ElevenMusic and ElevenCreative Guide 2026: Music v2, References, Finetunes, Rights and SynthID

Jack Righteous AI Music Platform Guide

ElevenLabs has moved beyond basic text-to-music generation. Music v2 now combines stronger prompt following, Audio Reference, longer section-based composition, improved vocals, inpainting and personalized Music Finetunes with ElevenLabs’ wider voice, dubbing and sound-effects ecosystem.

Originally published July 27, 2026 · Fully updated July 30, 2026

Eleven Music, ElevenMusic and ElevenCreative: the names explained

The naming is confusing because ElevenLabs uses “music” in several connected contexts. The easiest way to understand the ecosystem is to separate the company, model, workspace, audience platform and developer layer.

An ElevenCreative subscription does not necessarily become an ElevenMusic platform subscription. Confirm which product, account and plan you are using before assuming a feature or commercial permission carries across the entire ecosystem.

What changed with Music v2?

Music v2 is not simply a quality upgrade. It changes how a creator can direct and revise a song. ElevenLabs says the model improves prompt adherence, composition, multilingual output and vocal delivery while adding Audio Reference, section-by-section composition, mid-track genre transitions, stronger inpainting and embedded sound effects.

Stronger prompt understanding

A prompt can now function more like a compact production brief. Describe the purpose, emotional movement, tempo, instrumentation, vocal character, arrangement and transitions rather than relying on a genre label alone.

Longer, section-based composition

Creators can build intros, verses, choruses, breakdowns and outros with more deliberate changes across a track. This is particularly useful for film, games, advertising, podcasts and theatrical storytelling where the music must evolve with a scene.

More capable vocals

Music v2 is designed for faster delivery, denser lyrics, rap and multilingual performances. The model can perform more complex material, but lyric quality still determines whether the result sounds intentional. AI can deliver a weak line more fluently; it cannot make that line meaningful.

Mid-track transitions and embedded effects

A composition can shift genre, energy or instrumentation inside the same track. Sound effects can also be placed inside a musical generation, creating possibilities for trailers, games and narrative audio. Complex transitions should be requested explicitly and tested section by section.

Music length and exports

ElevenLabs currently lists generated durations from approximately three seconds to five minutes, with MP3 and WAV exports. Music v1 remains available during the model transition, but Music v2 is the newer default in the website interface.

What can Eleven Music generate?

• complete songs with vocals;
• instrumentals and background music;
• cinematic and theatrical cues;
• podcast themes and beds;
• advertising and social-media music;
• game music and transitions;
• multilingual songs;
• experimental genre blends;
• short effects-led pieces and longer section-based compositions.

A productive prompt identifies the intended use first. “Dark reggae” is a category. “A tense, restrained reggae-dub cue for a detective entering an abandoned church, beginning with hand percussion and bass before a distorted violin rises in the final third” is a direction.

How Audio Reference works

Audio Reference lets a creator upload approximately 30 seconds of audio to guide Music v2. ElevenLabs says the reference can influence sound, production style, instrumentation, tempo and mood. The upload is screened for copyright compliance.

A responsible reference workflow

1. Choose a recording you created, own or are clearly permitted to use.
2. Select the short passage that best demonstrates the relevant traits.
3. Upload the reference and add a written prompt explaining what matters.
4. Generate several versions and compare them against the requested traits, not against the identity of the source song.
5. Preserve the reference file, ownership evidence, prompt and chosen output.

Strong reference sources

• your own demo or instrumental;
• a collaborator’s recording covered by a clear agreement;
• a licensed reference that permits this use;
• a custom rhythm, texture or mood study;
• a track from your own catalogue.

Higher-risk reference sources

• a commercial recording you do not control;
• a famous singer’s isolated vocal;
• a song selected primarily for a recognizable melody;
• a master recording with unclear label or publisher rights;
• audio deliberately altered to avoid platform screening.

Music Finetunes: training ElevenMusic on an authorized sound

A Music Finetune is a personalized model trained from music the user is authorized to upload. ElevenLabs says it can learn patterns involving instrumentation, production character, tempo, timbre, arrangement and vocal character, with a model typically prepared in approximately five to ten minutes.

The difference from a prompt is important. A prompt requests traits for one generation. A Finetune grounds future generations in recurring patterns learned from a body of authorized work.

Who should consider a Finetune?

• artists with an established catalogue and recognizable production identity;
• brands requiring consistent sonic direction;
• game, film and podcast teams producing related assets;
• agencies creating recurring campaign music;
• production libraries developing a coherent collection.

What should not be used to train one?

• commercial songs you merely streamed or purchased;
• beats with unclear ownership;
• collaborator recordings without written permission;
• licensed music where model training was never granted;
• unauthorized celebrity or identifiable voices.

ElevenLabs also provides curated Finetunes, with examples including Afro House Beats, Reggaeton, Arabic Groove, 70s Cambodian Rock, 80s Nu-Disco Revival and Mozart-style symphonic music. These can offer stylistic consistency without uploading a private catalogue.

Editing, inpainting and human contribution

Generation is only the beginning of a professional workflow. Music v2 can regenerate a selected portion while preserving the rest of the composition. A creator can revise lyrics, add or remove a section, repair a transition, change a local style or replace a weak instrumental passage.

The strongest editing process is selective:

1. generate the complete concept;
2. identify the weakest specific section;
3. change one problem at a time;
4. save a version before every major revision;
5. move into BandLab or another DAW when generation is no longer the main problem.

Those decisions form part of the creator’s human contribution. Keep records of what you rejected, rewrote, regenerated, rearranged, mixed and mastered. The output file alone cannot explain the creative path.

Reusable vocals and creator identity

A reusable Vocal is intended to preserve singer identity across songs. It is not the same as a Reference or a Finetune. A Reference guides the musical direction of one generation. A Vocal focuses on the performer. A Finetune learns from a broader catalogue.

Use your own voice, a platform voice or an authorized collaborator. Reusability does not create ownership of another person’s identity. For the deeper consent and rights questions, read Who Owns a Voice in AI Music?

ElevenMusic rights and commercial use

ElevenLabs describes Eleven Music as cleared for nearly all commercial uses, including film, television, podcasts, social media, advertising and gaming, while directing users to the Music Terms for plan-specific details.

That statement should not be converted into a universal ownership guarantee. Commercial use still depends on several separate layers:

• the account plan and terms active when the generation was created;
• rights in lyrics, scripts and uploaded references;
• voice consent and performer permissions;
• collaborator and client agreements;
• distributor and marketplace policies;
• copyright and publicity law in the relevant jurisdiction.

Before commercial release, verify

1. the active plan and generation date;
2. the applicable ElevenLabs terms;
3. ownership or permission for every reference;
4. voice and performer consent;
5. lyric, script and composition rights;
6. collaborator and client assignments;
7. distribution and disclosure requirements;
8. your human contribution record.

Use the AI Music Rights and Ownership Guide to separate platform permission from copyright, voice rights and distribution eligibility.

Can you publish and earn from ElevenMusic songs?

Potentially, yes. A creator may be able to publish and commercially use ElevenMusic output under the applicable plan and terms. Publishing eligibility, however, is not decided by ElevenLabs alone.

A streaming distributor, client, production library or licensing marketplace may apply its own rules. A valid ElevenLabs commercial permission does not erase questions about uploaded references, cloned voices, collaborators or required AI disclosure.

Possible earning paths

• creator-owned releases;
• film, game and podcast licensing;
• advertising and social-media production;
• custom client commissions;
• production libraries;
• ElevenMusic platform engagement or royalty programs;
• ElevenCreative Music Marketplace licensing.

Marketplace participation is not a copyright registration, a guaranteed sale or automatic streaming distribution. Package each track around a clear use—such as cinematic tension, upbeat advertising, gaming loops, podcast themes or a consistent branded sound—rather than publishing an undifferentiated volume of generations.

For a deeper business breakdown, read How to Make Money With ElevenLabs in 2026.

ElevenMusic versus Suno

Neither platform eliminates the need for post-production, selection and rights records. For a multi-tool production approach, see Beyond Suno: ElevenLabs and BandLab AI Music Workflow.

Who ElevenMusic is best for

The Jack Righteous beginner workflow

1. Define the use. A release, advertisement, scene, podcast or client asset requires different music.
2. Write the brief. Specify mood, tempo, instruments, vocals, structure, transitions and exclusions.
3. Choose Music v2. Use the current model unless a controlled comparison requires v1.
4. Decide whether a Reference is necessary. Do not upload audio simply because the feature exists.
5. Generate and compare. Judge prompt adherence, emotional movement and production usefulness.
6. Edit the weak section. Preserve what already works.
7. Document and finish. Save the source, rights information, human contribution and final master.

The ElevenMusic production record

Frequently asked questions

Official sources and related guides

• ElevenLabs Music capability documentation
• ElevenLabs Music Terms
• ElevenLabs SynthID Watermark Guide
• Who Owns a Voice in AI Music?
• AI Music Rights and Ownership Guide

Educational notice: This guide reports current product features and public terms for creators. It is not individualized legal, financial or tax advice. Platform features, plans and terms can change.