AI in music is the talk of the town. Whether you’re a music fan, watching from the sidelines as streaming services make continual updates to their AI policies, or a musician fighting for their individuality in a lawsuit against an AI music company, it has become blatantly clear that this particular technology isn’t going away any time soon.
Indeed, with the recent investments of major labels such as UMG, Warner and Sony in AI music companies, it looks as though the role of AI in the music industry is only likely to expand.
Naturally, opinions are divided. Some users of the text-to-music generation software Suno are reportedly abandoning human-made music altogether, in favour of their AI-generated tracks. Meanwhile, others have taken to the streets in protest against licensing recorded music to generative AI companies.
As a 50-year-old hi-fi publication, music is at the heart of our work. While testing products, our review team listens tirelessly to a broad range of tracks, carefully assessing how human emotions and real-life recordings are conveyed by the equipment at hand. As such, we wanted to find out more about the role AI is playing in the records we listen to.
To do so, we spoke to British sound engineers about the ways that AI is changing their work.
Exposing Flaws
In the view of Jake Murray, director and studio manager at Strongroom Studios, generative AI within music is uncovering pre-existing issues within the industry.
“People try to spin it as this new thing on the horizon,” says Murray, “but what I’m seeing – much like during the Covid-19 pandemic when artists lost all touring revenue and it exposed problems with streaming and digital remuneration – is that this new software is exposing flaws with the commercialisation of music and the behaviour of corporations.
“In the case of recent lawsuits, the record labels aren’t happy because data they could monetise is being scraped on a massive scale,” he says. “But for a massive corporation, AI music is incredible, it is product, through and through. There are no creator rights, no unpredictable behaviours on tour, no ‘I’m tired, I want out of my contract.’ I think anyone who believes that major labels are going to continue fighting generative AI once they’ve got a deal in place is wrong.”
Izzy McPhee, a freelance mastering engineer who has worked with acts such as Dream Wife, Walt Disco and Marla Kether, feels that AI’s encroachment presents artists with new challenges. “The pace at which music is created and released now, definitely puts pressure on artists to prioritise singles and virality over the integrity of their music,” she says. “The instantaneous results of AI affect how quickly everyone expects information, and of course, music.”
Such pressures feed back into the use of AI, as musicians reach for tools that allow them to create faster and more cheaply. The attraction of such tools isn’t hard to appreciate given that nearly half of professional musicians were found to earn less than £14,000 a year in a 2023 census by Help Musicians and the Musicians’ Union.
One of Murray’s friends, who is a dance musician, recently approached him in a crisis about having used Suno to create a vocal topline. “He found himself really conflicted about whether to use it,” says Murray. “It sounded so close to what he had imagined, but it wasn’t sung by a singer, recorded by a sound engineer, mixed, balanced, processed and submitted to the production house. It hadn’t generated those three or four stages of revenue. It had only cost him the comparatively meagre subscription fee.
“It’s vital that we remember we are talking about lives and livelihoods here,” he says. “I think it’s going to be left to the individuals to decide what they think is kosher or not.”
"Human creativity is what makes beautiful music"
AI has seen widespread adoption beyond music. Even at Strongroom, Claude AI has proven handy in aggregating the studio’s vast discography.
“It has been incredibly useful for data management,” says Murray. “And best of all, nobody would have wanted to do that job. Previously, when I’ve sat down with new employees and broken the news that they are going to spend a few weeks sorting through the discography, the look on their faces is like they are about to die.”
Murray adds that, within musical work, there is a “grey area” of audio effects plug-ins that use AI.
“I’m a bit unclear about what people count as AI in the studio,” he says.
Recent software such as Synplant 2, which uses AI-powered features to turn audio samples into editable synthesiser patches, walks a fine line. “Depending on who you talk to, these are either plug-ins or neural networks or AI. But I don’t think anyone really has a problem with any of those [pieces of] software, aside from whether they like the sound or not.”
When it comes to the creative, musical and technical parts of her workflow, McPhee takes a stronger stance on using AI tools: “I have made a conscious decision not to use AI, and not to master songs that have been AI-created. A big part of my ethos is that human connection, collaboration and creativity is what makes beautiful music, and the bottom line is that I don’t think AI can replicate this.”
Charlie Francis, a mastering engineer who has worked with artists such as REM, Del Amitri and Gruff Rhys, says that he hasn’t yet seen a reason to use AI in his work, but acknowledges that some of the software he uses has been developed with machine learning.
“My perception is that my clients value my instincts and experience above anything else, so I would be very wary of allowing AI to make any judgments for me,” says Francis. “Automated systems in rendering and other areas are very useful, but they don't involve any decisions and have no artistic influence.”
"Heaven forbid, we might be able to define an era"
A common trait in almost all conversations around AI is their tendency to focus on the future. As technological advances outpace all but the experts’ comprehension (and sometimes even theirs), many of us are left guessing about the state of things to come.
Speculating on AI’s musical future, Murray says: “The people who want to produce homogeneous rubbish aren’t going to need anyone to do it for them anymore. As a result, I think we could see a new era of Muzak (background music played in public spaces such as elevators, shops and waiting rooms).
“On the flip side, I think the people who care are going to make more human-sounding music. I think music made by people whose priorities are expression, connectivity and interaction is going to become more characterful and less perfect. We might see a broader spread in terms of how music can sound, perhaps a release of the tension around things sounding ‘perfect’. Heaven forbid, we might actually be able to define an era again.”
McPhee echoes Murray’s sentiment: “I think there will be a corner of the music industry where people are consuming AI music, and are happy to follow algorithms and playlists that are AI-generated. However, I also think a lot of people will tire of the rapid turnaround, viral side of music and seek out human-made music and connection – whether through listening or attending gigs and concerts.
“In a world of constant overstimulation, I already find myself trying to find moments where I can sit down and immerse myself in longer form albums that take me on an emotional musical journey,” she says.
Francis adds: “A musician friend of mine thinks songwriters regurgitate iterations of everything they have heard before, which would make them indistinguishable from AI, but I think the human genius for making unseen connections and making distinctive work can't be written off so easily. We discount our own emotional and experiential life at our peril.”
The consensus among the engineers seems to be that, while AI threatens revenue streams and professional opportunities for musicians, its mass-produced genericism could drive creative innovation among those who view music as an essential, communal art form. The question is: will musicians find the time to create on an income yet further reduced?
Let us know your thoughts in the comments.
MORE:
What will hi-fi look like in 2076? We asked some of the most influential experts in the industry
Our pick of the best streaming services tested by our expert reviewers
