ElevenLabs has officially launched Music v2, a significant upgrade to its artificial intelligence music generation platform. This latest iteration introduces the ability to transition between disparate musical genres within a single track, allowing for complex shifts—such as moving from operatic vocals to heavy metal instrumentation—without sacrificing audio coherence. The model also supports the integration of non-musical sound effects and maintains high-speed lyrical delivery, such as rap, with improved clarity.

Beyond simple generation, the new model offers granular control for creators. Users can now isolate specific segments of a composition to regenerate them via text prompts while keeping the remainder of the track intact. This modular approach allows artists to construct songs section by section, stitching together intros, verses, and choruses to build a complete arrangement. The company noted that the model has been optimized for greater reliability across various languages and vocal styles.

As the competition in the generative AI music space intensifies, with major players like Google and Stability AI rolling out their own sophisticated tools, ElevenLabs is positioning its product as a commercially viable solution. The company emphasized that Music v2 is built upon licensed data, ensuring that the resulting tracks are cleared for commercial use. This focus on copyright compliance serves as a strategic differentiator in an industry currently grappling with legal challenges regarding training data.

The new technology is currently accessible through the ElevenCreative suite, designed for marketing and branding professionals, as well as the dedicated ElevenMusic platform. ElevenLabs plans to expand access further by integrating the model into its API in the near future.