Remixing the Electronic Music Landscape: Tralie Creates Revolutionary Digital Synthesizer
Music has been a central part of Assistant Professor Chris Tralie’s life since second grade. A violinist through graduate school at Duke University, Tralie’s eclectic music taste has evolved over the years.
“My parents played a lot of 80s music, then I got into classical when I started playing the violin, then came skater pop rock in high school, then electronic dubstep in college,” Tralie said.
Now, Tralie has combined his love of music and expertise in computer science to create software that is revolutionizing electronic music production.
Tralie’s research lies in the field of musical information retrieval (MIR), the interdisciplinary science of extracting information from music. His latest work involves manipulating a collection of sounds, known as a corpus, to recreate virtually any sound—a process called ”concatenative synthesis”. This technology enables music producers to transform sounds, from bees buzzing to leaves crunching, into elements that can be used in mixes and songs.
The initial idea for this research began in 2015, when Tralie attended a presentation at the International Society for Music Information Retrieval conference. Intrigued by the concept, he found the code unavailable for public use and decided to create his own version, which he published online.
“I thought it would be fun, and I wanted to share it with the community, but I didn’t think much of it after that,” Tralie said.
Tralie’s open-source implementation went live in 2018. Years later in the spring of 2021, Tralie received an email from electronic music producer Ben Cantil, known as Encanti. Cantil expressed interest in Tralie’s work, noting that although it was the only code available online, it was slow.
“A good way to get a computer scientist’s attention is to call his code slow,” Tralie said.
The code’s speed issue stemmed from an inherent limitation in the mathematical technique and algorithm from the 2015 paper. Although Tralie was eager to collaborate with Cantil, he struggled to find time to dive deeper into the problem.
“I pulled out every trick that I possibly could, but I couldn’t make it any faster,” Tralie said.
Two years later, Tralie found the inspiration to revisit the project after listening more closely to Cantil’s music while teaching similar concepts in his digital music processing class. Realizing that no one had solved the problem of speeding up the code, Traile decided to take on the challenge.
“I wanted to challenge myself and make it faster,” Tralie said. “It became a new research problem because the 2015 technique existed but still couldn’t run in real-time.”
Tralie developed a completely new mathematical approach, building on ideas from the original code. By the summer of 2023, he had a prototype, which he refined during a sabbatical in the spring of 2024. Tralie and Cantil presented their findings at a conference in San Francisco last fall.
“I solved the problem,” Tralie said. “Ben and I presented it together because I did the scientific side of it, and he brought the artistic vision.”
Tralie’s new code, called “The Concatenator,” has since been turned into a plug-in by Cantil’s company, DataMind Audio. This plug-in is designed to be user-friendly, allowing electronic musicians to use the technology without needing to interact with raw code.
“It’s the first time in my life that I’ve had my research become a tangible product,” Tralie said. “I’ve heard through the grapevine that some of my research has made its way into some companies like Spotify, but it’s still the first time I knew for sure my creative idea was at the center of a product.”
Tralie’s next goal is to refine the code, addressing timing and pitch issues that remain. He’s also exploring whether the technology could help identify cover songs, a topic he previously researched.
“I’ve had ideas, Ben’s had ideas—it’s really cool because now musicians are coming up with algorithm ideas,” Tralie said. “I think it’s all amazing, and I’m excited for the future.”