How to Write Suno Prompts: Structure, Style Tags, Templates

Updated 2026-09-11 · Suno-API team

Learn how to write Suno prompts that control genre, mood and instruments. Includes a fill-in template, a style tag cheat sheet and three common mistakes.

Take the same lyrics, swap in a different set of prompts, and you may end up with two completely different songs. A prompt is not a pile of adjectives. It works more like a brief for an arranger: the more specific you are, the better the model knows what you want.

The five parts of a prompt

A prompt that works usually stacks these in order, from broad to narrow:

OrderWhat it describesExample
1Genre and eracity pop, 90s Hong Kong style, lo-fi hip hop
2Mood and atmospherehealing, letting go after a breakup, alone late at night
3Lead instrumentselectric piano, fingerpicked acoustic guitar, saxophone
4Rhythm and tempomid-tempo, slow rock, light drums, no obvious climax
5Vocals and production texturefemale vocals, breathy, close to the mic, vintage tape feel

A template you can use as-is

Replace what is in the brackets with your own details and you have a usable prompt:

[Genre] + a song about [subject], the mood is [mood], the lead instruments are [instruments], the rhythm is [fast/slow description], sung by [male/female] vocals, overall it should sound [production texture].

A filled-in example: city pop, a song about walking on the beach on a summer evening, the mood is relaxed with a touch of nostalgia, the lead instruments are electric piano and saxophone, mid-tempo rhythm with light, upbeat drums, female vocals, overall it should sound like a vintage 80s recording.

Style word cheat sheet

DirectionCommon words
Poppop, city pop, synth pop, dream pop
Raphip hop, trap, boom bap, lo-fi hip hop
Electronichouse, techno, ambient, future bass
Rockrock, indie rock, punk, post rock
Folkfolk, acoustic, country, ballad
Chinese styleguofeng, traditional Chinese instruments, pipa, dizi, pentatonic scale

Style words are usually more accurate in English, while mood and scene descriptions read more naturally in Chinese. Mixing the two is fine.

Three techniques that noticeably improve results

  • Limit the number of elements: cram in seven or eight instruments at once and the usual result is that none of them can be heard clearly. Three or fewer lead elements is the most reliable.
  • Use "don't" to exclude: instead of describing what you want over and over, add exclusions like "no electronic drums" or "no high-note belting".
  • Keep half fixed, change only half: when you are hunting for the best version, keep the genre and mood description and change only the instruments or the rhythm. The difference is easier to hear.

Three common mistakes

  • Writing only "a good song". There is too little information, so the model can only guess.
  • Putting lyrics in the description box. The description box controls style; lyrics have their own input field.
  • Throwing everything out after one unsatisfying generation. The same prompt produces two different takes each time, so listen to both before you adjust.

¥0.6 per generation, 2 tracks, unlimited downloads

Runs on the Suno V6 engine. Download MP3 or WAV as often as you like. Signing up needs an email address, not a phone number.

Start creating Read the API docs