Suno Prompt & Style Tag Playbook

Updated: 2026-09-04. Prompt field names, lyric-mark support, and model behavior follow Suno that day. The same prompt can yield different arrangement density on V5.5 / V5 / V4.5—pick the tier for the use.

Introduction

Effective Suno prompts are not fifty English genre words—they reliably deliver five signals: mood, genre, instrument texture, vocals, tempo/energy, plus clear lyric section marks. This playbook gives a reusable prompt anatomy, tag habits, lyric marking, and four copy-ready templates (short-form BGM, ads, podcast idents, indie singles) plus anti-patterns. Pipeline place: Song workflow. Absolute beginners: Getting started.

What this guide solves

• Fixed slots so generations stop feeling empty and chaotic
• Style tags and lyrics each do one job
• Four high-frequency templates you can paste and retarget
• Spot mutually exclusive tags, infringement-risk prompts, and over-control habits

Prompt anatomy: five slots

Write the style prompt (or Style / Prompt field) in this order; commas on one line are fine—don’t scramble the logic:

1. Mood: warm, melancholic, uplifting, dark, playful…
2. Genre: indie pop, lo-fi hip hop, synthwave, cinematic orchestral…
3. Instruments / texture: acoustic guitar, 808, soft pads, piano ostinato…
4. Vocal: soft female vocal, warm male vocal, duet, choir, instrumental…
5. BPM feel: slow ballad feel, mid-tempo groove, fast driving beat… (feel words first; approximate BPM only when you have a real target)

Shortest usable skeleton:

[mood], [genre], [2–4 instrument/texture words], [vocal or instrumental], [energy feel], [one use/space sentence], [bans]

Filled example:

warm, indie pop, acoustic guitar, light drums, soft pads,
clear female vocal, mid-tempo, cozy cafe vibe,
no screaming, no heavy metal distortion

Why English genre tags often behave better

Training corpora and UI habits circulate genre/instrument labels heavily in English. Put use notes or lyrics in your language. Practice: English style tags + lyrics in your target language. If an entry accepts non-English style prose, still keep common English genre names to cut ambiguity.

Where the use sentence goes

After slots, before bans—natural language for “who it’s for”:

..., mid-tempo, background music for 15s product demo, leave space for voiceover, no vocals

Use sentences move dynamics and vocal density more than three synonym genre words.

Style tags: write like arrangement notes

Recommended combinations

Goal	Tag strategy
Clear-vocal single	Genre + vocal description + clear vocals + catchy chorus
Voiceover-safe BGM	instrumental + minimal/soft + low dynamics
Cinematic trailer	cinematic + concrete texture (brass hits / pulse strings)—not empty “epic”
Retro electronic	Pick a subclass (synthwave / city pop), then drums and synth words

High-value modifiers (add 1–2 per pass)

• Space: dry mix / spacious reverb / intimate close vocal
• Era: 90s, early 2010s, modern streaming mix
• Dynamics: punchy, gentle, building arrangement, sparse intro
• Timbre: warm analog, bright pop sheen, dusty vinyl texture

Low-value piles (delete)

• Synonym stacks: epic, cinematic, grand, huge, massive orchestra five deep
• Contradictions: silent sparse with wall of sound in one line
• Empty flex: masterpiece, Billboard hit, Grammy (no controllability)
• “Fully imitate living artist X”: use genre + instruments instead to lower infringement and policy risk

Lyric marks: sections that do their jobs

In Custom lyrics, use section marks (how strongly the model interprets them follows the product that day; even weak interpretation helps your own edits):

[Verse]
…

[Pre-Chorus]
…

[Chorus]
…

[Bridge]
…

[Outro]
…

Marking tips

• Keep [Chorus] on repeats—don’t rename into prose headings
• Hook lines can stand alone and repeat in chorus
• Dialogue-style lyrics prefer short lines; don’t pack punctuation when you need sustained melody
• Humming/backing: ooh / nah plus background vocals in style
• Declare language: English lyrics only (or your language) to cut language bleed

Instrumental: leave lyrics empty (or as UI requires); style says instrumental, no vocals.

Copy-ready templates (retarget the slots)

Paste into the style field; supply your own sectioned lyrics. Practice entries: Suno V5.5 · AI music studio.

1) Short-form BGM (15–30s material)

upbeat, modern lo-fi pop, soft keys, light finger snaps, muted guitar,
instrumental, mid-tempo, instant hook in first 2 seconds,
low dynamics, leave space for voiceover,
no vocals, no sudden drops, no harsh frequencies

Acceptance: rhythm in first 3 seconds; peak loudness doesn’t crush voiceover.

2) Ad / product demo score

bright, positive, clean pop, acoustic guitar and light electronic beat,
instrumental, mid-tempo, polished modern commercial feel,
clear mix, uplifting but not childish,
no vocals, no dark minor mood, no rock distortion

Need a sung slogan: drop instrumental, use short male/female vocal hook, supply a tiny [Chorus] slogan you have rights to use.

3) Podcast ident (8–20s recognition)

bold, minimal electronic, punchy kick, short synth motif, modern podcast ident,
instrumental, cold open, strong motif in first second, quick ending,
no long intro, no vocals, no busy fill every bar

Normalize loudness in edit; lock “intro 1 / outro 1” so you don’t re-roll every episode.

4) Indie single (full structure)

Style field:

bittersweet, indie pop, electric guitar, warm bass, live drums feel,
clear female vocal, mid-tempo, intimate verse and bigger chorus,
modern streaming mix, emotional but not melodramatic,
no screaming, no autotune artifacts emphasis

Minimum lyric structure:

[Verse]
…

[Chorus]
… (theme line)

[Verse]
…

[Chorus]
…

[Bridge]
…

[Chorus]
…

Finals prefer V5.5; direction hunting can use V4.5-class. Tiers: V5.5 guide.

Anti-patterns: stop these

1. Prompt as novel
   Three hundred words of story, zero genre/vocals—model guesses. Story belongs in lyrics; style stays tag-dense.

2. Mutually exclusive tags
   silent ambient + festival EDM drop → split into two versions and A/B.

3. Full mix manuals in one prompt
   Per-track dB and exact EQ Hz rarely help. Prefer vocal forward / bass light; detail goes to the DAW.

4. Copyrighted lyrics or “make it like [hit song]”
   Legal and platform risk, and you learn nothing reusable. Mood + structure + original theme lines instead.

5. Five variables at once
   Genre, vocal, lyric language, and tempo together destroy attribution. Keep Getting started single-variable discipline.

6. Infinite Extend on a bad baseline
   Perfect prompts can’t rescue a collapsed seed. Re-roll, then extend.

7. Ignoring bans
   No no vocals then complain BGM fights VO; no language then complain bilingual chaos. Bans are first-class.

Iteration patches: keep… only change…

After baseline passes:

Keep the same indie pop melody intent and female vocal.
Only make the chorus drums punchier and raise energy slightly.
Do not change lyrics language. No EDM drops.

Keep patch language consistent with the original style field when possible.

Slot rewrites: vague → executable

Vague	Executable rewrite
Make it nicer / premium	warm modern pop, clean mix, vocal forward
Cafe vibes	cozy, indie pop, acoustic guitar, soft pads, low dynamics
Make it hype	high-energy, driving drums, bigger chorus, keep same key intent
Like a commercial	bright, positive, polished commercial pop, instrumental
Quiet but not boring	sparse arrangement, soft pulse, simple motif, instrumental

Rewrite rule: swap subjective adjectives for hearable arrangement decisions (who plays, how dense, where vocals sit, energy curve).

Five slots by use (cheat sheet)

Use	Mood	Genre bias	Vocal	BPM feel	Extra emphasis
Short-form BGM	upbeat or chill	lo-fi / light pop	instrumental	mid	Hook in 2s, voiceover space
Ads	bright, positive	clean pop	often instrumental	mid	No gloom, no distortion walls
Podcast ident	bold, minimal	electronic ident	instrumental	short & clear	Cold open, fast stop
Indie single	one clear primary mood	precise subclass	clear lead	serve lyric breath	Chorus lift; ban empty epic

Bias is not a cage—you can put indie folk on short-form, but update dynamics and vocal decisions or acceptance fails.

Advanced lyric marks: less is more

Beyond brackets, light conventions (human-readable first):

• Theme line alone with blank lines so you can count repeats
• Shout ad-libs only on last chorus lines—don’t flood verses
• Brand slogan sing: slogan-only [Chorus]; verses aren’t product spec sheets
• Multilingual: declare e.g. Verse in Spanish, Chorus in English in style

If the model sings the marks aloud, remove brackets for blank lines; keep structure in your draft for the next paste.

Diagnose → patch flow

When output disappoints, classify before adding words:

1. Wrong mood → Mood slot only; freeze the rest
2. Genre bleed → Genre + one instrument word
3. Vocal issues → Vocal description or switch to instrumental
4. Intro too long → cold open / instant hook; don’t touch lyrics
5. Wrong lyric content → lyrics only; leave style
6. OK but no hook → strengthen chorus words only (catchy chorus, vocal hook)

Log “next round only changes” to match the Getting started record template.

Quick access

• Third-party flagship try: Suno V5.5
• AI music studio: AI music studio
• Official: suno.com

FAQ

Are longer prompts always better?

No. Five slots filled + one use sentence + bans usually beat adjective essays. Long prompts earn their keep only under multi-constraint delivery—and you must check contradictions.

Must I write BPM numbers?

No. slow / mid-tempo / fast usually suffice. Approximate BPM when syncing to fixed edit hits or existing drums—and accept residual model drift.

Can I write style entirely in non-English?

Yes, but keep common English genre/instrument tags; put use and lyrics in your language. A/B with the same acceptance card on your entry.

Why does the same prompt differ so much on V5.5 vs V4.5?

Generations differ on vocal clarity defaults, section consistency, and style following. Final listen prefer flagship; exploration on common free defaults is a rational quota strategy.

Will [Chorus] appear in the finished vocal?

Usually it should not be sung; if it is, remove brackets for blank-line sections or follow UI help. Finals need a proof pass for odd readings.

Can I paste someone else’s whole prompt?

As a learning sample—retarget use, mood, bans; strip named artists and unauthorized brands. Blind copies often miss your acceptance card and inherit bad habits.

Instrumental game battle music—how to bias slots?

Mood tense/high-energy; genre a precise electronic/orchestral mix; instruments pulse, percussion, brass hits yes/no; vocal instrumental or tiny vocal samples; BPM feel driving. Use sentence: combat loop, avoid long intros. Still single-variable iterate.

Same template, different results on third-party vs official?

Normal. Model version, defaults, and fields differ. Log against the entry you bound for evaluation; before commercial, re-check plans and terms on official.

Official resources

• Suno

Further reading

• Suno Guides Hub
• Suno getting started
• Suno song workflow
• Suno V5.5 complete guide
• Suno commercial rights

Summary

A working Suno prompt = mood + genre + instrument texture + vocals + energy, plus use sentence and bans; lyrics use section marks for their jobs. Paste the four templates, then single-variable iterate; avoid exclusive tags and “imitate that hit” prompts. Models and fields follow official that day; third-party studios are fine for practice with different terms.