Music meets math in senior Justin Rogers’ research. For his senior project, Rogers is attempting to make a computer program that can listen to a tone and identify its note and what instrument made it.
Although it is an area most closely related with voice-to-text and speech recognition, such software would have other uses, too.
“I envision something where you can record yourself and an app that can record sheet music for you,” Rogers said. “I think that would be really cool, and it’s not too far from what I have now.”
Rogers, a math major and psychology minor, has played the bass for five years and percussion for 10 years. He will present his research, titled “Harmonic, Statistical, and Topological Methods for Audio Classification,” at noon on Thursday in Dow Science 113.
“I like music. I like math,” Rogers said. “I wanted a project that was the intersection of both of them.”
When a computer is recording a sound, it actually records the changes in air pressure on a microphone as a series of numbers. These numbers are then used to play back the sound. Rogers is working a program that can use these sets of numbers to identify the pitch and instrument of a note.
So far, the program is 85 percent successful at identifying notes and 80 percent successful identifying instruments.
“I’m not sure I would be able to identify instruments that well,” said Mark Panaggio, assistant professor of mathematics and Rogers’ adviser.
Panaggio said the research Rogers is doing is preliminary research that could be used in the future, though there are no plans to do so at this point.
“Solving the problem is not the ultimate goal,” Panaggio said. “It’s about learning.”
Even so, some musicians said that such a program would be useful.
“I think it’s a good idea if it helps people compose easier,” sophomore musician Montie Montgomery said. “It would allow them to composer music a lot more quickly.”
Both Panaggio and Rogers had to work outside their areas of expertise, dabbling in fields like algebraic topology and zero-dimensional persistence diagrams.
Rogers said he originally studied algebraic topology, which he says tries to give patterns to an “amorphous blob” of geometry, for its own sake.
“You get this crazy abstract stuff, and I’m pretty happy it’s useful in application here,” Rogers said. “I never expected to use this material.”
Still, Panaggio said they have a ways to go.
“We’re not quite at Google’s level,” he said, “but we’re having fun exploring the problem.”