If you’ve not been living under a rock these past few years, you’ll have heard an awful lot about AI. Chances are you might have picked up on some of the unease it’s causing in the music industry.
The ability AI has in mimicking human voices is uncanny. There have been reports of calls to parents from highly convincing fake children asking for money after car accidents, for instance. The implications for this in music are obvious, and a few high-profile examples have hit the headlines recently. A few weeks ago, AI designer Roberto Nickson unveiled a Kanye West soundalike, and this week it’s been Drake’s turn (again.) “This is the final straw, AI,” he said, after a previous example was posted by a fan.
If the technology can already carry off reasonable impressions of distinctive industry voices, imagine what it’ll be able to do with generic-sounding pop vocals, many of which are already auto-tuned or otherwise digitally treated.
Money to be made
For some reason, AI engineers have so far shown a preference for rap, but it won’t be long before this affects all genres. Want something that sounds close enough to Ed Sheeran or Taylor Swift to satisfy the casual listener? There are a lot of parties who might, is the fear, not least the streaming services – AIs could be much cheaper to work with.
It’s easy to imagine where this could be going. Hologram versions of artists like Tupac didn’t really take off, but could new vocals from Prince, Freddie Mercury, Whitney Houston or John Lennon? There could be fans that will pay good money for spectacle, and these “new” songs could be performed live using robotics or reviving those holograms. Deceased artists’ estates aren’t always known for holding back on approvals when there’s money to be made.
Much of this is a legal grey area. While music companies are doing their best to sound future-forward around AI, they’re also taking action. Universal Music Group, Drake’s label and the biggest of the majors, has already demanded that streaming services prevent AIs from accessing its copyrights. This points to the rather prosaic truth that, rather than any real intelligence being involved, these AIs are essentially scraping and imitating artists’ copyrights. It’s not absolutely clear how copyright law relates to imitate human voices, especially if whoever makes a song using someone else’s is explicit about the fact. So there are a lot of burning questions, not least around who gets paid.
No need to panic?
There may be no need to panic. While there’s always a market for the new (and the tacky), fans are not stupid. What music listeners buy into is all too human: a story, a creative world to escape into and a personality. Sure, AIs will soon be attempting to provide all of these, but very highly evolved search-and-replicate technology is a long way off competing with real artists. Often, the messiness and unpredictability of human life is a key part of the appeal of artists. It’s also key to how music gets made in the first place.
If forced to guess, I’d say that AI-based music would struggle to reach double figures in terms of its market share, and recorded music is growing by more than that annually.
Currently, the only notable new technological advance is that AI has learned to ape the human artists fans actually care about. So perhaps we can keep calm and carry on – for now.
