YouTube Rolls Out AI-Generated Vocal Clone Feature

Charlie Puth, Charli XCX, Demi Lovato, John Legend, Sia, T-Pain and Troye Sivan are among the artists whose voices are available to clone using YouTube’s new Shorts feature powered by artificial intelligence. Here’s the low-down:

What is it? The experimental feature, dubbed “DreamTrack,” is now available for a small group of select creators in the United States. Capable of generating up to 30-second snippets of music using AI replicas of the voices of participating artists, the tool is fueled by Google DeepMind’s Lyria music-generation model. 

How does it work? The text-based tool combines user ideas with AI technology, allowing creators to type out their own ideas for songs into a prompt. From there, DreamTrack generates an audio clip mimicking the tone of the chosen artist; notably, more than half of the nine total singers included in the trial are signed to Warner Music.

YouTube is also continuing work on a slew of song creation and production tools designed in collaboration with artists to “bolster their creative process.”

What they’re saying: “The Lyria model excels at generating high-quality music with instrumentals and vocals, performing transformation and continuation tasks, and giving users more nuanced control of the output’s style and performance,” reads a description on Google DeepMind’s website. “Each participating artist has partnered with us and will have a hand in helping us test and learn to shape the future of AI in music.”