MIT Professor Paris Smaragdis Advances Machine Learning for Audio Perception
Paris Smaragdis, a researcher specializing in machine learning for sound, has joined the faculty at MIT to explore advanced audio perception. His work focuses on developing systems that allow machines to interpret complex sound environments, with applications ranging from industrial monitoring to music production. Smaragdis previously contributed to the audio restoration for the 2021 documentary series The Beatles: Get Back, which earned an Emmy Award in 2022.
Key points
- Paris Smaragdis joined the MIT faculty in 2025 to research machine hearing and audio processing.
- His research group explores machine learning for sound source separation and smart audio augmentation.
- Smaragdis previously held positions at Mitsubishi Electric Research Labs and Adobe.
- He contributed to the audio processing for the 2021 documentary The Beatles: Get Back, winning an Emmy in 2022.
- Current research includes developing flexible signal processing algorithms and integrating large language models into audio systems.
Advancing Machine Hearing
Paris Smaragdis, who joined the MIT faculty in 2025, is focused on teaching computers to listen and understand sound. His research aims to create systems capable of identifying specific audio events, such as pinpointing traffic accidents or detecting mechanical failures in industrial equipment.
Smaragdis notes that while computer vision has made significant strides, machine hearing remains an underexplored domain. His group is currently investigating algorithms that operate without fixed sampling rates or frame sizes to create more flexible and efficient audio processing systems.
Background and Industry Experience
Before his academic career, Smaragdis worked at Mitsubishi Electric Research Labs, where he developed systems to analyze video soundtracks, and at Adobe, where he led an audio research team. His work often bridges the gap between technical research and practical application for artists and musicians.
His academic contributions include a project utilizing neural networks for sound separation, which was applied to the 2021 documentary series The Beatles: Get Back. This work helped the production team win an Emmy Award for sound editing in 2022.
Why it matters
The research aims to move beyond traditional signal processing by using machine learning to enable computers to interpret sound in ways that mimic or exceed human perception.
What we know
- Paris Smaragdis joined the MIT faculty in 2025.
- Smaragdis previously worked at Mitsubishi Electric Research Labs and Adobe.
- Smaragdis contributed to audio processing for the 2021 documentary mini-series The Beatles: Get Back, which won an Emmy Award in 2022.
- Smaragdis holds joint appointments in the Music and Theatre Arts program and the Department of Electrical Engineering and Computer Science at MIT.
What remains unclear
- The long-term efficacy of using large language models for audio enhancement remains a subject of ongoing research.
- The timeline for the potential revival of the MIT underwater hockey club is not specified.
