Writing Suno Style Prompts
YAML--- name: writing-suno-style-prompts description: Crafts genre-precise Suno AI style prompts (positive + negative) for electronic music tracks by modeling specific artist production signatures. Use when generating AI music prompts for electronic subgenres (future bass, melodic techno, tech house, etc.) that need distinct tempo, instrumentation, vocal delivery, arrangement, and negative-prompt exclusions to avoid generic or genre-bleeding outputs. ---
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).