ElevenLabs Music v2.5 Prompt Guide: 10 Tips + Copy-Paste Templates

You type upbeat electronic track and hit generate. Thirty seconds later you have a song that sounds like every other upbeat electronic track. Nothing is wrong with it, and nothing is yours. ElevenLabs says why on its own best-practices page: any question you leave open gets answered with “the most average one.”

People searching ElevenLabs Music v2.5 prompt already picked the model. What they want is a string they can paste. This page is that: ten tips, each a full block, then five templates for songs that will end up under a music video.

The facts below come from ElevenLabs’ own docs, read on 2026-10-28. Where the docs say nothing (for example, plan fields inside a composition plan), this page says nothing either.

Table of Contents

• What the docs say v2.5 takes
• 10 copy-paste tips
• 5 templates for music-video songs
• Vague prompt vs a v2.5 prompt
• FAQ
• Copy, pick the model, make the MV

What the docs say v2.5 takes

Per ElevenLabs’ music capability page:

• Select it in the API with model_id="music_v2_5". Music v2 stays available.
• Length runs from 3 seconds to 5 minutes.
• Prompts can be natural language; the model also understands musical terminology.
• Output can have vocals or be instrumental only. Multilingual output covers English, Spanish, German, Japanese and more.
• v2.5 supports composition plans (sections, global and local styles, lyric formatting), and you can regenerate one section without touching the rest.
• Audio Reference (paid plans) takes a clip of about 30 seconds to steer sound, mood, instrumentation, tempo and production. It does not copy or remix the audio, and every reference is screened for copyright.
• The docs call Eleven Music “cleared for nearly all commercial uses” and point to the music terms for plan-specific details. Read those terms before a client ad.

On SunoMV the same model is the ElevenLabs Music v2.5 row in the picker, at 40 credits per song. For how it compares with Suno V6, see ElevenLabs Music v2.5 vs Suno V6.

Image: SunoMV Team · the prompt is the only part of the studio you control before the first listen

Official launch clip, in ElevenLabs’ own words. Source: YouTube / ElevenLabs, 2026-09-11.

Practical rule: Decide genre, mood, instruments, tempo and era yourself. ElevenLabs’ own line: “Decide genre, mood, instruments, tempo, and era before the model decides for you.”

10 copy-paste tips

Every example below is quoted or adapted from the ElevenLabs best-practices page. The anti-example is the same intent left vague.

1. Name genre, BPM and the sound, not a category

French house, 122 BPM, filtered disco sample chops, sidechained bass, tight
four-on-the-floor kick, rooftop-at-sunset mood, warm analog glue

Anti-example: upbeat electronic track

The docs use exactly this pair as weak versus strong. Tempo is worth stating: the page says the model holds a stated BPM and key precisely enough to layer with other material.

2. Add a key when you will cut to it

synthwave arpeggio anthem, F minor, 128 BPM, neon nostalgia, analog polysynth pads

Anti-example: 80s vibe

A stated key like “in A minor” is also worth writing when you plan to layer a second part later.

3. Use production words to change the mix

Same ballad, three treatments, taken from the docs:

slow soul ballad, 68 BPM, female vocal — bone-dry drums, close-mic'd vocal, dead room
slow soul ballad, 68 BPM, female vocal — cavernous plate reverb, tape echo throws, gospel room
slow soul ballad, 68 BPM, female vocal — tape saturation, wow and flutter, dusty vinyl crackle

Anti-example: slow soul ballad, nice sound

If you lack the studio terms, the docs say to describe the space instead.

4. Narrate the arrangement in order

UK garage, 132 BPM — start with just a shuffled drum loop, add a warm sub bassline
after four bars, then bring in chopped vocal stabs for the drop

Anti-example: UK garage with a good drop

Order is the lever. For a music video, this is how you get a verse that is quiet enough for dialogue-free visuals and a chorus that lands on a cut.

5. Say how the vocal is delivered

whispered indie folk, close and intimate, fingerpicked acoustic guitar

stadium rock, belted powerhouse vocal, huge drums, wall of guitars

Anti-example: rock song with singing

For harmonies, the docs’ phrasing is “two singers harmonizing in C”.

6. Write your own lyrics, or let it write them

Most music prompts include lyrics by default. If you write lyrics, the model combines them with the prompt length to decide vocal structure and placement. If you do not, it generates structured lyrics that fit the length.

indie pop, 104 BPM, bright guitars, female vocal
lyrics begin at 15 seconds

Anti-example: a song about missing someone (and then complaining the chorus is generic)

The docs give timing cues such as “lyrics begin at 15 seconds” and “instrumental only after 1:45”. Use them to leave room for an opening shot.

7. Ask for instrumental only when the video needs room

dreamy guitar loop, 140 BPM, F sharp minor, emo trap, four bars, instrumental

Anti-example: background music (you get a vocal and a story you did not ask for)

The docs say to add “instrumental only”, and for loops to state “instrumental” and exclude melody explicitly: boom bap drum break, 90 BPM, dusty and swung, four bars, no melody — just drums.

8. Set a length instead of accepting one

State a duration such as “60 seconds”, or use auto mode. The model accepts 3 seconds to 5 minutes.

cinematic pop, 96 BPM, 60 seconds, instrumental intro, lyrics begin at 15 seconds

Anti-example: short song

For a clip-length MV hook, 15 to 30 seconds is a common ask; whatever you pick, write the number.

9. Isolate a part with “solo” or “a cappella”

solo piano in C minor

a cappella vocals in A major, 90 BPM, soulful and raw

Anti-example: just the piano part please

The docs recommend “solo” before instruments and “a cappella” before vocals, with key, tempo and tone added for better results. The same page notes that the v1 model does not generate stems from a full track, so this is a prompting technique, not a split of an existing song.

10. Iterate on the prompt instead of starting over

ElevenLabs’ help center says small changes can have a big impact. Change one thing per take.

Take 1: slow soul ballad, 68 BPM, female vocal — bone-dry drums
Take 2: slow soul ballad, 68 BPM, female vocal — cavernous plate reverb

Anti-example: rewriting the whole prompt each time, so you cannot tell what helped.

Practical rule: A good prompt has genre, BPM, a production word, an arrangement order and a vocal instruction. Missing slots become the most average choice.

5 templates for music-video songs

Image: SunoMV Team · write the take, listen, change one thing

Fill the brackets and keep the structure. These are adapted from the docs’ examples toward music-video use. They are starting points, not guaranteed results.

Pop chorus-first (short-form MV)

bright indie pop, 110 BPM, [key], clean electric guitar, claps, female vocal, close-mic'd
start with a four-bar instrumental, lyrics begin at 8 seconds, chorus by 20 seconds
duration: 45 seconds

Rain-and-neon synthwave (instrumental bed)

synthwave arpeggio anthem, F minor, 120 BPM, neon nostalgia, analog polysynth pads
instrumental only, builds for sixteen bars, then drops the arpeggio
duration: 60 seconds

Intimate acoustic (lyric close-ups)

whispered indie folk, close and intimate, fingerpicked acoustic guitar, 72 BPM
male vocal, dry and close-mic'd, no drums until the second verse
lyrics begin at 6 seconds

Stadium rock (big finish)

stadium rock, belted powerhouse vocal, huge drums, wall of guitars, 138 BPM, E minor
start with just drums and bass, add guitars after four bars, vocal enters on the downbeat

Drum-loop sketch (to build on)

boom bap drum break, 90 BPM, dusty and swung, four bars, no melody — just drums

Swap the brackets for your own key and tempo.

Practical rule: A template is finished when genre, BPM, key (if layering), vocal delivery, arrangement order and a duration are all on the page.

Vague prompt vs a v2.5 prompt

	Vague	Copy-paste
Genre	“upbeat electronic”	named subgenre + BPM
Sound	“nice mix”	production words or a described space
Structure	“has a drop”	“start with…, after four bars, then…”
Vocals	“with singing”	delivery style, or “instrumental only”
Lyrics timing	(hope)	“lyrics begin at 15 seconds”
Length	“short”	“60 seconds”
Iteration	rewrite everything	change one thing per take

FAQ

Do I write “ElevenLabs Music v2.5” in the prompt?

No. Choose the model in the picker (or model_id="music_v2_5" in the API) and keep the prompt about the music.

How long can a song be?

ElevenLabs lists 3 seconds as the minimum and 5 minutes as the maximum.

Can I get a track with no vocals?

Yes. Add “instrumental only” to the prompt.

Is the output cleared for commercial use?

The docs describe Eleven Music as “cleared for nearly all commercial uses” and point to the music terms for plan details. Check those terms for your plan before shipping a client ad.

Can I upload a reference track?

On paid plans, Audio Reference accepts a clip of about 30 seconds to guide sound, mood, instrumentation, tempo and production. It is not designed for genre transformation and it does not copy the audio.

Does this page measure which model sounds better?

No. These tips come from ElevenLabs’ documentation, not from a matched-prompt listening test.

Copy, pick the model, make the MV

1. Copy one block and fill the brackets. Keep genre, BPM, an arrangement order and a duration.
2. Open SunoMV, pick ElevenLabs Music v2.5, and paste the block.
3. Listen, change one thing, regenerate. When the song is right, turn it into a lyric-synced video with the audio-to-video generator.

For the licensing and length comparison with Suno V6, read ElevenLabs Music v2.5 vs Suno V6. For the video side of the workflow, see the AI music video creation guide.