Musical Disruption

Duke Music Professor John Supko's Music Changes With Each Playing

When he was supposed to be practicing the piano, 10-year-old John Supko would instead write his own classical music, scrawled on scraps of paper he would then present to his music teacher at his next lesson.

It was rudimentary, but the young man clearly had promise. It wasn't long before his piano teacher handed the budding musician off to a composition teacher at a local music school in Washington, D.C., where Supko lived with his parents.

Supko's first piece for flute, written at age 11.

The composition bug that bit Supko more than two decades ago has remained, though his love of writing music on paper has certainly waned. Supko, now 34, is the Hunt Family Assistant Professor of Music at Duke. He has carved out a niche in generative music, a growing yet still obscure corner of the art world that uses computer software and other digital technologies to create unpredictable music that can sound different each time it is played.

For Supko, this music can be exhilarating and mysterious. But it has also proven frustrating as he's struggled to learn computer programming, a world apart from his area of expertise.

“I was more interested in making my own music.”

Why do it? He knew early that composing music was more fun than performing it.

“It was an ego thing,” he said, recalling the scraps of music he insisted his piano teacher review. “I was more interested in making my own music than playing other stuff.”


As a youngster, Supko wrote what you might think of as “traditional” music – pen to paper, classical arrangements for a single instrument or a duet. At 11, he wrote his first piece, an 87-second short for the flute. Two years later he wrote a short sonata for violin and piano. When he recorded the piece, he played the piano part himself.

Supko says that this composition was influenced by the CDs he was listening to at the time, including Copland's Third Symphony.

He continued to write music, studying composition as an undergrad at Indiana University before getting his doctorate from Princeton.

But all that time, a foundational tenet of music gnawed away at his creative insides: A song is the same each time – the same collection of notes performed in the same order. There’s beauty in that, of course, but also predictability.

But what if you could inject unpredictability into music? What if the end and the beginning might swap? What if a stanza today disappears tomorrow? What if someone – or something – other than the composer decided which notes were played?

John Supko sets up equipment to play his composition: 3 discmans.

Supko first tested this idea while in graduate school. He recorded roughly 75 sounds – each of a different length between five and 45 seconds – on each of three CDs. He put the CDs into three Sony Walkman devices – remember those? – which were then set to shuffle. Out came a tinny, haunting, metallic-sounding series of tones. He would eventually use it as the foundation of a 10-minute piece of recorded music called “Without Stopping,” which featured a live guitarist, organist and vocalist riffing along with the unpredictable, randomly generated music. 

Supko made 'Without Stopping' by randomly shuffling more than 200 varying sounds using three Sony Discmans.

Supko had an even more profound eureka moment at Princeton while a fellow graduate student discussed a new software program. It was Max/MSP, a programming language for music and multimedia. This was when Supko first realized there may be a tool to help him realize his ideas regarding unpredictability in music.

“I felt like I was struck by lightning,” Supko said. “I thought a lot of my vague ideas could be realized. But I was also dejected, because I had no experience with it.”

“It felt like I was struck by lightning.”

Thus began Supko’s grand professional challenge – programming. While in high school, Supko had rejected his father’s suggestion that he take a computer science course – a decision the composer regrets to this day. In lieu of any formal computer science education, Supko set out to teach himself programming. Lots of trial. Lots of error. It took him a good six years before he attained anything close to mastery.

“I’m self-taught,” he said. “I just bumble around with this technology stuff.”

With every new generative music project, Supko has further honed his skills and broadened his personal definition of ‘music.’ He travels a lot, often toting a portable sound recorder he can activate with a switch on a pair of ear buds. Using this innocuous technology, Supko surreptitiously records noises and sounds of all sorts, from conversations in art museums to insects buzzing to the whoosh of a train zipping through a subway station. These sounds often end up as small, chopped-up, sonic moments in his music, each a single element to be randomized.

“Everything,” he said, “is fair game.”

Part of the attraction of generative music is the possibility of using the computer to discover musical ideas from a vast amount of unlikely sounds. For a recent album called s_traits, Supko and Duke art professor and media artist Bill Seaman compiled more than 110 hours of audio over three years, ranging from recordings of their own piano performances to soundtracks from decades-old documentaries.

“Everything is fair game.”

Supko wrote a software program to create some order from that chaos. The result was a 26-track, 77-minute audio recording that the New York Times called one of the top classical recordings of 2014.

“We provide way too many ingredients, and the generative software lets us try things out,” said Seaman, who co-teaches a generative art, text and music course with Supko. “John is such a perfectionist, he sometimes struggles to realize his ideas because he keeps changing things. Generative software lets him try a lot of things very quickly.”

Supko’s interest in generative music has helped him hone his teaching as well. It lets him show students how to use unpredictable processes to develop ideas and encourages them to push past musical boundaries, he says.

“There’s often a spark of musicality in the things the computer comes up with, and it’s this spark that I want my students to find in their own music,” he says. “ It has to do with doing the ‘wrong’ things imaginatively, the things that wouldn’t normally come to mind, a way of breaking the rules creatively and exuberantly.  Students seem to like this, and I find it has applications not only in the purely music-related courses I teach but also in a course like Arts Entrepreneurship, where disruptive thinking sometimes leads to successful breakthroughs.


Supko has learned at least one valuable lesson about technology – it doesn’t always make things easier. Earlier this year, Supko created an hour-long acoustic guitar concerto performed by 15 musicians at a Paris conservatory.

“... breaking the rules creatively and exuberantly.”

The musicians rehearsed written material to accompany the work's 300 sound files in advance, but they didn’t know the precise order in which they would appear during the performance. Exciting, right?

Instead of sheet music, each musician read off a laptop screen, all of which were linked via a 5G wifi network to Supko’s master computer, which determined the order of the music to be performed. Even Supko, who had created the music and written the algorithms directing his software to put the piece together, didn’t know precisely how it would turn out.

Paris Conservatoire

Students in a sound engineering program at the Paris Conservatoire took part in a two-day recording session in May 2015 using a software system Duke's John Supko created that sent notated music to each performer's computer at every moment of the piece.

“It required an enormous amount of work,” he said of the piece, called “L'Imitation du sommeil" ("The Imitation of Sleep.”) “I was just out of my mind with work. But it worked out extremely well.”

It worked out well, Supko says bluntly, because he liked how it sounded. And that very basic explanation lies at the core of his work. The pieces Supko creates aren’t for the masses; you won’t slide them over to your Workout Playlist anytime soon, and they won’t sell a zillion copies on iTunes. This music is, almost by definition, a constant work in progress. And for Supko, there’s as much thrill in the process as in the result.

But when it all works – when the algorithm he creates spits out music that rings true to him – he’s won.

“It’s like what Duke Ellington used to say,” Supko says. “If it sounds good, it is good.”

“MUSICAL DISRUPTION” is a project of Duke University’s Office of News and Communications. It was reported and written by Eric Ferreri with videography by Carson Mataxis, design by Jonathan Lee, assistance by Sonja Foust and editing by Keith Lawrence.

It is the fifth installment in the “Taking Note” series, which features profiles of Duke humanities faculty.

© Duke Today, November 13, 2015