Why describe a build rather than just a genre
Saying only "worship song" gives the model a genre label but no arrangement. Worship tracks are recognised by their shape: a sparse verse (often piano or acoustic guitar alone), a chorus that adds drums and bass, and a bridge that strips back before lifting again. Describe that shape directly, for example "piano verse, full band chorus, bridge drops to vocal then rebuilds", rather than relying on the word worship to carry all of it.
How to make the chorus sound congregational
A congregational chorus usually has a simple, repeatable melody and layered or group voices rather than a single lead showing off range. Use phrases like "layered voices", "group of voices singing together" or "easy to sing along to" rather than technical terms. Keep the chord pattern simple by picking a genre anchor like pop (see our pop prompt guide) since worship choruses often share pop's four-chord structure and tempo range of 90 to 130 BPM.
How to write the bridge so it actually lifts
A lifting bridge needs contrast before the lift. Describe a moment of space first, such as "drums drop out" or "bare vocal line", then describe the instruments returning: "strings swelling underneath", "drums entering after a bare vocal line". Without that contrast, models often just repeat the chorus arrangement and the bridge does not feel different.
| Section | What to describe |
|---|---|
| Verse | sparse instrumentation, one or two instruments, intimate vocal |
| Chorus | full band, layered voices, simple repeated melody |
| Bridge | drop in instrumentation, then a build back up |
| Outro | return to a sparse texture, or a final held chord |
Choosing tempo, key and mood words
Contemporary worship generally sits between 70 and 130 BPM: slower, more reflective songs around 70 to 90 BPM, and more energetic, full-band songs around 110 to 125 BPM. Major keys (C major, G major) tend to read as hopeful or triumphant; minor keys (A minor) suit more reflective or pleading songs. Pick one mood word (hopeful, tender, triumphant, reflective) rather than stacking several, since mixed moods confuse the model.
Using section tags for structure
Square-bracket tags such as [Verse], [Chorus], [Bridge] and [Outro] are a common convention that models follow often but not always, so pairing them with a written description of the build (as in the prompts above) gives the clearest result. If you want a fully instrumental section, such as a quiet moment for prayer, write [Instrumental] and say "no drums" or "piano and strings only" so vocals do not appear unexpectedly.
Lyrics skeleton
Original, for structure
[Verse] In the quiet I am found Steady hands upon this ground [Chorus] Here we stand with open hands Light is breaking, still we stand [Bridge] Every valley, every height Carried through the darkest night [Outro] Still we stand
Common mistakes
- Writing only "worship song" with no arrangement, which gives a generic ballad: describe the piano-to-band build instead.
- Stacking many moods (peaceful, epic, sad, joyful) in one prompt, which confuses the model: pick one dominant mood per section.
- Forgetting to describe the voice, which often defaults to a single soloist: say "layered voices" or "group of voices" for a congregational feel.
- Putting the lyrics inside the style or prompt box instead of the lyrics box: keep sound description and words separate.
- Expecting a dramatic bridge without describing contrast first: say what drops out before you say what comes back in.