MEDIA, Pa. — Does exact timing or imprecise emotion make for better music? Artificial intelligence (AI) might be able to help answer the question, according to Michael Yatauro, associate professor of mathematics at Penn State Brandywine, and Arnav Pandey, a second-year mathematics major who has transitioned to University Park from Brandywine.
The two are collaborating to see how a drummer’s playing compares to exact metronome time, and they were selected to present their work at the American Mathematical Society’s Joint Mathematics Meeting in Washington, D.C., in January.
“Arnav and I are both drummers. He has marching band experience and is in the Blue Band, while I have more of a rock band background,” Yatauro said. “We ended up landing on this project that was inspired by papers we read that looked into how music can be more about emotion than exact timing. There’s precision, but it’s not perfect or on the exact beat. We’re trying to look at how to analyze the different beats statistically. We’re combining our passions for mathematics and drumming, which I think is great.”
Yatauro and Pandey used an AI tool called Moises, a toolkit for musicians that takes a soundtrack and breaks it down by instrument. Using this tool allowed them to listen only to the drum beat in the songs.
“The nice thing about drumming is that it’s very staccato in terms of its sound. When the drum is struck, it creates a waveform that can be analyzed more accurately,” Yatauro said, explaining that they analyze the waveforms using Moises. “Since we’re doing this for a full drum set, we don’t want to just listen and pause, since it ends up creating a lot of discrepancies and errors. Instead, we use Moises to separate the drum track, and we feed the waveform into Google Colab, another AI tool that allows you to write and execute Python code — a programming language — in your web browser, to get the exact moments the drum is struck.”
They are still planning on what the findings from this research will be used for; as of right now, they are running tests on sample data — song clips — to test the limits of the process and identify areas for improvements or adjustments.
In terms of applications, Yatauro said this research could be used to produce more realistic sounding drum machines for styles of music that use them.
“We are also looking to understanding the type of timing that produces music that is perceived as ‘appealing.’ This gets into a process of studying randomness in the context of music,” he said. “Some call this randomness ‘pink noise’ to distinguish it from other forms of randomness, such as white or brown noise.”
Since Pandey is studying at University Park, he and Yatauro meet once every week over Zoom.
“The research project has been very fulfilling for me personally because it’s allowed me to prove to myself that math can be applied in creative outlets,” Pandey said. “Combining two huge spheres in my life — music and academics — has always been a goal of mine, and since Dr. Yatauro and I are both drummers who like math, this project was almost waiting to happen.”
Pandey also noted that the project provided an education beyond the classroom.
“Learning about the research process, manually collecting data on various music charts and trying out a variety of different methodologies has provided me with an invaluable learning experience unlike anything I have gotten from a traditional class,” Pandey said.