YouTube announced it has started testing Dream Track for Shorts, a suite of experimental features that lets you create 30-second soundtracks with the help of generative AI. Dream Track was announced last year and it"s being developed in partnership with Google DeepMind.
The streaming giant collaborated with nine artists, including Charlie Puth, Demi Lovato, John Legend, Sia, Papoose, T-Pain, Troye Sivan, and Alec Benjamin, whose AI-generated voices can be used to create soundtracks.
YouTube"s first Dream Track experiment allowed select participating creators in the US to generate soundtracks using one of the artists" voices. They could write a prompt and choose from a carousel of AI-generated voices to create an original soundtrack to use in their Shorts.
Dream Track is powered by Google DeepMind"s music generation model Lyria. As part of its newest experiment, YouTube said it is allowing a small group of creators to generate instrumental-only soundtracks of up to 30 seconds in length. The company stated:
Now, creators in the new experiment group can enter English-only text prompts + select a music genre (pop, hip hop, country, etc.) to generate 30-second, non-looping instrumental soundtracks.
We’re excited to continue to ideate more Dream Track features that we hope enable deeper engagement between music fans and artists on YouTube.
YouTube is embracing AI in its products just like many other tech companies. Earlier this month, it tested an AI-powered feature called Jump Ahead, which lets users skip ahead to a commonly viewed portion of the video by double-tapping on the screen.
It uses AI to organize and summarize comments on YouTube videos which at the moment is available on the YouTube mobile app for large comment sections, with comments in English. The company also put rules in place that require creators to properly label AI-generated content they post on YouTube.