the tokens Stable Audio honors most reliably, so put it right after the genre.
How do I get an instrumental track from Stable Audio? ▾
State it explicitly — "instrumental, no vocals" — near the front of the prompt, and again in the negative/exclusion field where the version you use supports it.
Can I write a Stable Audio prompt based on a song I like? ▾
Yes — describe the song's measurable properties (BPM, rhythm density, brightness, dynamics, mood, instruments) instead of naming the artist. Music DNA Studio analyzes any track in-browser and generates a Stable Audio-formatted prompt automatically.
Why does my prompt keep producing the same generic result? ▾
Usually a missing tempo, a missing production layer, or contradictory genre tags. Change one layer at a time — rerolling an unchanged prompt tends to produce nearly the same output.