AI audio start-up ElevenLabs on 5 August introduced Eleven Music, a generative model that creates full-length songs, including vocals, from simple text prompts. The company says the service can deliver studio-quality tracks in minutes and gives users fine-grained control over genre, structure and lyrics, with support for multiple languages. Music produced under the platform is billed as royalty-free for most commercial applications, from films and games to advertising. To minimise copyright risk, ElevenLabs has struck licensing agreements with Merlin Network, which represents about 30,000 independent labels and distributors, and publishing giant Kobalt Music Group. Rights-holders who opt in will allow their catalogues to be used in AI training and will receive a pro-rata share of a royalty pool. A more advanced "Eleven Music Pro" model trained on those works is slated for release in the coming months. The launch comes as regulators and the music industry scrutinise AI-generated content. Competitors Suno and Udio are fighting lawsuits filed by the Recording Industry Association of America over alleged use of copyrighted material. ElevenLabs says its model is built only on licensed data and incorporates watermarking and other safeguards to prevent sound-alikes and unauthorised distribution.