If you’ve ever wished you could sing like a mega-star, a new YouTube feature may let you do just that, because the company is working on an artificial intelligence (AI) tool that lets creators record audio in the voices of famous musicians.
Details are still a little sparse, as no potential artists or labels have been announced. But YouTube has reportedly approached Sony Music Entertainment, Universal Music Group, and Warner Music Group about getting the rights to songs to train the tool.
Also: How to write better AI prompts
First reported by Bloomberg, the move comes not long after YouTube (and other music-streaming services) banned a viral AI-generated song featuring a realistic-sounding version of Drake and The Weeknd.
There are certainly murky waters around the use of AI in music, especially when it comes to mimicking the likeness of a real person. AI is a powerful tool, but its capabilities also pose a big problem in terms of copyright and licensing.
It appears YouTube, though, is taking the lead in overcoming the challenges by embracing emerging technology. It was just a few months ago that the company established its “first-ever set of AI music principles” to allow for creative expression while still protecting artists on its platform. Now, it’s adding AI-generated mimicry as an actual feature.
Also: How AI helped get my music on all the major streaming services
This development, of course, raises a few questions. First off, what limits will be in place on what content can be generated? Will creators be able to sing anything they want? Second, how will monetiztion be handled? Any revenue generated on an AI video that mimics someone’s voice will certainly be shared with the artist, but what will that split look like? Will an artist have any recourse if someone creates content where they’re singing a song that’s legal, but with a message the artist disagree with personally?
YouTube is no stranger to AI integration, rolling out new AI video tools for creators last month that allow for automatic dubbing into other languages and automatic background generation, plus AI video summaries for viewers. This new artist sound-alike feature was apparently supposed to be a part of that rollout, but rights issues have slowed the process.
