Acoustic Emotion Recognition Using Linear and Nonlinear Cepstral Coefficients

Abstract

Recognizing human emotions through vocal channel has gained increased attention recently. In this paper, we study how used features, and classifiers impact recognition accuracy of emotions present in speech. Four emotional states are considered for classification of emotions from speech in this work. For this aim, features are extracted from audio characteristics of emotional speech using Linear Frequency Cepstral Coefficients (LFCC) and Mel-Frequency Cepstral Coefficients (MFCC). Further, these features are classified using Hidden Markov Model (HMM) and Support Vector Machine (SVM).