Recent breakthroughs in Artificial Intelligence technology are reshaping the way we look at online content and providing exciting new possibilities to every major industry across the board. It also generates challenges the likes of which industry professionals have never faced before. The music industry is no different- with a new era of AI comes a brand new generation of music streaming, possibilities, challenges, and all.
Earlier this year, streaming giant Spotify rolled out its AI DJ feature for premium subscribers. Dubbed “DJ X,” the service features an AI generated voice guiding you through music picked based on your own personal music taste as well as selections from Spotify’s top charts, interspersed with moments of commentary from DJ X themself. The experience is meant to feel like your own personal radio station, free of commercials and completely tailored to your own music taste.
But AI disk jockeying doesn’t stop at streaming services. Recently, the “world’s first” AI radio DJ went live on Portland’s Live 95.5, based off one of the station’s real life DJs, Ashley Elzinga. While Ashley (human Ashley, that is) expressed enthusiasm at the breakthrough, not all reactions to the news were positive, with some listeners taking to twitter to voice their concerns.
If a world in which human DJ personalities have been replaced by AI is on the horizon, it should come as no surprise that AI generated music tracks are rising in popularity. At the forefront of this is AI Streaming service WAVs.ai, which hosts popular songs sang by AI recreations of popular artists’ voices.
While it’s quite amusing to satisfy the curiosity of “what would Wonderwall have sound like if it was sung by Kurt Cobain,” the technology certainly isn’t quite up to snuff yet. The backing tracks are simply karaoke versions of the original songs themselves, and the audio quality of the AI generated vocals is poor. The result is something that is amusing, maybe even impressive, but decidedly uninspired.
Despite this, WAVs.ai is still doing quite well for itself, raising $20 million from investors. The company has dubbed itself the “world’s leading AI music provider” and claims to work with the likes of Apple, Google, Amazon, and Meta.
It seems that nowhere on the WAVs.ai website are any of the original artists or producers credited for any of the AI mashups they host. This example begs the question: What are the implications of AI generated music on the greater music industry?
While some major artists, most notably Grimes and Paul McCartney, are embracing AI music generation, most of the deepfake mashups such as the ones hosted on WAVs.ai are unauthorized by the original artists. This is likely why WAVs.ai doesn’t require a sign-in and is completely ad-free… for now. The service’s website promises royalties to those who upload these AI recreations in the future.
Already, AI generated music has been cropping up on major streaming platforms. Recent startup Figaro plans to help these services identify AI generated tracks. They also plan to help them identify music that could infringe upon artists rights; deepfake Kurt Cobain’s beautiful rendition of “Wonderwall,” for example, wouldn’t fly.
The Recording Academy is responding to the uptick in AI generated music by issuing new guidelines for Grammy eligibility, stating that Grammys will only be awarded to music with significant human contribution. While this isn’t a blanket ban on AI generated music, it does mean that a song could feature AI generated music or composition, but not lyrics.