Prompt Guide
Write better prompts,
get better songs
The story you write is the most important thing you control. This guide shows you exactly what separates a forgettable song from one that makes someone cry.
✦ Start CreatingStep 1 — The Story Field
The story field is everything
The AI reads your story and writes original lyrics from it. The more specific your story, the more personal and emotional the output. Generic inputs produce generic songs. The AI cannot invent personal detail — saying 'I love her' gives it nothing to work with. What she actually did gives it everything.
Name them — full identity
Not 'my girlfriend' — 'Adunola from Ibadan'. Not 'my father' — 'my father Baba Ade, a tailor in Mushin for 38 years'. The AI cannot invent personal identity. What you name, it uses. Named people become characters in the song; unnamed people stay placeholders.
Use numbers, not feelings
Not 'we've been through a lot' — 'four years, three breakups'. Not 'she worked hard' — 'double shifts for eleven years'. Not 'things are better now' — 'last month I got the promotion'. Numbers carry emotion that adjectives cannot. Specific facts become specific lyrics.
The turning point — before and after
The most powerful stories describe a change, not a state. Not 'he loves me' — 'he drove to Abuja at midnight when I called him crying'. Not 'she sacrificed for me' — 'she sold her last gold earring to pay my WAEC fee'. One specific action beats a paragraph of 'he's amazing'. What did they actually do?
Lyrics Tip — Write to be Sung
How to write lyrics the AI sings clearly
Whether the AI writes your lyrics or you paste your own, every song goes through the same voice engine. These three rules make the difference between crisp vocals and mumbled, slurred delivery.
Max 8 words per line
Short lines give the AI clear phrasing boundaries. A line with 12+ words gets crushed onto too few notes and sounds rushed or slurred. Ajasa automatically splits any line longer than 8 words — but writing short lines from the start gives the AI more control over where melody lands.
Use section markers
Put [Intro], [Verse], [Chorus], [Bridge], or [Outro] on its own line before each section. Without them the AI sings the whole song at the same energy level — no build, no drop, no contrast. These markers are how the song tells the AI when to lift and when to hold back.
End lines on open vowels
Lines ending on -ove, -ay, -ee, -oo, -ah, -o, -me, -you hold melody well — the mouth stays open and the note can ring out. Hard consonant endings like -nt, -ck, -st, -nd cut the phrase short. Wherever possible, land your last syllable on a vowel sound.
✗ Hard to sing clearly
“I never thought that I would find someone who truly understood me the way that you do”
✓ Sings cleanly
[Intro] Somewhere in the rain [Verse] I never thought I'd find you Someone who understood me You stayed when I was nothing You saw what I could be [Chorus] You are my home You are my free
Step 2 — Genre, Mood & Tempo
Combinations that always work
Genre sets the musical foundation. Mood gives the AI emotional direction. Together they determine the feel of the song before a single lyric is written.
Slow acoustic-driven ballad with talking-drum undertones and gentle call-and-response. The AI writes intimate, conversational lyrics — meant to be felt, not danced to. Best for anniversaries, proposals, and deeply personal dedications.
High-energy Afrobeats production with syncopated percussion and anthemic chorus structures. The AI writes celebration-forward, crowd-ready lyrics. Best for birthdays, graduation, owambe dedications, and any moment that deserves a banger.
Full choir swell with organ lead and testimony-format lyricism — from trial to breakthrough. The AI structures the song as a praise narrative. Best for memorials, thanksgiving songs, church celebrations, and songs of faith.
Deep log-drum bassline with piano-house chord progressions and township-rooted energy. The AI generates dance-forward, feel-good lyrics. Best for South African weddings, Afro-fusion parties, and cross-continental celebrations.
Smooth neo-soul production with slow-burn phrasing and space for vocal runs. The AI generates reflective, memory-driven lyrics. Best for long-distance dedications, reunion songs, late-night mood pieces, and throwback stories.
Lead electric guitar over West African percussion with a bright, celebratory arrangement. The AI writes community-oriented, praise-forward lyrics. Best for 60th–80th birthday songs, elder tributes, hometown dedications, and naming ceremony songs.
Talking-drum lead with electric guitar and Yoruba cultural phrasing. The AI generates oriki-inspired praise patterns — naming the subject by their qualities and lineage. Best for Yoruba elder tributes, naming ceremonies, and cultural deep-cut dedications.
British-Nigerian production — Afrobeats groove with melodic R&B phrasing and smooth contemporary delivery. Best for diaspora love stories, city romance, London-Lagos relationship songs, and modern dedication tracks.
How to write your Sound Direction
The Sound Direction field in the Style section goes straight to the music AI — separate from your story. Here is how to write it so the beat comes out the way you heard it in your head.
Name the instruments you want
The most direct way to control your sound: just name the instruments. 'Talking drum, bass guitar, electric guitar, tenor sax.' 'Log drum kick, piano chord stabs.' 'Grand piano, gospel choir, organ.' The music AI reads these as production instructions and builds those sounds into the beat. Don't be vague — 'African drums' is weaker than 'talking drum and dundun'. The more specific, the closer the result is to what you imagined.
Add texture words
After naming instruments, add texture: heavy, deep, sparse, dense, layered, stripped, intimate, warm, bright, dark. These shape how the instruments feel against each other. 'Deep bass, sparse percussion' = lots of space and air. 'Heavy bass, dense drums, layered percussion' = a full wall-of-sound. 'Soft piano, no drums' = completely intimate. Two words of texture on top of your instruments can change the whole character of the beat.
Describe how the groove moves
You can also describe the rhythm pattern. 'Rolling hi-hats' tells the AI the hi-hat should move in triplets. 'Syncopated groove' means the kicks hit off the expected grid. 'Steady 4/4 beat' means a locked, predictable pulse. 'No drums at all' removes percussion entirely and leaves only melodic instruments. 'Live band feel' makes the rhythm section feel like real musicians in a room, not a drum machine. Mix these with instrument names for the most precise result.
Ready-made examples to copy
Afrobeats ceremony: 'talking drum lead, deep bass guitar, electric guitar fills, no hi-hats, warm and sparse'. Pure Amapiano: 'log drum kick, piano chord stabs, violin melody, no live drums'. Gospel ballad: 'grand piano, full gospel choir, organ swell, no percussion until chorus'. Party banger: 'heavy 808 bass, rolling hi-hat triplets, brass stabs, shekere'. Acoustic-only: 'nylon string guitar, soft shaker, no bass, no drums, intimate'. Write these exactly as shown — short comma-separated phrases, no full sentences needed.
A complete prompt, filled in
This is what both fields look like when they work together. The story tells the AI what to say. The Sound Direction tells it how to sound.
Story
My Aunty Funke is turning 60 this Saturday. She has been selling fabric at Balogun Market in Lagos for 35 years. Her stall is number 14 and it has a yellow shade at the front that everyone in the market knows. With money from that stall she paid school fees for six of us — her nieces and nephews — including me. She never got married but we all call her Mama. One day I asked her if she ever gets tired. She said: 'Tiredness is for people who are not yet done.' I want this song to celebrate her and show her that her stall changed all our lives forever.
Sound Direction
lead electric guitar, warm talking drum, bass guitar, brass stabs on the chorus, no hi-hats, live band feel, celebratory
Why this works
The story gives the AI a real person (Aunty Funke), a specific place (stall 14, yellow shade, Balogun Market), real numbers (35 years, six people), and her actual words to work from. The Sound Direction sends completely separate instructions to the music generator — it names the instruments, describes when to hit hard (brass stabs on the chorus), and sets the energy (live band feel, celebratory). The story never needs to mention music. The Sound Direction never needs to mention people. They do two completely different jobs.
Step 3 — Language tips
Writing for Nigerian & African languages
When you choose Yoruba, Hausa, Igbo, or Pidgin, the AI writes lyrics natively in that language — not a translation. Your story is still in English; the AI handles the cultural adaptation.
- ·Reference Yoruba proverbs or family values (ẹbí, ìdílé) if relevant
- ·Mention a town — Ibadan, Osogbo, Ilé-Ifẹ
- ·The AI uses authentic Yoruba tonal language and prosody, not transliteration
- ·Northern Nigerian cultural references land naturally — Kano, Sokoto, market life
- ·Family and faith themes are especially strong in Hausa lyrics
- ·BibleTTS-trained voice gives native Hausa pronunciation in the vocals
- ·Pidgin works best for upbeat, street-level, relatable stories
- ·You can mix English and Pidgin cues — the AI will naturalise the blend
- ·Reference Lagos slang culture freely — the model is trained on it
- ·Works especially well for Côte d'Ivoire, Senegal, Cameroon, and DRC context
- ·Afrobeats + French is a fast-growing sound — lean into it
- ·The vocals are fully French — not accented English
Cover Art
Choosing the right cover art tier
Cover art is generated entirely from your song data — genre, mood, title, and lyrics. You never need to write a prompt for the AI mode. You only write a prompt if you choose Augment mode.
FLUX.2 Klein — fast 1MP generation. Good for testing layouts.
Best for
Quick previews, demos, lower-budget songs
How the prompt works
Automatic — Ajasa writes the prompt from your song data. No input needed.
FLUX.2 Pro — 4MP photorealistic output. GPT-4o writes a cinematographic prompt from your song.
Best for
Songs you plan to share on Spotify, Apple Music, social media
How the prompt works
Automatic — GPT-4o reads your title, genre, mood, story and opening lyric, then writes a FLUX-optimised prompt with specific lighting, lens, and colour palette.
Best-of-2 FLUX.2 Max — generates two top-tier image candidates, scores both for quality, returns the winner.
Best for
Release-ready commercial singles, professional portfolios, high-stakes gifts
How the prompt works
Same GPT-4o prompt as Pro — but two completely different FLUX.2 Max images are generated from it and the higher-quality one (measured by visual richness, sharpness, and colour depth) is returned.
Augment mode — using your own photo
Ready to try it?
You now know more about writing great prompts than 95% of users. Go make something.