Music industry

AI was in the music industry long before rapper AI FN Meka appeared

Music has always been influenced by technology – from Bob Dylan’s invention of the phonograph to transitioning from the acoustic guitar to the electric guitar, to the ubiquity of streaming platforms and, most recently, an ambitious attempt to crossover ‘ia with commercial music.

FN Meka, billed in 2021 as a “virtual” rapper whose lyrics and beats were constructed with “proprietary AI technology”, has had a promising rise.

But just days after signing with Capitol Records – the label that carried The Beatles, Nat King Cole and The Beach Boys – and released its debut track “Florida Water”, the record label dropped it. Her pink briefs were partly a response to fans and activists widely criticizing her image — a digital avatar with face tattoos, green braids and a gold grid — and decrying her mix of stereotypes and infused lyrics.

The AI ​​Artist, voiced by a real person and created by a company called Factory New, was not, technologically speaking, a groundbreaking experience. But it was a driver for an imminent discussion within the industry: how AI will continue to shape our experience of music.

The music/AI partnership: crucial for diverse listening experiences

In 1984, classical trombonist George Lewis used three Apple II computers to program Yamaha digital synthesizers to “improvise” with a live quartet. The resulting record – a syrupy, spatial co-creation of computer and human musicians – was titled “Rainbow Family” and is considered by many to be the first example of artificially intelligent music.

In the years that followed, advances in mixing desks popularized the practice of sampling and interpolation – sparking debates about remixing old songs to create new ones (art form or trick good market?) – and Auto-Tune has become a central tool in recordings and on stage for singers. performances.

FN Meka is not the only AI artist. Some were introduced and lasted, with less commercial support. YONA, a “virtual singer-songwriter and AI poet” created by Ash Koosha, has performed live at music festivals around the world, including MUTEK in Montreal, Rewire in the Netherlands and Barbican in the UK .

In fact, the most crucial and successful partnerships between AI and music have been “under the hood,” said Patricia Alessandrini, composer, sound artist and researcher at the Center for Computer Research in Music and Acoustics at the Stanford University.

During the pandemic, the music world has relied heavily on digital tools to overcome the challenges of remote music sharing and playback, Alessandrini said. JackTrip Virtual Studio, for example, was an online platform used to teach college music lessons while students were remote. It minimized time delay, making audio-visual synchronicity much easier, and was born out of machine learning sound research.

And for producers dealing with large music files and digital compression, AI can play a role in signal processing, Alessandrini said. This is important for sound engineers and musicians, as it saves them time and makes it easier to create or export large discs.

There are beneficial apps for technology and music that intersect when it comes to accessibility, she said. Instruments have been made using AI to require less force or pressure to generate sound, for example, allowing injured or disabled people to play with eye movements only.

Alessandrini’s own projects include the Piano Machine – which uses computers and voltages as “fingers” to create new sounds – and Harp Fingers, a technology that allows users to play the harp without physically touching it.

At the meta level, algorithms are the ubiquitous drivers of online streaming platforms – Spotify, Apple Music, SoundCloud, YouTube and others are constantly using machine learning, in less transparent ways, to personalize playlists, outputs , nearby gig listings and music recommendations.

In discussion: the very concept of an AI artist

Less agreed upon is the concept of an AI artist itself. Reactions were mixed among those faithful to the humanity of the art; some who argued that if certain performers were indistinguishable from AI, then they deserved to be replaced; others who invited novelty; and many whose feelings fall somewhere in between.

“With any cultural form, part of what you’re dealing with is people’s expectations of ‘what things look like or what an artist looks like,'” said Oliver Wang, music writer and professor of sociology. at California State University, Long Beach. Gate.

Some experts say these questions leave out a critical point: no matter the technology, there’s always a human behind the work – and that should matter.

“Sometimes people don’t know or see how much human labor is behind artificial intelligence,” said Adriana Amaral, a professor at UNISINOS in Brazil and an expert in pop culture, influencers and fan studies. “It’s a team of people – developers, programmers, designers, production and marketing people.”

But that misunderstanding isn’t always the public’s fault, Alessandrini said. This often comes down to marketing. “It’s more exciting to say that something is created entirely by AI,” Alessandrini said. This is how FN Meka was marketed and promoted online – as an AI artist. But while its lyrics, sounds, and beats were AI-generated, they were then performed by a human and animated, like a cartoon.

If it seems odd for one to become a devoted fan of a virtual character, it shouldn’t, Amaral said. The competitive video game world, which is nothing without its on-screen characters, is a multi-billion dollar industry that sells arenas all over the world.

Still, music purists and audiophiles — and anyone who values ​​music as an experience, rather than just entertainment — may very well resist AI musicians. In particular, Alessandrini said, AI is better at generating content faster and copying genres, though it’s unable to innovate new ones — a result of training their computational models, in large part, using music that already exists.

“When a rapper has these different influences and his own specific cultural experience, then that’s the kind of magic thing he uses to create,” Alessandrini said. “You can tell Bobby Shmurda is one of Brooklyn’s first dig artists because of one song in particular. So that’s a [distinctly] human capability, versus AI.

Alessandrini compares this artistic experience to advances in AI in medicine – applications of robotic technologies used during surgeries that are more efficient and mitigate the risk of human error. But, she says, there are some things humans do better – caring for a patient, understanding their pain.

It’s hard to imagine AI voices ever reaching the emotional and beautifully human depths of, say, a Nina Simone or an Ann Peebles; or channeling the genuine camaraderie and bounce of a band like OutKast.

What’s next for artist personas created using AI?

In 2017, the French government commissioned mathematician and politician Cédric Villani to lay the ambitious foundations for the country’s artificial intelligence (AI) future.

Its strategy, which takes into account the economy, ethics and education, is above all straddling the dividing line between creation and consumption.

“The division between the uncreative machine and the creative human is becoming increasingly blurred,” he writes. Creativity, he continued, was no longer just an artist’s skill – it was a necessary tool for a world of cohabitation, machine and human together.

Is this happening?

You can’t talk about music on a large scale without also talking about money. Although FN Meka was a failure, AI has strong ties to the music sphere that won’t be severed because an AI rapper was kicked out of a label. And it seems inevitable that another major record label or music festival will launch.

Why? It could all come down to cost, say experts and music listeners who run the gamut of cynicism.

Wang said he slyly suspected that record companies and executives were looking at AI musicians as a way to save money on royalty payments and travel expenses in the future.

Beyond the money-hungry music industry, there’s also room for a lot of progress with AI, Amaral said. She hopes that FN Meka’s image and the way it was received was a wake-up call for any AI artist who will inevitably come next. She also mentioned YONA, whom she saw in concert in Japan, as a slim, white, capable pop star — not unlike many who dominate the music scene today.

“We have all the technological tools to create someone who could be green, or fat, or whatever we want, and we’re still stuck on those models,” she said.

“What will the landscape look like in five, 10 or 15 years?” Wang asks. “Pop music, despite people’s cynicism, rarely remains static. It’s constantly changing, and maybe these computer attempts to “create” artists will be part of that change.

Thanks to Dave Tepps for writing this article.