Writing Suno Style Prompts
A Suno style prompt has two parts: Positive (everything the track should contain) and Negative (everything to exclude). Structure the positive prompt in this fixed order:
- Genre + mood descriptors (3-6 words, non-overlapping)
- BPM + tonality/key character
- Rhythm section (kick, bass, percussion)
- Melodic/harmonic elements (synths, pads, instruments)
- Vocal profile (who, tone, delivery style, what's excluded — e.g. "one lead only")
- Arrangement/structure (intro → verse → build → drop → breakdown → final drop)
- Mix character (one closing phrase: e.g. "wide stereo, modern digital mix")
Negative prompt: list genre-adjacent sounds and vocal types that would bleed into the wrong style, using a consistent format (hyphen or plain comma list — pick one and stay consistent).
Progress:
- Identify the reference artist/track's actual subgenre (don't default to "EDM" — distinguish future bass vs melodic techno vs tech house vs peak-time techno etc.)
- Lock the BPM to the genre's real range (future bass ~100, melodic techno ~122-126, tech house ~124-128, peak techno ~126-130)
- Select 4-6 mood words that don't overlap in meaning (avoid stacking synonyms like "nostalgic, bittersweet, wistful, sentimental" — pick the two sharpest and cut the rest)
- Define the rhythmic backbone first (kick character + bass relationship) — this is the genre's fingerprint
- Define ONE vocal profile in detail: gender/register, tone, delivery mode (sung/spoken/chanted), and explicitly state "one lead only" or similar to prevent duet bleed
- Write arrangement as a single flowing sentence tracing intro → tension build → drop → breakdown → final drop
- End positive prompt with a one-line mix descriptor
- Write negative prompt: exclude the 2-3 nearest-neighbor genres, opposite vocal types, and competing production techniques (e.g. if melodic techno, exclude trance supersaws and dubstep growls specifically, not generically)
Example 1: Input: "Anyma - Samsara" (melodic techno, cinematic, female vocal) Output:
Electronic, melodic techno, dark, hypnotic, cinematic, haunting, 124 BPM, minor tonality, deep four-on-the-floor kick, pulsing sub bass, rolling synth bass sequence, metallic synth plucks, evolving arpeggios... One low-register female lead, smoky resonant tone, intimate controlled delivery... Opens with an eerie synth motif... rising synth modulation builds into a powerful melodic drop.
Negative: Male vocals, duet, choir, cheerful pop, acoustic folk, tropical house, funky slap bass, rap verses...
Example 2: Input: "Mau P - Drugs From Amsterdam" (tech house, spoken vocal hook) Output:
Electronic, bass-driven tech house, underground club energy, cheeky, gritty, 125 BPM, punchy four-on-the-floor kick, thick rubbery syncopated bassline... One deep male spoken vocal, casual conversational phrasing, mischievous confidence... a short original phrase repeated as the central hook...
Negative: Female vocals, duet, melodic pop singing, emotional belting, long rap verses...
- One mood pair per axis: pick one word for emotional tone (dark/euphoric/cheeky) and one for energy (hypnotic/relentless/swaggering) — don't stack four synonyms.
- Make the bass-kick relationship the genre's signature: "rolling synth bass sequence" (melodic techno) vs "rubbery syncopated bassline" (tech house) vs "sidechained supersaw" (future bass) are what actually differentiate genres to the model.
- Always specify vocal exclusivity ("one lead vocalist only," "single tracked," "no duet") — this is the single most effective way to prevent unwanted harmonies/duets.
- Tie the drop description to the hook mechanism: state what carries the hook (vocal chops, repeated chant, synth motif) so the climax is specific, not generic "big drop."
- Keep negative prompts genre-aware, not generic: exclude the specific neighboring genres and techniques that the model is likely to confuse with the target (e.g., exclude "trance supersaw wash" for melodic techno, not just "trance").
- Reference tempo as a range anchor, not dogma: note "reference tempo" so the prompt reads as a production direction, not a rigid lock.
- Stacking near-synonym moods ("nostalgic, bittersweet, wistful, sentimental, tender") — dilutes direction; pick 2-3 max.
- Forgetting to exclude the opposite vocal gender/style in the negative prompt — causes duet/harmony bleed.
- Generic arrangement language ("builds to a drop") instead of naming the specific build mechanic (snare roll, filtered riser, silence cut).
- Reusing the same instrumentation palette across different genres — future bass pads ≠ melodic techno pads ≠ tech house percussion; each needs its own vocabulary.
- Negative prompts that just invert the positive list word-for-word without considering real genre-adjacent confusions (e.g. excluding "hardstyle kicks" for a genre that was never at risk of that bleed).