Adam Zewe writes at MIT News about Paris Smaragdis, a professor who wants machines to hear and understand sound better than people do. Smaragdis holds joint appointments in MIT’s Music and Theater Arts program and the Department of Electrical Engineering and Computer Science. He joined MIT in 2025 after 15 years at the University of Illinois at Urbana-Champaign.

His path started with classical guitar in Greece and a small synthesizer his father brought home. He studied bass at Berklee, then did his master’s, PhD, and a postdoc in Barry Vercoe’s group at the MIT Media Lab. Most students there used signal processing to handle audio. Smaragdis picked machine learning instead.

At Mitsubishi Electric Research Labs he built systems that identified what was happening on screen by listening to the soundtrack, such as spotting a sports highlight from the crowd noise. At Illinois his group used neural networks to pull individual sounds out of mixtures.…