AI Song Prompts: 30 That Actually Work (and the Formula Behind Them)

Most bad AI songs are bad prompts. The model is not guessing at random — it is filling in everything you left out, and a vague description leaves it a lot of room.

This is a working list of prompts you can paste directly, plus the formula underneath them so you can write your own.

The formula

Every prompt below is built from the same five parts, in this order:

Genre → Tempo → Vocal → Instruments → Mood/occasion

[genre], [BPM], [vocal type], [instruments], [mood] — [what it is about]

You can drop any part. Each one you drop is a decision handed back to the model. In practice, vocal type and tempo are the two people forget most and regret most, because they are the hardest to fix by re-rolling.

One more lever worth knowing: adding a key ("E minor", "C major") nudges the emotional colour. Minor keys read as sadder or more serious; major keys as brighter. It is three words for a real effect.

Pop

1. Upbeat pop, 118 BPM, bright female vocal, synths and handclaps, feel-good and summery — about finally moving out
2. Pop ballad, 88 BPM, emotional female vocal, piano and strings, heartfelt and cinematic — about missing someone at Christmas
3. Dark pop, 100 BPM, breathy female vocal, sub bass and sparse percussion, moody and defiant
4. Pop rock, 130 BPM, energetic male vocal, big electric guitars and live drums, triumphant — about starting a business
5. Dream pop, 92 BPM, soft layered vocals, reverbed guitars and warm pads, hazy and nostalgic
6. Euphoric dance pop, 126 BPM, bright vocal hook, driving synths and a big drop, festival energy

Acoustic and singer-songwriter

7. Acoustic singer-songwriter, 100 BPM, warm male vocal, fingerpicked guitar and light percussion, intimate and honest
8. Folk, 96 BPM, gentle female vocal, acoustic guitar and mandolin, wistful — about leaving a small town
9. Country ballad, 76 BPM, weathered male vocal, steel guitar and brushed drums, tender and plainspoken
10. Indie folk, 108 BPM, close-mic'd vocal with harmonies, banjo and stomps, hopeful and communal
11. Acoustic apology song, 84 BPM, soft male vocal, single guitar, sincere and unpolished — asking someone to forgive you

Hip-hop and R&B

12. Laid-back hip-hop, 90 BPM, confident male vocal, warm bass and crisp drums, smooth and self-assured
13. Boom bap, 94 BPM, gritty vocal delivery, dusty samples and vinyl crackle, nostalgic
14. Contemporary R&B, 72 BPM, silky female vocal with harmony stacks, Rhodes piano and soft trap drums, late-night and tender
15. Trap, 140 BPM, hard-hitting male vocal, 808s and rolling hi-hats, cocky and celebratory
16. Neo-soul, 80 BPM, rich vocal with runs, live bass and organ, warm and grown

Cinematic and orchestral

17. Cinematic orchestral pop, 85 BPM, soaring vocal, strings and brass swells, epic and hopeful
18. Animated-movie ballad, 82 BPM, expressive female vocal, orchestra and harp, magical and yearning
19. Broadway showtune, 112 BPM, theatrical vocal, brass and strings, big and joyful
20. Epic trailer choir, 70 BPM, layered choral vocals, taiko drums and low strings, ominous and building

Occasions

21. Warm celebratory birthday pop, 110 BPM, joyful vocal, big sing-along chorus with their name — for a 60th birthday
22. Comedic upbeat pop, 125 BPM, cheeky vocal, a funny birthday roast that still ends sweet
23. Romantic ballad, 74 BPM, tender male vocal, piano and strings, emotional — for a wedding first dance
24. Feel-good acoustic, 104 BPM, warm mixed vocals, guitar and claps, grateful — a thank-you song for a retiring teacher
25. Gentle ballad, 68 BPM, soft female vocal, piano and cello, bittersweet and comforting — a memorial song for a grandmother
26. Bouncy pop, 116 BPM, playful vocal, ukulele and glockenspiel, celebratory — for a new baby's arrival

Kids and lullabies

27. Soft lullaby, 60 BPM, gentle female vocal, music box and warm strings, soothing and tender
28. Cheerful nursery rhyme, 100 BPM, sweet bright vocal, ukulele and glockenspiel, sing-along and simple
29. Playful kids pop, 115 BPM, giggly bright vocal, toy piano and claps, silly and energetic — about a dog who steals socks
30. Bedtime acoustic lullaby, 66 BPM, near-whispered vocal, fingerpicked guitar, calm and slow — using the child's name

Four fixes for when it comes out wrong

Wrong voice. This is the most common complaint and the easiest fix — the vocal type was left out or left vague. Name it directly: "warm male vocal", "bright female vocal", "child-like vocal", "husky low female vocal".

Too cluttered. You probably named too many instruments. Cut the list to two, and add "sparse arrangement" or "minimal production".

Too generic. Add the specific thing. A name, an occasion, a place, a detail only you would know. Generic in, generic out.

Wrong energy. Words like "upbeat" are doing less work than you think. Replace with a BPM. "Upbeat" spans 100 to 140, and those are very different songs.

Where the lyrics fit

Every prompt above describes the sound. The words are a separate question, and you have three options: leave them empty and let them be written as part of the song, write them yourself with [Verse] and [Chorus] tags, or write the lyrics with AI first so you can read and edit them before the music is built around them.

For anything with a name in it — birthdays, weddings, dedications — write the lyrics first. It is the difference between a song about a birthday and a song about their birthday.

Try one

Copy any prompt above, change one detail so it is about someone real, and generate it. Then, if you want, put a face to it and let a photo sing the result.