Toward emotional speaker recognition: framework and preliminary results

Authors:
Yingchun Yang;Li Chen
Affiliations:
College of Computer Science & Technology, Zhejiang University, Hangzhou, China;College of Computer Science & Technology, Zhejiang University, Hangzhou, China
Venue:
CCBR'12 Proceedings of the 7th Chinese conference on Biometric Recognition
Year:
2012

Citing 5
Cited 0

Emotions, speech and the ASR framework

Speech Communication - Special issue on speech and emotion
An overview of text-independent speaker recognition: From features to supervectors

Speech Communication
Emotional speaker identification by humans and machines

CCBR'11 Proceedings of the 6th Chinese conference on Biometric recognition
Applying emotional factor analysis and I-vector to emotional speaker recognition

CCBR'11 Proceedings of the 6th Chinese conference on Biometric recognition
Speaker and Session Variability in GMM-Based Speaker Verification

IEEE Transactions on Audio, Speech, and Language Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Besides channel and environment noises, emotion variability in speech signals has been found to be another important factor that degenerates drastically the performance of most speaker recognition systems proposed in the literature. How to make current GMM-UBM system adaptive to emotion variability is one consideration. We thus propose a framework named Deformation Compensation (DC) for emotional speaker recognition, which viewing emotion variability as deformation (some sort of distribution distortion in the feature space) and trying to take deformation compensation by making dynamic modification on the feature, model and score level. This paper reports the preliminary results which have been gained so far, including our proposed Deformation Compensation framework together with the preliminary case study on GMM-UBM.