Voice AI company ElevenLabs has launched Music v2, an update to its music-generation model that can switch genres mid-track. The company claims the model handles complexity in vocals and composition, allowing it to transition from opera to heavy metal and back, deliver fast rap without losing coherence, and even add non-musical sound effects to a track. That’s right - your AI-generated symphony can now abruptly become a death metal riff, just because it can.
The new release comes nearly 10 months after the startup’s first music generation model. Now, artists can select a part of a song and recreate it using prompts without touching other parts. Instead of short clips, they can build songs by sections - intro, verse, chorus - and stitch them together. ElevenLabs adds that the model performs more reliably across languages, lyrics, vocals, and arrangements, which is handy when your AI opera singer suddenly decides to rap in Klingon.
This launch joins a race among AI labs - including Google, Stability AI, and Suno - to release professional-grade music generation models. At Google I/O, Google added cover creation, section-based editing, and music videos to its Flow Music tool. But ElevenLabs is betting on its licensed data: the model is built on licensed data and cleared for commercial use, so users can freely use the tracks. This is crucial given that rivals Suno and Udio face court cases over copyright issues. Striking deals with labels is key, because nothing says “creative freedom” like avoiding a lawsuit.
The new model is available on ElevenLabs’ ElevenCreative tool for marketing and branding teams, and on the newly launched ElevenMusic platform for AI-generated songs, with API access coming soon. Because what the world needs is an AI that can write a jingle for your cat food brand, then segue into a power ballad about your tax return.