Suno Lyrics Tags: The Complete Section + Directive Reference
Everything in square brackets in a Suno lyric is an instruction, not a word to sing. There are two kinds — section tags that name the structure, and performance directives that shape the production — and they follow different rules. This is the complete vocabulary our forge emits on every song, with the formatting rules that keep an engine from singing your stage directions out loud.
The 14 section tags
Section tags name the structural blocks of the song. Our format layer recognizes fourteen, and they cover essentially every section a modern song uses:
[INTRO] · [VERSE 1] · [PRE-CHORUS] · [CHORUS] · [POST-CHORUS] · [HOOK] · [BRIDGE] · [BREAK] · [BUILD UP] · [INTERLUDE] · [TRANSITION] · [SOLO] · [INSTRUMENTAL] · [OUTRO]
Three formatting rules, all load-bearing:
- One tag per line, nothing else on the line. A tag buried mid-line stops being a tag and becomes a lyric. Our normalizer only treats a bracket as structural when it is the entire line.
- Numbers go inside the bracket, after the name.
[VERSE 1],[VERSE 2],[PRE-CHORUS 2]. Numbering matched sections helps any engine map verse 2 onto verse 1's melody. - Spelling variants are tolerated, but pick one. Our layer accepts
[PRE-CHORUS]and[PRE CHORUS],[BUILD UP]and[BUILD-UP],[INST]and[INSTRUMENTAL], and folds each pair to one canonical form. A lyric that alternates spellings reads as two different sections to a less forgiving parser.
If you invent a section name outside this list, it usually won't break anything — our normalizer passes unknown tags through — but you're gambling on the engine's interpretation instead of using vocabulary it has seen a million times.
The five directive families
Performance directives are the second bracket type: ALL-CAPS production instructions placed at the start of the section they govern. Our forge's writing room debates these explicitly — they're locked alongside the structure, not sprinkled on at the end. The five families:
- Instrumentation shifts:
[CLEAN ELECTRIC GUITAR. ONE STRUM PER BAR.]·[BASS + DRUMS ONLY. TIGHT, ANXIOUS GROOVE.]·[FULL BAND DETONATION.] - Dynamics/intensity:
[MAXIMUM INTENSITY.]·[GENTLE, BARELY THERE.]·[BUILDING — EACH LINE LOUDER THAN THE LAST.] - Silence/space:
[DEAD SILENCE. NO BREATH.]·[MUSIC CUTS OUT FOR ONE BEAT] - Vocal delivery:
[WHISPER]·[SPOKEN, NOT SUNG]·[VOICE CRACKING] - Re-entry/impact:
[MUSIC SLAMS BACK IN.]·[DRUMS HIT LIKE A WALL.]·[EVERYTHING DROPS EXCEPT PIANO.]
The placement rules we enforce internally: a directive sits at the start of its section, before the first sung line; it's specific enough that a producer who can't hear the song could build the moment from it; and not every section gets one — directives go where the production shift serves the emotion, with intros, outros, and deliberate silences earning them most often.
The saturation problem: when every song whispers
Directives obey an economy. We audited 200 of our own generated songs and found [whispered]-style delivery directives in 80% of them (159/200) and a voice-crack directive in 44% (88/200). When everything whispers, nothing whispers — so we now hard-cap both at one per song, and the default is none. A four-stanza ballad about a walk to the mailbox does not need a whisper.
The same logic applies to your hand-written lyrics: a directive per section is a sensible ceiling, and the most powerful directive in the vocabulary is usually a silence placed right before the line that matters most.
How the tags translate to other engines
Not every music engine reads the same dialect. Our provider layer normalizes the format per engine: section tags get re-cased ([VERSE 1] becomes [Verse 1] for engines that expect title case), and performance directives get converted into parenthetical directions — [FULL BAND DETONATION.] becomes (full band detonation) — for engines that treat unknown brackets as singable text. The structure survives translation; the directive syntax is the part that varies. That asymmetry is the whole subject of Udio vs Suno prompt formatting.
The four mistakes that garble a render
The failure modes we see most, in order:
- Punctuation as performance direction. Semicolons, em-dashes, parentheticals in the lyric line — engines read the raw word sequence. If a pause matters, give it a line break or a bracketed tag, not typography. (Mistake #3 in the seven Suno-killing mistakes.)
- Lowercase or mid-line tags.
my heart [whisper] breaksis a lyric containing the word "whisper" in costume. Tags live on their own line, in caps. - Stage directions written as lyric lines. "The music swells as she remembers" outside brackets will be sung, verbatim, in full sincerity.
- Directives that contradict the style string. If the style string says "intimate acoustic ballad" and the chorus directive says
[FULL BAND DETONATION.], you've given the engine a paradox. Decide which document owns the dynamic arc — see the style string checklist — and make the other one agree.
A formatted skeleton to copy
Here is the shape every forge output follows — structure tags on their own lines, directives at section starts, sung text everywhere else:
[INTRO]
[DEAD SILENCE. THEN A SINGLE PIANO NOTE.][VERSE 1]
[CLEAN ELECTRIC GUITAR. ONE STRUM PER BAR.]
First lyric line here…[CHORUS]
[FULL BAND DETONATION.]
Chorus lyrics here…[BRIDGE]
[DROP ALL MELODY. PURE VOCAL.]
Bridge lyrics here…
Tags handle the format. The words still have to be worth singing — that part is covered in writing lyrics Suno can actually sing and the full Suno playbook.