ElevenLabs has released Music v2, a major update to its music generation model. The model can now switch genres mid-track, reconstruct isolated sections without touching the rest, and embed non-musical sound effects directly into a composition. An architecture that fundamentally changes what is possible to create with a prompt.
Key Takeaways
- ElevenLabs Music v2 can transition from opera to heavy metal and back within a single track
- Section-based editing lets users regenerate a specific part without altering the rest of the composition
- The model is available on ElevenCreative and ElevenMusic, with API access coming soon
A Section-Based Architecture That Changes Everything
Music v2 arrives nearly ten months after the first version, which launched in August 2025. That is a short gap for an audio generation model, but the advances are substantial. The most striking capability: mid-track genre transitions. ElevenLabs demonstrates a shift from opera to heavy metal and back, without artificial breaks. The model also maintains coherence through fast rap across an entire track, which is a distinct technical challenge from stylistic transitions.
Another feature reshapes the creative workflow entirely: section-based editing. A user can select a specific portion of a track, say a chorus or a verse, and regenerate it using a prompt without affecting the rest of the composition. This level of editorial granularity brings the tool closer to traditional audio production software. Until now, music generation models required regenerating an entire track for every modification.
The model also supports non-musical sound effects embedded directly into a composition, opening the door for content creators who need immersive soundscapes without stitching together multiple sources. ElevenLabs says Music v2 performs more reliably across multiple dimensions at once: languages, lyrics, vocals, and arrangements. A step toward overall coherence rather than peak performance on a single metric.
The composition process itself has also evolved. Users can now build a track section by section, generating the intro, verses, and chorus separately before combining them. A modular approach that mirrors how music producers actually work, rather than treating the song as a single undifferentiated prompt.
Licensing as a Competitive Weapon
ElevenLabs states that Music v2 is trained on licensed data and cleared for commercial use. The distinction matters. Suno and Udio, two direct competitors in AI music generation, are facing lawsuits from major labels over alleged unauthorized use of copyrighted recordings. ElevenLabs is explicitly positioning itself as the legally secure option for businesses that cannot afford the litigation risk.
The model is available now on ElevenMusic, the company’s new platform for AI-generated songs, and on ElevenCreative, its tool for marketing and branding teams. API access has been announced for the near future, which will open integration into automated production pipelines.
The company crossed $330 million in annual recurring revenue at the end of 2025. In that context, Music v2 is not a standalone product but one more layer in an audio ecosystem that already covers voice synthesis, transcription, sound effect generation, and voice cloning. ElevenLabs is working to cover the entire audio pipeline, from voice to music to immersive soundscapes.
Competition is moving quickly. Google, Stability AI, and Suno have all released competing models in recent weeks, featuring extended generation lengths and new editing capabilities. The AI music generation market is consolidating fast around a small number of players who are setting their own technical standards.
Also on Horizon:
- Meta One: Paid Subscriptions Land Across All Platforms
- Zara and H&M Are Using AI to Replace Their Models
- DuckDuckGo Up 30%: Users Reject Google’s Forced AI Search
What Music v2 Signals for Music Creation
In the short term, the immediate beneficiaries are marketing teams and video content creators. Having commercially licensed, section-editable, genre-flexible music on demand removes a major friction point in content production. Royalty-free music libraries built a multi-hundred-million-dollar market around that same promise. Music v2 makes it available on demand at marginal cost.
The reach of AI audio already extends well beyond music. AI voice reconstruction capabilities pushed the NTSB to shut down its public dockets to protect the identity of crash victims. Music v2 illustrates the same trajectory: audio models are gaining precision and control at a pace that legal and ethical frameworks struggle to keep up with.
Over the medium term, the arrival of the API is the most significant signal. Once Music v2 is available as a direct integration, post-production studios, creative agencies, and streaming platforms will be able to automate entire sections of their music pipeline. Functional music composers, already under pressure from royalty-free libraries, will face a new wave of automated competition on short formats and budget-constrained projects.
AI music generation does not yet replace artists on long-form or high-identity projects. But it progressively narrows the perimeter where a human remains irreplaceable. Music v2 is not a breaking point. It is one more step in the ongoing compression of time and cost in audio production.
Follow the story on Horizon.
