How Suno takes direction
Suno is a two-part brief: a style field describing the sound, and a lyrics field carrying the words. A third field, Exclude Styles, subtracts things. Two sliders — style influence and weirdness — decide how literally and how adventurously the model reads your tags. Treat them as four separate dials. Folding them into one long paragraph is the most common reason output sounds like everything else on the platform.
Layer by layer
- Genre and vocal type first. Early tags pull hardest, so put the two things that decide the song at the front. Four to eight tags beats a paragraph.
- Tempo and key as numbers. "104 BPM" and "F# minor" are instructions; "mid-tempo" is a mood. If you have measurements, use them.
- Tag every section. [Intro], [Verse], [Pre-Chorus], [Chorus], [Bridge], [Outro]. Structure is recruited from the tags rather than guessed, and the hook lands where you put it.
- Name the voice you want. "Female lead vocal, airy delivery" or "male baritone, close-miked". Leave it blank and you get the default voice, which is the tell most people hear first.
- Exclude artefacts, not feelings. "EDM drops, autotune, distorted 808s" is useful. "Sad" is not something a negative field can remove.
- Read the sliders as dials. Style influence around 40–60% keeps a result natural, 70%+ forces the tags through. Weirdness under 30% stays usable; over 50% is for experiments.
Suno setups you can paste
Paste the style block, the lyrics block and the exclude line into the matching fields, then set the sliders.
Indie pop · F# minor · 104 BPM
Lo-fi study beat · A minor · 82 BPM
Cinematic trailer cue · D minor · 140 BPM
More ready-made prompts across moods and engines: AI music prompt examples.
What to leave out
- Artist and band names. They describe less than genre plus instrumentation plus vocal type, and they borrow someone else's rights problem.
- One long sentence in the style box. Bury your genre at word 40 and it may not survive the weighting.
- Lyrics with no section tags. Without [Chorus] the model has no map of where the hook belongs.
- Contradictions. "Aggressive lullaby at 60 BPM double-time" asks for two songs. If you want contrast, build it as a stripped [Bridge] instead.
- A reference upload. Do not feed it a commercial recording. Measure the recording instead and describe the numbers.
The rule underneath all of them: describe the music, never the record. Reference an artist or a track and you inherit someone else's rights problem along with a vaguer brief.
Build a Suno brief from a reference track
The hard part is the honest numbers: tempo, key, brightness, dynamics, and the instruments actually carrying the groove. Music DNA Studio measures them from any MP3, WAV, FLAC or MP4 — locally in your browser, nothing uploaded — then outputs a Suno-shaped prompt alongside versions for the other ten engines. You make the creative calls; the spec underneath is measured, not guessed.
A common use: a track you like the feel of, measured into tempo, key, brightness and instrumentation, then written into the style box as tags rather than a name.
Guides for the other engines
Eleven Music · Suno · Udio · Mureka · MiniMax Music · YuE2 · Stable Audio · Soundraw · Mubert · Beatoven · AIVA
Build The Suno Brief
Drop in a reference, pick Suno, copy the style, lyrics and exclude blocks. Free, no signup, nothing leaves your device.
ANALYSE YOUR AUDIO — FREE →FAQ
What is the difference between style influence and weirdness? ▾
Style influence is how literally the tags are followed; weirdness is how far the result departs from the obvious answer. Raise style influence when output ignores your tags. Raise weirdness only when a take sounds too safe and you want an experiment — it is a creative dial, not a quality dial.
Do I have to write my own lyrics? ▾
No. Leave the field empty and Suno writes them, or paste a scaffold with section tags and replace the placeholder lines. If you do supply words, tag the sections — that is what maps the hook to the chorus.
Why does my Suno track sound generic? ▾
Three usual causes: the default vocal because none was named, no tempo or key in the style box, and tags vague enough to fit any song. Fix them in that order — name the voice, add the numbers, then cut the style list to the tags you would defend.
Can I use a song I like as the starting point? ▾
Not by uploading it. Measure it instead: tempo, key, brightness, dynamics and the instruments doing the work. Those numbers are facts about the music rather than a copy of the recording, and they get you closer than "sounds like" ever does.
Can I release Suno output commercially? ▾
It depends on your plan and their current terms — free tiers typically carry non-commercial conditions. Check the terms before you plan a release around a track, and keep your receipts.