What is the future of AI in music?

The evolution of modern music has been inextricably linked with advances in technology. A few examples: in 1929, George Beauchamp and Adolph Rickenbacker patented the Rickenbacker Frying Pan, the first electric guitar, paving the way for rock ‘n’ roll. Fast forward to October 1964, when Bob Moog unveiled the first modular voltage-controlled synthesizer, effectively kickstarting electronic music. The recording software ProTools was launched in 1989 and would go on to make home recording affordable, totally changing the way future generations approached music. 

Technology has been the biggest catalyst for constant changes in the way music is composed, performed and distributed, but the biggest shake-up in music may be yet to come as Artificial Intelligence (AI) comes to the fore. AI has already been used to mimic the music of popular artists.  

In 2021, a “new” Nirvana track, named “Drowned in the Sun”, was posted online. Casual listeners might’ve been forgiven for thinking a lost Kurt Cobain song had been unearthed but “Drowned in the Sun” had actually been generated using two AI frameworks. Google’s Magenta produced the music, based on data gleaned from dozens of original Nirvana recordings, while a neural network generated plausibly Cobain-like lyrics, which were then delivered by the singer of a Nirvana tribute band.  

Hardcore Nirvana fans won’t have been tricked by “Drowned in the Sun”, but it’s undeniably close to the grunge pioneers’ sound, and suggests AI isn’t far away from producing plausible fakes. 

Pianist, composer and technologist Zubin Kanga (a lecturer in Musical Performance and Digital Arts at Royal Holloway University, London and head of the pioneering Cyborg Soloists project), agrees. “From what I’ve experienced so far, I don’t think it’s sophisticated enough yet,” Kanga told Reader’s Digest. “But it is remarkable how well it can imitate someone else. I’ve heard some very good imitations—things done in the style of Beethoven, Chopin, Frank Sinatra or Britney Spears. These things are very good at resembling a particular composer or artist for a short period of time.”  

Still, according to Kanga, AI has some way to go until it can successfully imitate composers and performers of longer pieces of music, “I think it’s often about the whole thing, the whole shape of it. And the sense that you can hear a performer’s thought processes going through it—the way they think about the entire architecture of a piece as they’re performing it.

“That’s something that these AI models are nowhere near replicating yet. At a certain point, it will be able to replicate people, the sound or the style of someone very well. But I’m not sure how much that will move you, compared to listening to the real thing.” 

While music composed by AI might not yet have the emotional impact of music written and performed by humans, Kanga believes there’s an area in which it is almost advanced enough to usurp real-life composers. “I think it won’t be long until AI tools are used to produce commercial music for very specific purposes. If you have a 30-second commercial, that’s an easier thing for an AI to imitate.” 

But the use of AI in performing and composing music is not without its controversies. As the technology advances, it’s becoming increasingly urgent that ethical and legal debates about its usage are held.  

“There are definitely issues around future use of AI,” says Kanga. “Say someone’s recorded something which can then regenerate their voice or their style of music, then who owns the copyright for that model that they’ve done? And that’s something that I’m already dealing with, because in my Future Cyborgs project we’re doing things where someone’s recorded something and that’s being used by someone else. Or we’re using recordings that exist and then putting them in a new context and they might not be that recognisable once they’ve been mixed in with all sorts of other things. But still, someone’s recording has contributed to that. So, I think this this is an area of copyright law that will have to develop very quickly.” 

Still, musicians have always adapted to new technologies and the most forward-thinking among them will grasp the opportunities that AI will open up. Kanga is already relishing the prospect, “I’m looking at live generation, where you have something that resembles a live performer on stage—AI that you can interact with and improvise with, like a jazz musician. I think that’s also very exciting. And again, that’s developing very fast but it’s not yet at the stage wehere it’s going to replace people, but at the stage where you can have a really interesting interaction. 

“It’s a tool and I think that’s the way it should be used—like a lot of these technologies. They’re not replacing people, they’re just useful tools for composers and performers to use.” 

Read more: Meet the composers using AI to create music

Read more: Is music licensing changing the industry?

Keep up with the top stories from Reader's Digest by subscribing to our weekly newsletter

*This post contains affiliate links, so we may earn a small commission when you make a purchase through links on our site at no additional cost to you.