Robot Rock: Does AI Have a Music Problem?

In April of this year, enormous news broke in the world of music. A lost album had been discovered from Oasis, the iconic and beloved Britpop legends. Taken from the period between the band’s third and fourth albums, a set of newly-released songs perfectly captured late-90s culture between their crashing choruses and the unmistakable Mancunian drawl of Liam Gallagher. Fans were delighted with this surprise Easter present, even if it had one glaring problem: 

It wasn’t real. 

The tracks, released under the moniker of AIsis, were - naturally - artificial renditions of what unreleased Oasis songs from that era might have sounded like, had they ever been made. Except, that’s not where the intrigue of this story ends. Because, scratching below the surface, these songs weren’t actually AI, either. Or at least not in the way we’ve commonly come to understand artificial intelligence.

It’s an anecdote which comes to mind when speaking with Kristoffer Roggemann and Steph Altman - creative producer and creative director, respectively - of the music and sound studio Mophonics. The pair have both seen how generative AI in the audio world is still some distance off its counterparts in other creative mediums. And, as Kris explains, there are a couple of good reasons why this is the case. 

“One is scientific, which is that human hearing is so much more discerning than sight”, he says. “For example, it only takes 24 frames per second to trick the human eye into believing it’s seeing movement, but it takes more than 40,000 audio samples per second to build a waveform that the ear believes is a real sound”. 

The Mophonics partner continues to make this point: "I once heard an audiologist claim that if our eyes were as powerful as our ears, then we could look up from Earth and be able see a quarter sitting on the surface of the moon!”. 

And, as Steph continues, there’s another crucial reason why audio AI is currently outpaced by its visual counterparts. “The VFX and animation industries have had several decades-long histories of creating sophisticated computer-generated graphics”, he notes, “so those tools are inherently much farther ahead of the audio equivalents”. 

So, in that case, how did the technology produce something as impressive as AIsis? Well, as mentioned earlier, it didn’t. The songs we hear are in fact written by humans - originally intended for another band - and have been performed and produced in the style of Oasis. The AI element of the project came in the form of replicating Liam Gallagher’s distinctive vocals. Impressive, no doubt, but not on the same level as producing entirely original content from scratch. 

Or, as Steph puts it, “we’re not yet at a stage where AI can remake a well-known song in a believable way. So to imagine it could go way beyond that and write and produce an original song is just not realistic right now”. 

For an industry currently riding atop a wave of AI hype, this is an arresting message to hear. And, to be clear, neither Kris nor Steph are technological naysayers. “AI has the power to continue democratising creativity”, posits Kris. “Just as a kid with an iPhone and CapCut can put together an edit that would have only been possible with a professional film editor in 2013, so music AI will soon equip a non music-maker with the ability to create basic AI-generated music for their non-professional project, for pennies”. 

But, within the context of our industry, there’s a sense that we may be expecting AI to run before it’s able to walk. “Today with the abundance of sounds and tools at our fingertips, throwing together loops is so easy that bottom-rung music beds are extremely accessible - and they’re priced accordingly”, explains Steph. “There’s AI stock music products out there that are pretty competitive already. But the bar is low. This kind of music is for the least sophisticated content creators”. 

For brands operating in the unforgiving and relentlessly competitive landscape of modern media, that kind of music simply isn’t going to cut it. “There's a giant, yawning chasm between that and the music we get paid to make, and I don’t see it closing soon”, continues Steph. “The difference between good music and great music is evident to skilled ears (i.e. to the people who are hiring us) and musicAI doesn’t even do good music yet. That’s not going to work for serious brands”.  

Striking a note of agreement, Kris foresees a future in which the prevalence of AI will underline the importance of a guiding human hand. “The musical low water mark will rise, but as a result, pro music-makers will become even more original, with the good taste and ingenuity to make the right choices - including the surprising choices that don’t have the sterility of algorithmically created music”, he says. 

Steph agrees about the need for a human touch. “If and when AI is ready to craft more sophisticated pieces of music, human prompt engineers will be needed, just like with ChatGPT or Midjourney”, he says. “It will require some kind of ‘operator’ with musical chops who is able to adjust compositions to satisfy the tastes of a client, or to manipulate it to achieve their own vision. These people might well be the ‘composers’ of tomorrow”.

And so, there may be some expectation management in order when it comes to AI in the world of audio. Although the technology may already be a key creative partner, it’s hard to envisage top-tier brands turning to it in order to create music which helps their campaigns to stand out. There’s still an art to music which remains, perhaps thankfully, human. 

“Even mixing and mastering is such an art. It’s the secret sauce that can make the difference between a good song and a hit”, says Kris. “‘Billie Jean’ was famously mixed 91 times before they settled on the final mix, showing how delicate a balance you need to strike to make that hit sound. I simply don’t see a world where a computer will be able to mix like a human can - or sift through 91 mixes of ‘Billie Jean’ to pull the trigger on the magic mix like Quincy Jones did”. 

Human Melody

To fully appreciate the intricacies of producing genuinely top-tier music, we need to broaden the conversation out. In an article penned back in 2019, the legendary Australian singer/songwriter Nick Cave mused on the potential of AI to write a “great” song. In doing so, he hit on a crucial point for today’s AI debate: That the difference between a song that is ‘good’ and one that is ‘great’ is the human experience. As he wrote when reflecting on the impact of Nirvana’s ‘Smells Like Teen Spirit’:

“It is perfectly conceivable that AI could produce a song as good as Smells Like Teen Spirit.

But, I don’t feel that when we listen to Smells Like Teen Spirit it is only the song that we are listening to. It feels to me, that what we are actually listening to is a withdrawn and alienated young man’s journey out of the small American town of Aberdeen – a young man who by any measure was a walking bundle of dysfunction and human limitation – a young man who had the temerity to howl his particular pain into a microphone and in doing so, by way of the heavens, reach into the hearts of a generation. We are also listening to Iggy Pop walk across his audience’s hands and smear himself in peanut butter whilst singing 1970. We are listening to Beethoven compose the Ninth Symphony while almost totally deaf. We are listening to Prince, that tiny cluster of purple atoms, singing in the pouring rain at the Super Bowl and blowing everyone’s minds. We are listening to Nina Simone stuff all her rage and disappointment into the most tender of love songs. We are listening to Paganini continue to play his Stradivarius as the strings snapped. We are listening to Jimi Hendrix kneel and set fire to his own instrument.

What we are actually listening to is human limitation and the audacity to transcend it. Artificial Intelligence, for all its unlimited potential, simply doesn’t have this capacity. How could it?

So - AI would have the capacity to write a good song, but not a great one. It lacks the nerve.”

“I love that quote”, reflects Steph. “It feels like a work of art in its own right. Of course, not every song needs to be Smells Like Teen Spirit or Purple Rain. If you just want something to accompany your YouTube video of how to unclog a drain, then you don’t need that kind of memorability or artistic genius. But if you do want your content to stand out, and you want to be remembered, that level of craft is what it takes”. 

Whether AI will ever get to the standard of a Prince or Nirvana is an open question. As Kris acknowledges, “Even pre-AI, innovation in music has been pretty awesome to be a part of: things that were simply not feasible 15 years ago, like isolating specific elements of a mixed track to edit or remove, are now possible thanks to natural tech developments. But unlike in the text and visual space, AI is not performing breathtaking miracles in music… at least not yet.”

Yet, if nothing else, it’s a great opportunity to celebrate the essential humanity of great music-making. And for Steph, Kris, and the team at Mophonics, that’s never going to change.