ElevenLabs Music v2
ElevenLabs' text-to-music model — studio-grade songs with multilingual vocals, section-level editing, and training built entirely on licensed data.
curl https://api.venice.ai/api/v1/audio/queue \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "elevenlabs-music-v2",
"prompt": "An uplifting cinematic orchestral build with soaring strings"
}'
# Use the returned queue_id with https://api.venice.ai/api/v1/audio/retrieve.
# Call /audio/complete after downloading if needed.Overview
What is ElevenLabs Music v2
ElevenLabs Music v2 is ElevenLabs' text-to-music model, released in May 2026. It generates studio-grade full songs with vocals or instrumentals from natural-language prompts, sings in multiple documented languages, handles mid-track genre transitions, and lets you regenerate individual sections via inpainting — all trained only on licensed data and cleared for commercial use.
Using it anonymously on Venice
On Venice, ElevenLabs Music v2 runs under the anonymized privacy tier — your prompts and lyric drafts are not stored, profiled, or used for training, so unreleased lyrics and commercial briefs stay yours. There is no subscription to manage: you pay per track, from $0.69 for a 60-second clip, scaling with length. The model is permissionless to start with — no credit card, no generation history attached to your identity.
Specifications
Datasheet
- Maker
- ElevenLabs
- Modality
- Text-to-music — complete songs with vocals or instrumental
- Open weights
- No — proprietary
- License
- Proprietary
- Released
- May 26, 2026
- Singing languages
- Multilingual, including English, Spanish, German, Japanese and more
- Audio reference
- Upload a reference track of up to ~30 seconds to guide style, instrumentation, tempo and mood
- Training data
- Licensed data only, created in collaboration with labels, publishers and artists
- Privacy on Venice
- Anonymized — prompts not stored
- Available on Venice since
- Sep 2026
Assessment
Strengths and limitations
- The clearest rights story in AI music: trained only on licensed data and cleared for nearly all commercial uses, from ads and podcasts to gaming.
- Strong vocal delivery: it sustains fast rap and dense lyrical delivery and keeps supplied lyrics intact through a generation.
- Granular section control: improved inpainting lets you regenerate just the bridge without touching the chorus.
- Long-form composition: build a full song section by section (intro, verse, chorus and beyond) with structure maintained throughout.
- Handles dramatic mid-track genre shifts (opera to heavy metal and back) and embeds non-musical sound effects without breaking musical coherence.
- A documented, published language list for vocals: English, Spanish, German, Japanese and more — rather than vague multilingual claims.
- Closed and proprietary: no open weights, so it cannot be self-hosted or fine-tuned.
- Arrangements lean conservative on busy genres; reviewers find it safer and more predictable than Suno or Udio rather than more adventurous.
- Fine timing is not always exact: instrumental hits can land early, breaks can run long, and requested endings may resolve as fades.
- Commercial clearance has asterisks: film, TV and studio-game use requires an Enterprise plan, and marketplace tracks cannot be distributed to streaming services — read the music terms before building a business on it.
Use cases
What it is good for
- 01Ad and brand soundtracks where licensing certainty matters more than experimental arrangements.
- 02Background music for podcasts, social video and product demos that must sit under a voiceover.
- 03Full songs with original lyrics in English, Spanish, German, Japanese and other supported languages.
- 04Iterative songwriting — regenerate a weak bridge or chorus while keeping the rest of the track intact.
- 05Game and app audio via API, with per-track pricing that scales with clip length.
Prompting
Getting better results
Tag sections explicitly — [Intro], [Verse], [Chorus], [Bridge] — and Music v2 will follow the structure you lay out.
Put your exact lyrics in the prompt and name the language; the model preserves supplied words and sings in documented languages.
Specify genre, tempo, key and instrumentation in plain musical terms — it understands terminology, not just vibes.
Upload a ~30-second audio reference when you want the output to match a particular production style, tempo or mood.
For long songs, describe the arc section by section (intro → verse → chorus → bridge) instead of asking for 'a full song' in one phrase.
Request mid-track shifts explicitly ('transitions from opera to heavy metal at the bridge') — v2 handles genre changes that break other models.
Samples
Sample outputs
Generated on Venice with our standard prompt suite — the same prompts we run through every model of this type, so you can judge it like-for-like.
“An uplifting cinematic orchestral build with soaring strings, warm brass, and a hopeful resolution.”
“A mellow lo-fi hip-hop beat with a soft jazzy piano loop, vinyl crackle, and a relaxed late-night mood.”
Alternatives
How it compares
| Model | Best for | Voices | Open weights | Price (Venice) |
|---|---|---|---|---|
| ElevenLabs Music v2 | Licensed, commercial-safe songs with vocals | Multilingual vocals (EN, ES, DE, JA + more) | No | from $0.69 / track |
| ElevenLabs Music v2.5 | Newest quality tier from ElevenLabs | Multilingual vocals | No | from $0.69 / track |
| ACE-Step 1.5 | High-volume, budget music generation | Vocals and instrumentals | No | from $0.03 / track |
| ElevenLabs Sound Effects | SFX and stingers, not songs | None — sound effects only | No | $0 / sec |
ElevenLabs Music v2 is the right pick when the track ships commercially and licensing risk is the dealbreaker — cleared training data, documented languages, and per-track pricing that scales with length. For bulk drafts on a budget, ACE-Step 1.5 is the cheaper engine.
API
Call it from your code
Venice exposes this model through the REST API. Queue a generation with the model id.
curl https://api.venice.ai/api/v1/audio/queue \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "elevenlabs-music-v2",
"prompt": "An uplifting cinematic orchestral build with soaring strings"
}'
# Use the returned queue_id with https://api.venice.ai/api/v1/audio/retrieve.
# Call /audio/complete after downloading if needed.Pricing
What it costs on Venice
Pay per track on Venice — price scales with length, from $0.69.
New Venice accounts include a free daily allowance and 500 welcome credits — no credit card required.
FAQ
Frequently asked questions
ElevenLabs Music v2 is a text-to-music model released by ElevenLabs in May 2026. It generates studio-grade full songs with vocals or instrumentals from natural-language prompts, supports multilingual singing, mid-track genre transitions, embedded sound effects, and section-level inpainting, and was trained only on licensed data.
You pay per track, with the price scaling by length: $0.69 for a 60-second clip, $1.38 for 120 seconds, $2.08 for 180 seconds, $2.76 for 240 seconds, $3.45 for 300 seconds, and $4.15 for 360 seconds. There is no subscription required.
It is neither free-software nor open-weights: the model is proprietary to ElevenLabs and cannot be self-hosted or fine-tuned. On Venice you pay per generated track rather than a flat fee, and new accounts can start without a credit card.
Yes — it was trained only on licensed data in collaboration with labels, publishers and artists, and is cleared for nearly all commercial uses including ads, podcasts, social video and gaming. Note that film, TV and studio-game use requires an Enterprise arrangement with ElevenLabs, and marketplace tracks cannot be distributed to streaming platforms, so check the music terms for your specific use.
It is multilingual with a documented language list that includes English, Spanish, German and Japanese, among others. ElevenLabs is unusual in publishing exactly which languages its vocals support, which matters when a mangled non-English vocal is not acceptable.
Yes. You can upload a reference track of up to approximately 30 seconds in common audio formats to guide the sound, production style, instrumentation, tempo and mood of a generation. Uploaded references are screened for copyright compliance before use.
Yes. Improved inpainting lets you select any section of a track and regenerate just that part — tweak the bridge without touching the chorus, leaving everything else exactly as it was.
Both produce broadcast-usable tracks. ElevenLabs Music v2 wins on rights certainty (licensed training data, commercial clearance) and section-level control; Suno offers a deeper production toolkit with stems, a multitrack editor and longer tracks. A common workflow is drafting in Suno and producing the final cleared deliverable in Music v2.
Venice runs it under the anonymized privacy tier: prompts and lyrics are not stored, profiled, or used for training, and generations are not tied to a personal history. That matters when you are prompting with unreleased lyrics or client briefs.
Use ElevenLabs Music v2 anonymously
Venice does not store your prompts. Chat history stays in your browser.