YouTube Music has introduced a revolutionary feature called Dream Track, which allows users to generate music in the styles of popular artists such as Sia, John Legend, and T-Pain. However, this tool comes with a crucial caveat: artists’ permission is required to use it. This bold move hands over the creative reins to artificial intelligence, making it a groundbreaking feature in the music industry.
Powered by Google’s DeepMind music generation model, Lyria, Dream Track is tailored to produce unique backing tracks for creators on YouTube shorts, reminiscent of the popular platform TikTok. However, this innovative step comes at a delicate juncture in the already tense relationship between AI engineers and the artists and writers whose creations underpin these cutting-edge models.
Dream Track operates by auto-generating 30-second audio clips based on textual prompts such as “a ballad about how opposites attract, upbeat acoustic.” Nine artists, including Alec Benjamin, Charlie Puth, and Demi Lovato, have given their consent to have their vocal and instrumental styles emulated by the tool.
Another YouTube tool, Music AI, goes a step further, transforming a hummed tune into a guitar riff or converting a song into a different genre. Google employs its SynthID watermark to subtly differentiate tracks produced by these tools, making the changes imperceptible to the ear.
The battleground for generative AI music cloning has been active since at least 2020, marked by the release of OpenAI’s Jukebox program. However, recent advancements in artificial music realism, exemplified by a viral fake Drake song in April, have intensified the existing tensions between artists, the music industry, and tech companies seeking to capitalize on AI-generated content.
Notable incidents include Universal Music Group’s lawsuit against Anthropic for the unauthorized distribution of song lyrics through a chatbot and superstar Bad Bunny expressing discontent over an AI track utilizing his vocal likeness.
These controversies prompted Ed Newton-Rex, head of audio at generative startup Stability AI and a composer himself, to resign, citing disagreement with the company’s stance on “fair use” of copyrighted works in training generative AI models.
YouTube Music is actively positioning itself as a responsible AI collaborator for artists, collaborating with Universal Music Group in August to establish principles for ethical AI music generation. This week, the platform unveiled rules concerning deepfake videos, permitting only labels and distributors of artists engaged in AI music experiments to request removal.
If YouTube Music successfully navigates a constructive partnership with the music industry, it could become a powerful tool for rival platforms like TikTok, which introduced its own music generator earlier this year. However, ongoing debates surrounding copyright norms and revenue distribution underscore that tensions around this new technology are escalating.