Prompt generation
Generate music from a creative brief
- Input
- Genre, mood, instruments, tempo, use, and lyric direction
- Output
- A vocal song or instrumental track
| A vocal song or instrumental track |
| Control verses, choruses, and other sections | Composition plan | Section order, lyrics, style, and duration plan | A structured long-form composition |
| Guide a track with your own sound | Audio Reference | Reference audio and optional text direction | A new composition guided by sound, instrumentation, tempo, and mood |
| Regenerate one section without restarting | Inpainting | An existing track, target section, and revision direction | A revised section while the rest stays intact |
Prompt generation
Composition plan
Audio Reference
Inpainting
Music v2 supports several ways to move from an idea to a finished track. Generate from text, build a song section by section, use reference audio to guide the sound, or revise one selected passage.

Music v2
Prompt generation is the direct entry point. Describe genre, era, mood, tempo, instrumentation, vocal delivery, language, structure, and intended use, or request an instrumental track.
Best for: Branded content, creator soundtracks, song drafts, and in-product generation
Long-form composition
Music v2 can build a song through an intro, verses, choruses, bridges, and an outro. Sectional work is useful when lyrics and arrangement need more deliberate control than a single free-form prompt.
References and inpainting
Audio Reference guides sound, production style, instrumentation, tempo, and mood without copying the source. Inpainting regenerates a selected section while leaving the rest of the song in place.
State whether the track is for an ad, podcast, short video, game, product, or songwriting, then identify the listener, duration, and release context.
Describe genre, mood, tempo, meter, instruments, vocals, language, production texture, and dynamics instead of relying on one broad style label.
Use a concise prompt for quick music, or plan intros, verses, choruses, bridges, lyrics, and section goals for a complete song.
After generating, compare the melody, rhythm, vocals, lyrics, and transitions, then continue refining a chorus, bridge, or another specific section.
Compare the melody, lyrics, vocals, and arrangement, then select the version that best fits the intended use and complete the track.
Before you begin, decide where the music will be used and how it should sound, then prepare the lyrics, song sections, and any reference audio. These five steps will help turn the idea into a clear brief.
01
List the use, audience, emotional arc, duration, tempo, instrumentation, vocals, language, structure, and excluded elements.
02
For vocals, decide language, rhyme, point of view, sections, and sensitive wording. A full song can map intro, verses, choruses, bridge, and outro.
03
Use your own demo, melody sketch, or production reference to show the rhythm, mood, instrumentation, and overall sound you want.
04
Compare versions across melody, lyrics, pronunciation, brand fit, audio quality, and expression, then choose the one that best matches the intended use.
05
Keep the prompt, lyrics, and audio for each version so you can compare melody, vocals, and arrangement before choosing what to publish.
Generate a vocal song or explicitly request an instrumental. A brief can combine genre, mood, instruments, tempo, structure, and intended use.
Music v2 handles fast rap, dense lyrics, range changes, and more complex vocal delivery while improving multilingual lyrics and vocals.
Organize intros, verses, choruses, bridges, and outros while maintaining structure and continuity as a song grows.
Music v2 can handle style changes within one track and place non-musical effects inside the composition for narrative or scene changes.
Reference audio can guide the sound, production style, instrumentation, tempo, and mood so the result stays closer to the intended musical direction.
Regenerate a selected passage to revise a bridge or chorus without rebuilding the entire song from the beginning.
Create custom music from brand voice, audience, edit pace, and emotional arc for advertising, product launches, and branded content.
Create intros, outros, transitions, and background beds with defined instruments, rhythm, and memorable motifs for a repeatable channel identity.
Prototype level music from scene, tension, character, and pace while exploring different moods and rhythmic directions for an interactive experience.
Combine lyrics, section structure, vocal direction, and instrumentation into a draft for discussing melody, rhythm, arrangement, and production.
Prepare lyrics and pronunciation guidance for each language, compare vocal delivery and arrangement, then choose the version that best fits the intended audience.
Add music generation to creator tools, content platforms, or interactive apps so users can make songs and soundtracks from a description, lyrics, or scene brief.
Create a 35-second instrumental electronic track for a premium technology product launch. Start with restrained glassy synth pulses, build with precise percussion and warm sub-bass, then resolve into a confident three-note motif. Modern, spacious, no vocals, no retro arcade sounds.The use, duration, instrumentation, dynamic arc, motif, and exclusions provide more direction than a generic request for futuristic electronic music.
Write a Mandarin city-pop song about leaving the office late and choosing to walk home through summer rain. Female alto vocal, 112 BPM, clean electric guitar, warm analog synths, melodic bass, restrained verses and a bright sing-along chorus. Use natural contemporary Mandarin and avoid cliché neon imagery.Narrative, language, vocal range, tempo, instrumentation, section contrast, and a banned cliché all help shape the vocal and lyrical result.
Create a 12-second instrumental intro for a weekly science podcast. Curious but credible, 96 BPM, marimba accents, muted drums, soft modular synth texture, and a clean ending that leaves room for a spoken title. No cinematic boom and no vocals.A short intro should reserve space for speech and define the ending so it remains practical to edit.
Compose a seamless-feeling two-minute track for an underwater exploration level. Slow 6/8 pulse, bowed glass, low strings, distant metallic percussion and occasional whale-like synthetic tones. Begin calm, introduce unease after the first third, and end unresolved. Instrumental, immersive, not horror.Scene, meter, instrumentation, timed progression, and an emotional boundary define level music more effectively than a genre label alone.
You can prepare your account and API key now, then start using Eleven Music v2 after it launches.
Use one account for balance, API keys, task records, and models that are already live.
Keep the key in a server-side environment variable or secret manager, never browser code, screenshots, or a public repository.
Use MiniMax Music 2.6 to test music briefs, task submission, status checks, audio playback, and usage records.
Eleven Music v2 is being integrated
You can validate your brief with a music model already available on HiAPI and enable the Music v2 launch reminder to be notified as soon as it launches.
Sign up for 200 Credits
Use your trial credits on music and voice models that are already available.


Generate complete songs from a music brief and lyrics while comparing styles, languages, and vocal directions.

Create quick song drafts across languages, styles, and vocal directions from text.

Generate multi-speaker dialogue, character voices, and narration for spoken parts around a music project.
Eleven Music v2 is ElevenLabs' next-generation AI music model for vocal songs and instrumentals, with more complex vocals, multilingual output, long-form sectional composition, style changes, references, and inpainting.
Not yet. HiAPI is completing the Music v2 integration and launch checks. This page currently provides prelaunch information and a launch reminder, without task submission, pricing, or API parameters.
ElevenLabs music products and APIs cover short clips through long-form works. The actual duration range on HiAPI will be published in the final documentation after launch.
Yes. Music v2 supports vocal songs with lyrics, multilingual singing, fast rap, and dense lyrics, as well as instrumental music. Clear lyrics, language, vocal direction, and structure make the target more specific.
It builds and extends a song through sections such as intro, verse, chorus, bridge, and outro instead of stopping at a short clip. This helps with complete lyrics, defined structure, and an evolving arrangement.
No. Audio Reference guides sound, production style, instrumentation, tempo, and mood while generating a new piece rather than directly copying or remixing the reference track.
Inpainting regenerates a selected passage such as a bridge, chorus, or lyric section while keeping the rest of the song. A revision brief should identify the section, problem, and intended change.
Include the use, audience, genre, mood, tempo, meter, instruments, vocals, language, structure, dynamics, and exclusions. For lyrics, add the theme, point of view, sections, and writing style.
Commercial use depends on the selected ElevenLabs plan and distribution context. Check the latest Music Terms for the usage scope that applies to advertising, streaming, film, television, or games.
MiniMax Music 2.6 already generates songs from a music brief and lyrics, making it useful for comparing language, vocal, and arrangement directions. ElevenLabs Text to Dialogue covers spoken dialogue, character voices, and narration.