Could artificial intelligence produce a hit song?

Think:Act Magazine Artificial Intelligence
Could artificial intelligence produce a hit song?

Article

by Michael Hann
illustrations by Kama

Creativity is a human pursuit. Or is it? You might think that only musical geniuses can write the perfect pop song, but the algorithms are catching up fast.

Artificial intelligence in music on the rise

A lilting piano introduces the song. It's a pop song much like any other: Foreboding percussion joins in for the first chorus, an acoustic guitar gradually changes the texture of the music and for just short of three minutes, it builds to its climax. You wouldn't raise an eyebrow at hearing it on the radio, in a shop, in a bar. Yet Break Free by Taryn Southern is also a pop song unlike any other.

"Break Free" comes from Southern's upcoming album "IAMAI", the first mainstream pop album on which every single musical element, bar Southern's lyrics and vocal melodies, was provided by artificial intelligence; every note of instrumentation was composed by a program created by the US tech firm Amper Music. What's surprising isn't that this has happened, but that it has taken so long to happen. It's already been 60 years since Lejaren Hiller and Leonard Isaacson programmed the University of Illinois's ILLIAC computer to create "Illiac Suite", the world's first piece of computer-composed classical music. Yet only now is AI music a tool that anyone with a laptop can use.

Making music with artificial intelligence involves a huge amount of data and high processing power - and leads to the promise of exceptional virtuosity.
Making music with artificial intelligence involves a huge amount of data and high processing power - and leads to the promise of exceptional virtuosity.

For Drew Silverstein, Amper's CEO, that availability signals a revolution in creativity. "We are opening up the doors of expression to billions of individuals. History has shown that the more we are able to democratize creativity, the more dynamic and explosive our creative world gets, in a positive way." Is it the Gutenberg printing press of music? "That's a great analogy. The difference being that you still had to own the printing press to be able to publish the book. With Amper, not only does it help solve the question of how we share our music with the world, but it makes sure everyone has their own printing press."

Amper is one of a number of AI music companies whose focus isn't on creating works of art, but putting AI tools into the hands of as many putative musicians as possible. The main focus of these companies is providing functional music, usually for video: Amper, for example, was set up by film and TV composers responding to declining production company budgets available for bespoke music. Among Amper's competitors is London- based firm Jukedeck. Both Amper and Jukedeck work on similar principles: Users are offered choices about style, speed and instrumentation. Once those parameters have been set, the programs produce a piece of music. Both charge single license fees for commercial usage, with no royalties due back to them in the unlikely event you somehow get given a worldwide hit.

AI: keeping time
1957

Composers Lejaren Hiller and Leonard Isaacson unveil Illiac Suite, a string quartet composed by the ILLIAC computer at the University of Illinois. They set certain parameters – for example, an ostinato eight-note pulse as the rhythm. Then the computer assigned the pitches.

1965

The 17-year-old Ray Kurzweil – later an eminent inventor and futurologist – plays the piano on the US TV show I've Got a Secret. His secret: The piece he played was composed by a computer he had built.

1980

Composer David Cope, suffering from writer's block, turns to a computer for help. He becomes obsessed with the possibilities and in 1993 releases Bach By Design, an album made by a computer programmed to replicate Bach.

1996

Musician and composer Brian Eno releases the album Generative Music 1 made using SSEYO Koan software. "Generative" was used to describe system- generated music that would change each time it was played.

2016

Sony CSL Research Laboratory makes the first pop song created by AI. Daddy's Car was not without human stylistic input, however – arranged and produced by Benoît Carré, it was designed to sound like the Beatles.

AI music versus human-created compositions

There are a variety of reasons that AI music is becoming commercially viable now, suggests Ed Newton-Rex, the CEO of Jukedeck. Firstly, cheap, mass-produced AI music has a market owing to the growth of digital media. Then there's the technology itself: The availability of more data, the increased processing power to deal with that data, and the way these allow "neural networks" to become more efficient in replicating the workings of the human brain. It's now become a virtuous circle, he continues. "Improvements in data, processing power and models have led to more success, which leads to more funding, which leads to more expertise being drawn in, which in turn means there are more improvements in models and data and processing power. I think we will all be very surprised by how things continue to improve."

Six short pieces of music are the components for artist Jem Finer's Longplayer. The sections are played by computers to a set of rules and create a piece of music that will play without repetition for 1,000 years.

Then comes a philosophical question: What is music? Is it just a series of notes, or is it a reflection of humanity. Jem Finer, creator of "Longplayer" and "Score for a Hole in the Ground", two installations that create music without the need for human input after an initial setup, and formerly of the folk-punk band The Pogues, says: "It's both things. When I write a computer program that spews out loads of notes that are filtered into some semblance of composition, the machine's doing that of its own accord. But it's only doing that because I've told it to. It's an expression of humanity."

But what comes next? Will "Break Free" be an oddity or will it prove the harbinger of music's next great revolution? "Few people argue that AI music will be indistinguishable from human-created music in 100 years, or sooner," Silverstein says. Finer points out that so much of what we take for granted in pop music – programs such as AutoTune that pitch a singer correctly even if they can't hit the note or the GarageBand app that enables kids to form bands without being a musician – mean computers are already deeply embedded in pop.

It's one thing, of course, for Amper to create "Break Free". But will AI ever be able to create its own "Like a Rolling Stone" or "Purple Rain", or any of the songs where the genius of the creator is as important as the songs themselves? "I firmly believe we, as a human society, will forever value the creation and expression of art; it's core to our human existence," Silverstein insists. So it seems the Bob Dylans, Adeles and Beyoncés aren't on the way to being phased out. Not yet, anyway.

Further reading
Our Think:Act magazine
blue background
Think:Act Magazine

Think:Act magazine about artificial intelligence

{[downloads[language].preview]}

What exactly do people mean when they talk about AI in 2018? Where do I start if I want to embrace AI in my business? Get your questions answered in our Think:Act magazine on artificial intelligence.

available in
Subscribe now!

Curious about the contents of our newest Think:Act magazine? Receive your very own copy by signing up now! Subscribe here to receive our Think:Act magazine and the latest news from Roland Berger.

I agree that the above mentioned data will be used by Roland Berger to send me Think:Act Magazine newsletters and further information from the Roland Berger Group via email. I can revoke my consent at any time with effect for the future by email or to the contact data given in the Marketing Privacy Policy. This consent is in accordance with the Marketing Privacy Policy.
Portrait of Think:Act Magazine
Think:Act Magazine
Central Europe
  • Photos monsitj / iStockphoto; 123ducu / iStockphoto; Ana Kova; Roland Berger GmbH; Cheng Jiabei / Imaginechina / laif; Chad Hagen; Sarah Illenberger; Mungang Kim / The Noun Project; Tavis Coburn; Frank Hähne; PeopleImages / iStockphoto; Mark von Ulrich