Computer Science Location: Zoom (email presenter for link)
Add to Calendar 2020-06-09T14:00:00 2020-06-09T14:00:00 America/New_York PhD Qualifying Exam Presentation by Lahiru Nuwan Wijayasingha Frequency domain features and their robustness for spoken emotion classification with CNN Abstract: Zoom (email presenter for link)

Frequency domain features and their robustness for spoken emotion classification with CNN

Abstract:

Speech Emotion Recognition (SER) is an important task since emotion is an important dimension in human communication. But  SER  systems  are  very  sensitive  to  environmental  noises. CNNs are the state-of-the art machine learning technique used to  solve  this  ongoing  problem.   Literature  indicates  that  frequency  domain  features  obtained  via  Fourier  transformation and related techniques perform better than the other types such as time domain features.  Research indicates that various features and their combinations have varying impacts on the performance and noise-robustness of machine learning models. This aspect is not been thoroughly investigated for SER in the past.  The aim of this research is to study the performance of CNN based SER systems using several different frequency domain features and their combinations.

Committee:

  • Madhur Behl (Chair)
  • John Stankovic (Advisor)
  • Yangfeng Ji
  • Brad Campbell