What happened
Voice AI company ElevenLabs has launched Music v2, a new version of its music-generation model. According to TechCrunch, the model can switch genres mid-track — from opera to heavy metal, for instance — and add non-musical sound effects.
Artists can now build a song by composing an intro, verse, and chorus separately and then stitching them together. The model is built on licensed data, cleared for commercial use, and is available on the company's ElevenCreative and ElevenMusic platforms, with API access coming soon.
How the room's reading it
The release is seen as the latest move in the race between AI labs to ship professional-grade music tools. Competitors like Google, Stability AI, and Suno have all recently pushed models capable of generating longer, more complex tracks. The key distinction practitioners are spotting is the focus on workflow. While others chase length, ElevenLabs is offering granular, sectional control that mirrors how music is actually made.
There's also a significant business angle. With other AI music startups like Suno and Udio facing copyright lawsuits, ElevenLabs' emphasis on using licensed data is a clear signal to builders. The consensus is that this de-risks building on their platform.
Sailfish's take
We think the most important feature here isn't the genre-switching party trick — it's the sectional composition. For too long, generative audio has been a 'prompt-and-pray' affair. This is one of the first real steps toward a proper digital audio workstation workflow, which is what separates a toy from a tool you can actually ship with.
The combination of granular control and commercially-safe, licensed training data is the real story. It makes the platform a serious contender for production use cases like dynamic game soundtracks or personalised ad audio. We'd start prototyping with this immediately. It's worth watching, and worth building with once the API drops.