The Master Guide / 12 Music · Suno
Suno controls and starting values
Work in Advanced mode, with Variety and My Taste off, and change one slider at a time.
Work in Advanced mode, with Variety and My Taste off, and change one slider at a time.
The controls that matter
From Suno's own pages [official guidance], and as read in the app in October 2026. Locations may move.
Control | What it does | Use |
Songs, Speech, Sounds tabs | Songs is music. Speech mixes a spoken track and music into one file. Sounds makes one-shots and loops at a set tempo (BPM) and key | Songs for every option; Sounds for a missing hit; never Speech for a final |
Simple or Advanced | Only Advanced shows the lyrics, styles, sliders, Exclude and Duration | Advanced, always |
Model picker | v6 is Suno's main model, v6-wild the one Suno labels experimental, and v6-mini a lighter one | v6. v6-wild for at most one option of three. v6-mini is not used |
Credits | One Create makes two clips: 10 Suno credits, 20 with Max Mode | Pro holds 2,500 Suno credits a month |
Lyrics | Words to sing plus section tags in square brackets. Words in parentheses come out as ad-libs or chants. Empty means instrumental. Up to 5,000 characters | Any word in it may be sung: never the lyric helper for client lyrics |
Styles | Genre, era, BPM, key, instruments and their jobs, vocal type, mood, arc, ending; 1,000 characters | Musical terms only, never an artist's or a song's name |
Exclude styles | A negative list, shown on the song page with a minus sign | The brief's avoid list |
Vocal Gender | Steers male or female; Suno says it doesn't guarantee it | Only with vocals, beside a voice description in Styles |
Duration | The desired length, 10 s to 6 min | Custom, set to the cut length, as a target |
Max Mode | More compute for consistency, at twice the Suno credits | Off at 60 s or under; on beyond 2 min |
Weirdness | Pushes away from the expected; 0–100%, default 50% | See the starting values |
Style Influence | Pulls toward the style text; 0–100%, default 50% | See the starting values |
Audio Influence | The pull of an uploaded reference; appears only after an upload | High to keep a bed, low to take only a feel |
Variety | Rewrites your style text before generating; its default is not Off | Off (the reset arrow returns to the default, which is not Off) |
Personalize: My Taste | Tilts toward the account's listening history | Off for client work |
Edit menu | Extend, Crop, Remove Section, Fade, Replace Section, Remaster, Get Stems and more; each spends Suno credits | For endings, sections and hit points |
Remix menu | Cover (the same melody in a new style), Reuse Prompt, Mashup, Sample, Use as Inspiration | Cover only, for a big style change on your own song |
Get Stems | Auto Split: up to 12 stems, 50 Suno credits. Split from Mix: one part plus everything else, 10 Suno credits a stem (20 for both) | The chosen take only (see Music → Style-text examples, style formula 6) |
Download | MP3 on every plan; WAV on Pro; stems ride with their song | Plan every download (see Music → Downloads and the licence) |
Song info | Shows the style text, exclusions, sliders, Variety and Max Mode used | It doesn't show Duration: log it yourself |
Notes on the table
Never use Speech for a final: the voice must stay locked, and the music on its own track.
Never press the Personalize wand or the style chips.
For an instrumental, Exclude takes lead vocals, singing, lyrics and spoken words.
Duration is a target: the final fit happens in the edit.
Max Mode also goes on for close covers, or when a long track loses consistency.
The + inputs, Personas and Studio
Leave every + input empty unless your plan names it.
+ Audio uploads a reference, or the source for a Cover or an Extend, and brings up Audio Influence. Upload only your own material, where the NDA allows, never a commercial recording (up to 30 minutes on Pro).
+ Voice takes only your own verified singing voice. It can't carry a hired singer's voice, so a real singer always records the finals in a session (see Music → Sung lyrics and the real singer).
+ Inspo takes the feel of a playlist or a song (Use as Inspiration, on Pro); use it only when your plan names it.
Personas, inside the Voices menu, keep one vocal character across a jingle's versions (see Music → Sung lyrics and the real singer).
Studio, Suno's multitrack (clip stretch, fade curves), needs the Premier plan. It isn't part of the Pro workflow: fits and fades happen in the edit.
How the controls interact
Weirdness pushes away; Style Influence pulls toward your text. Low Weirdness with high Style Influence is the most literal take, and the likeliest to sound like library music. For surprise, move one option to v6-wild and keep Weirdness moderate, rather than pushing Weirdness high.
Change one slider per retry, by ±10. The two partly cancel, so moving both hides which one fixed or caused the problem.
Low Style Influence lets the lyrics, an upload or the model's habits lead; use it only when the style text is a light suggestion.
Variety and My Taste off keep every option traceable to the exact text you wrote and logged. The variety between three options comes from three different style texts. Raise Variety to Normal only when a right direction keeps giving near-identical clips.
An instrumental with structure [best available, not yet validated]: tags-only lyrics (intro, build, drop, button end) give sections with no words; add "instrumental" in Styles and the vocal exclusions as a second safeguard. If vocals still stray, switch to empty lyrics.
Duration sets the length; tags divide it. Each section needs time: at most 3 sections for 15 s, 4–5 for 30 s, 6–8 for 60 s. Timings written in the style text are followed only loosely.
Endings are cues, not commands. Fix a failed ending with an Edit tool, never more prompt text: Replace Section on the last bars, Crop to the downbeat, Fade Out, or Extend from a bar earlier with a hard-stop tag.
Songs have no BPM or key control. Write both in Styles as numbers, then measure the BPM on every take; never trust it.
Starting values
These are starting points: no creator publishes values, and none has compared them.
Job | Weirdness / Style Influence | Max Mode |
General bed | 35 / 75 | Off |
Voiceover-friendly bed | 25 / 80 | Off |
Cinematic build | 45 / 70 | Off at 60 s or under; on beyond 2 min |
Sung jingle | 30 / 80 | Off |
Every job | Variety off, My Taste off, model v6 | — |
On earlier projects (one account, a small sample): standard mode at Weirdness 35 landed within about 1 s of Duration in 10 of 10 clips; 50/50 came out up to 4 s short; Max Mode ran short every time, by up to 10–15 s, which is why it starts off at 60 s or under.
Before each Create
Songs tab, Advanced, the model. Leave + Audio, + Voice and + Inspo empty unless your plan says otherwise. Paste the lyrics (or leave them empty) and the style text. In More Options, the bin icon is Reset All: don't press it. Re-check every value just before each Create, because values may not carry over between sessions: Exclude, Vocal Gender, Duration, Max Mode, Weirdness, Style Influence, Variety and My Taste. Title each song by project, direction, length and round, and save it to one workspace per project.
Read once, at no credit cost, the first time a round meets it: Audio Influence's range and default, on an approved upload; whether + Voice and Personas work with v6; whether a take can be played in sync with the assembly in the browser.