Patrick J. Wolfe, an associate professor in electrical engineering, has been using statistical models to find new ways to reduce noise and improve sound quality in audio tracks.

Wolfe’s research examines specific audio imperfections—for example, the hissing of an analog tape, or the crackles and pops in a vinyl recording—and how the human mind perceives those imperfections.

Wolfe uses a special statistical technique called Bayesian methodology to model what people expect when they listen, classifying sounds by whether they are “signal” or “static”—in other words, whether they seem natural or jar the ear when heard.

By utilizing such statistical models, audio engineers can better determine the signal behind an audio recording and thus filter music from noise when restoring old tracks.

The applications of Wolfe’s research extend beyond remastering recordings. The same principles can be applied to improving hearing aid technology and developing better speech-recognition tools, as well as reducing background noise in situations as ubiquitous as cell phone conversations, according to the Harvard School of Engineering and Applied Sciences website.

Wolfe, who came to Harvard in 2004, has spent his career integrating diverse academic fields, including engineering, applied mathematics, statistics, and music.

Currently Wolfe is researching a wide range of topics, including signal processing in visual data, network interactions (including social networks), and the functioning of vocal mechanisms in humans.

Even though Wolfe teaches electrical engineering at SEAS, he is also affiliated with the statistics department and in 2004 founded the Statistics and Information Sciences Laboratory at Harvard.

“I’ve been doing [interdisciplinary research] for so long I kind of take it for granted. Even in college I had to shuttle between singing in class and taking tests on topics like electromagnetism,” Wolfe said, referring to his undergraduate years at University of Illinois Urbana-Champaign where he majored in both electrical engineering and music.

Wolfe also credits the structure of SEAS as a major boost to his interdisciplinary approach.

“The fact that we do not have separate departments is very important. We’ve very firmly retained that culture of not having siloed or stove-piped different fields.”

In his faculty profile, Wolfe cites another reason for joining the Harvard community six years ago.

“The chance to join Harvard is like what they say about the Mafia: It’s an offer you can’t refuse.”