Making machines that can listen
Professor Paris Smaragdis develops computers with superhuman hearing for applications like speech recognition, music understanding, and environmental monitoring.
MIT Professor Paris Smaragdis SM ’97, PhD ’01 is working on creating artificial perception systems that can hear and comprehend sound better than humans. His goal is to develop machines capable of rapidly identifying the source of a traffic accident or detecting potential failures in industrial equipment. Smaragdis, who holds joint appointments in music and electrical engineering, draws parallels between his work and computer vision research, but focuses on the auditory domain.
He has been particularly interested in developing machine learning techniques for sound source separation and applying these tools to enhance music creation and understanding. Growing up in Greece, Smaragdis developed a passion for music at an early age, particularly synthesizers and jazz fusion. His background in computers and synthesizers led him to study music technology at Berklee College of Music, where he collaborated with MIT Media Lab faculty to create software synthesizers.
He later pursued advanced degrees at the Media Lab, where his research shifted towards training machine-learning models to understand sound, a less conventional approach compared to traditional signal processing techniques. Smaragdis has applied his expertise to various projects, including creating tools for Adobe's audio research team and assisting with archival sound restoration for a documentary mini-series.
Written by urgent.news from MIT News AI's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
