Sung lyrics and the real singer

Write lyrics to the tempo, describe a singer's style in musical words, keep one vocal character across versions, and finish every sung track with a real singer.

Writing sung lyrics [best available, not yet validated], songwriting craft with working numbers:

1. Start from the tempo. At the song's BPM a 4/4 bar holds about 4–8 sung syllables at a comfortable pace: one a beat for a held hook, two a beat for a verse, fewer in a slow song. Count each line's syllables as sung in the dialect: a long vowel is one syllable; a doubled consonant adds none.

2. Stress on the strong beats. A word's stressed syllable falls on beat 1 or 3, never on a weak off-beat. Read each line aloud to a click at the BPM, and rewrite any line you have to rush.

3. The hook: 2–5 words, sung at least twice, built on open vowels (aa, ee, oo), landing on a downbeat.

4. The brand line sits on the hook's last bar or on the tag, on a held note, spelled as the brand is said, never split across a bar line.

5. The sub-dialect's own words, spelling and rhymes (see The brief and research → Dialect rules); no Modern Standard Arabic (MSA) unless the brief asks. Real sayings and pop-culture references only, each one verified.

6. Three lyric texts, like the three texts of a spoken line: the canonical lyrics, clean, for pages and captions; the Suno text, with tags, ad-libs in parentheses and vowel marks as the voice rules set them; and the singer's sheet, the lyrics bar by bar with the English meaning.

Sung Arabic spelling follows the voice rules (see Voice and Arabic dialects → Spelling for the engine); no source shows whether marks help Suno, so the voice rules hold [best available, not yet validated]. Every sung take is transcribed against its lyrics, flagging content words only, because sung delivery bends words. A native speaker of the dialect passes every sung line, never only the key lines, and Saudi, Gulf and every other non-Egyptian vocal stays a draft until it passes (see Voice and Arabic dialects → The native reviewer).

A singer's style in plain musical terms. Client pages may name the singers a song draws on (see The idea → Writing a concept); Suno's fields never do, and no song imitates a living singer's voice closely enough to pass as them. Turn the singer into properties, and write those in Styles:

• voice: register (low baritone, high tenor, alto), timbre (warm, bright, nasal, husky, breathy) and weight;

• ornament: melisma on held vowels, short runs, slides into notes, vibrato and where it comes;

• maqam colour, for Arabic songs: the scale's feeling in words, with its name if the brief wants it. Bayati is warm and slightly sad. Hijaz is dramatic, with its wide step. Rast is bright and noble. Nahawand sits close to a minor key. Suno reads maqam names loosely, so describe the feeling too;

• rhythm: the named rhythm (Saudi samri or khabeeti, Khaleeji 6/8, Egyptian maqsum or baladi), the tempo and the swing;

• delivery and era: conversational or declamatory, ad-libs, call and response with a chorus, the production sound of the decade.

Personas: one vocal across versions. When a jingle has versions (cut-downs, a second language, seasonal lyrics), save a Persona from the approved guide take and use it for every Suno version, so the guide vocal stays one character. Never make one from a commercial recording or a real singer's voice. Whether Personas work on v6 is read, free, on your first jingle. The finals are still the real singer.

The real singer's session. You approve the brief and the booking.

• Casting brief: dialect and sub-dialect, gender, range, the style in plain terms, the guide take, the lyric sheet; the cost quoted per project, with a buyout for paid media.

• The bed: the Suno instrumental stem of the locked take (Split from Mix, vocal plus everything else, 20 Suno credits), at the locked tempo, with a click from the tempo map; the guide vocal in the singer's headphones as a reference only.

• Delivery spec:

  • dry WAV, 48 kHz, 24-bit, mono;

  • the lead and each double or harmony as separate files, every file starting at bar 1 of the bed, so they drop into the edit in sync;

  • three full takes, plus pickups of any line;

  • 1 s of room tone;

  • no reverb and no tuning baked in (a tuned copy only as an extra file).

• Session: directed live by you or the native reviewer.

• Then: the singer's vocal over the bed replaces the guide on the first music track, with the guide muted beneath; sung shots are re-synced to the singer's stem; the hero version is mixed from the stems in finishing.

Two style translations, written for this guide. "Draws on the great Egyptian tarab divas" becomes:

Egyptian tarab-style ballad, 4/4, 76 BPM, rich low female alto, long ornamented phrases on held vowels, slow wide vibrato at the ends of lines, nahawand colour, longing and dignified, a string section answering each phrase, qanun runs between lines, 1960s orchestral sound

"The energy of a Saudi samri troupe" becomes:

Saudi samri, 6/8, 98 BPM, a strong unison male chorus, short declamatory phrases, call and response between a lead voice and the group, frame drums and deep hand drums, rhythmic group clapping, proud and celebratory, raw live-room sound

Sources

• Suno, What are Personas? and Can I use my own lyrics?.