November 05, 2009 —
A team of USC researchers from the Signal Analysis and Interpretation Laboratory (SAIL), directed by Professor Shri Narayanan, the Signal and Image Processing Institute (SIPI), and the Ming Hsieh Department of Electrical Engineering won the Interspeech 2009 Emotion Challenge Award. The team was led by Chi-Chun (Jeremy) Lee, a graduate student. Other team members included Emily Mower, also a graduate student, Sungbok Lee, a Research Assistant Professor, Shrikanth Narayanan, Professor, and Carlos Busso, a graduated former Ph.D. student, who is currently an Assistant Professor of Electrical Engineering at the University of Texas, Dallas.
From left: Carlos Busso, Jeremy Lee, Emily Mower, Shri Narayanan.
[Not in picture: Sungbok Lee]
The award recognizes the best performance in a competition for automated recognition of emotion in speech held as part of the 10th Annual Conference of the International Speech Communication Association held in Brighton, U.K., in September 2009. Competitors tested their emotion recognition algorithms on a standardized data base consisting of nine hours of spontaneous, emotionally colored, unscripted speech. The USC team won in the Classifier Sub-Challenge category for their work published in the paper entitled “Emotion Recognition Using a Hierarchical Binary Decision Tree Approach”.
A key technique used in this winning approach for multi-class emotion pattern classification, in contrast to conventional ways, is to do split the problem into a series of binary classifiers, and importantly, using a strategy of attacking the easiest categories first and dealing with the more ambiguous ones at later stages of the processing. Details of this research, and other on going efforts on emotions research, can be found at http://sail.usc.edu.