YouTube Music could soon make it a little easier for anyone to live out their pop star dreams, or at ... [+]
Becoming a successful musician is hard enough, but becoming a pop star is something that many will strive to be while very few will succeed. YouTube Music could soon make it a little easier for anyone to live out their pop star dreams, or at least feel like an A-list musical producer—with a lot of help from artificial intelligence.
This week, the Google-owned service introduced an experimental tool dubbed Dream Track in YouTube Shorts, powered by Lyria, Google's DeepMind's advanced music generation model. Users pick the concept for a track and then selecting the artists—with AI doing the hard part of "writing" the song. It is currently available to a handful of artists and creators, who will provide YouTube Music with feedback that could determine how AI could be employed in content creation.
Nine musical artists including Alec Benjamin, Charlie Puth, Charli XCX, Demi Lovato, John Legend, Papoose, Sia, T-Pain, and Troye Sivan have chosen to collaborate on the project. The small group of creators can employ the tool to produce wholly original AI tracks that are up to 30 seconds in length.
"By simply typing an idea into the creation prompt and selecting a participating artist that appears in the carousel, an original Shorts soundtrack featuring the AI-generated voice of that artist will be produced for the creator to use in their Short," Lyor Cohen, global head of music and Toni Reid, vice president of emerging experiences at YouTube, wrote in a blog post on Thursday.
A 22-second sample clip featuring T-Pain and another featuring Charlie Puth were also released this week,
Neither track sounds like anything that could climb to number one on the charts or earn a Grammy nomination, but it is noteworthy that AI was able to produce content that employs the cloned voices of the singers. The technology is likely to improve quickly, as we've already seen this with a lot of AI-generated content creation tools.
Embracing AI
YouTube Music's Dream Track test comes even as the video streaming service announced this week that it would require creators to disclose when AI was employed in the generation of content. YouTube Music may be taking a "if you can beat them, join them" approach.
"The company is engaging directly with a diverse group of musicians, songwriters and other artists who will use and evaluate the new AI tools," technology industry analyst Charles King of Pund-IT explains. "That's an important distinction between this effort and other Generative AI platforms that are facing increasing criticism for using copyrighted IP without consulting or compensating the writers and other artists who own that material."
King further suggested that given YouTube's size and leadership position in online content distribution, this is a model that other companies developing AI tools and services would be wise to emulate or follow.
"Technology is advancing so rapidly in the creation, distribution and promotion of popular music that artists and rights holders end up constantly having to play 'catch up,'" Bruce Barber, professional in residence at the University of New Haven, added.
"This happened when artists started using digital music 'samples' from other artists to create their own works: there were concerns about copyright, ownership and creative control—and it's not surprising that artists today have similar concerns surrounding the use of AI."
The core issues involving AI-generated music will continue to revolve around permission and intent.
Absent the approval of the artist/rights holder, AI-generated creations could become something that artists might not want to be associated with, especially if the new works cast the original creator in an unfavorable light," Barber suggested.
"This being said, I think the artists mentioned are taking the correct path: they see the inevitable use of AI in reinterpreting their work, and by engaging with the technology at an early stage should expect to gain more control over their work than artists that take a 'wait and see' approach," Barber continued.
Replacing Talent?
Of course, the other debate is whether AI can ever replicate the "talent" of any artists—musical or otherwise. Algorithms may be able to generate original musical content, but it doesn't mean it can recreate the "magic" that goes into songwriting.
There is the possibility that AI could create a timeless symphony or hit pop song. But it is just as likely to create something that sounds like it was written by a computer and performed by a cover band.
"In the end, listeners will decide whether or not AI-created music is worth their time and attention," Barber explained. "In the meantime, it seems like a good idea for artists to resist the temptation to bury their heads in the sand and hope that music created by AI will simply go away. Even at this early stage that prospect seems highly unlikely."
