Paris Smaragdis
Paris Smaragdis
Professor
(217) 265-6893
3231 Siebel Center for Comp Sci

For More Information

Education

  • Ph.D. Massachusetts Institute of Technology, 2001, Advisor: Barry Vercoe

Biography

Paris Smaragdis completed his masters (1997), Ph.D. (2001), and postdoctoral studies (2002) at MIT, performing research on computational audition. Prior to the University of Illinois he was a research scientist at MERL, and Adobe Research. His research is focused on machine learning approaches to solving various audio signal processing problems. In 2006 he was selected by MIT’s Technology Review as one of the year’s top young technology innovators (TR35) for his work on machine listening, he has won the IEEE Signal Processing Society Best Paper Award twice (2017 and 2020), he was elected an IEEE Fellow (class of 2015), and selected as an IEEE Signal Processing Society Distinguished Lecturer (2016-2017).

He has served in the IEEE Signal Processing Society Board of Governors (2017-2020), as chair of the Machine Learning for Signal Processing Technical Committee of the IEEE (2012-2014), as chair of the Audio and Acoustics Signal Processing Technical Committee of the IEEE (2018-2020), as chair of the IEEE SPS Data Science Initiative (2019-2020), and as chair of the Audio and Acoustics Signal Processing Technical Committee of the IEEE, and from 2012 to 2015 he chaired the steering committee of the international community on Latent Variable Analysis and Source Separation (2012-2015).

He is currently the Editor in Chief for the ACM/IEEE Transactions on Audio, Speech, and Language Processing. He has been a Senior Area Editor for IEEE’s Signal Processing Transactions and IEEE’s Open Journal of Signal Processing. His research has been productized multiple times in commercial software that is in use by millions of users worldwide, and he holds more than 40 US patents, as well as patents in Japan and Europe. He has been an active consultant on audio technologies with multiple Fortune 500 companies.

Resident Instruction

  • CS 498 - Audio Computing Lab
  • CS 598 - Machine Learning for Signal Processing

Research Statement

Paris Smaragdis is best known for contributions in the field of audio processing, more specifically on the problem of source separation (the process of extracting an isolated signal from a mixture). His work on Frequency Domain Independent Component Analysis resulted in the first practical real-time implementations of such systems in the 90s. His later work on Non-Negative Spectral Factorizations was widely adopted for many audio applications, and more recently his introduction of end-to-end deep learning methods for source separation and denoising has resulted in wide adoption.

Smaragdis’ current research focus is on very efficient on-device processing of audio using deep learning, as well as methods for distributed learning over thousands of sensors. He is currently investigating the use of graph models for parameter-free representations of time sequences, binary network signal processing systems, online on-device learning, and is also interested in fully-differentiable digital signal processing systems.

Research Interests

  • Audio Processing
  • Machine Listening
  • Signal Processing
  • Machine Learning

Research Areas

  • Audio, speech, music and auditory processing
  • Machine learning
  • Machine learning and pattern recognition
  • Robotics, vision, and artificial intelligence
  • Signal detection and estimation
  • Signal Processing
  • Speech recognition and processing

Research Topics

Journal Editorships

  • Editor in Chief - ACM/IEEE Transactions in Audio, Speech and Language Processing
  • Senior Area Editor, IEEE Open Journal of Signal Processing, 2020-present
  • Senior Area Editor, IEEE Transactions of Signal Processing, 2015-2020
  • Associate Editor, IEEE Signal Processing Letters, 2012-2016

Professional Societies

  • Chair, Machine Learning for Signal Processing Technical Committee, IEEE Signal Processing Society, 2013-2014
  • Chair, Steering Committee, International Community on Latent Variable Analysis
  • Member-at-large, Board of Governors, IEEE Signal Processing Society, 2018-2020
  • Chair, Audio and Acoustics Signal Processing Technical Committee, IEEE Signal Processing Society, 2019-2020
  • Chair, IEEE SPS Data Science Initiative, 2018-2020

Teaching Honors

  • List of teachers ranked as excellent by their students (Fall 2019)
  • List of teachers ranked as excellent by their students (Spring 2018)
  • List of teachers ranked as excellent by their students (Fall 2017)
  • Engineering Council Outstanding Advisors (2015)
  • List of teachers ranked as excellent by their students (Fall 2015)
  • List of teachers ranked as excellent by their students (Spring 2015)
  • List of teachers ranked as excellent by their students (Fall 2014)
  • Engineering Council Outstanding Advisors (2014)
  • List of teachers ranked as excellent by their students (Spring 2014)
  • List of teachers ranked as excellent by their students (Fall 2012)
  • List of teachers ranked as excellent by their students (Fall 2011)
  • List of teachers ranked as excellent by their students (Fall 2010)

Research Honors

  • Dean's Award for Excellence in Research (2019)
  • Dean's Award for Excellence in Research (2016)

Other Honors

  • Campus Distinguished Promotion Award (2016)
  • IEEE Signal Processing Society Distinguished Lecturer (2016-2017)
  • Adobe Distinguished Inventor (2015)
  • MIT Technology Review's World's Top 35 innovators under 35 years old (TR35) (2006)
  • C. W. Gear Outstanding Junior Faculty Award (2015)
  • IEEE Fellow (2015)

Recent Courses Taught

  • CS 448 - Audio Computing Laboratory
  • CS 498 PS3 (CS 498 PS4) - Audio Computing Lab
  • CS 545 - Machine Learning for Signals
  • CS 598 PS (CS 598 PSE) - Mach Lrng for Signal Processng