Emotion Recognition from Speech using Discriminative Features

Abstract

Creating an accurate Speech Emotion Recognition (SER) system depends on extracting features relevant to that of emotions from speech. In this paper, the features that are extracted from the speech samples include Mel Frequency Cepstral Coefficients (MFCC), energy, pitch, spectral flux, spectral roll-off and spectral stationarity. In order to avoid the 'curse of dimensionality', statistical parameters, i.e. mean, variance, median, maximum, minimum, and index of dispersion have been applied on the extracted features. For classifying the emotion in an unknown test sample, Support Vector Machines (SVM) has been chosen due to its proven efficiency. Through experimentation on the chosen features, an average classification accuracy of 86.6% has been achieved using one-v/s-all multi-class SVM which is further improved to 100% when reduced to binary form problem. Classifier metrics viz. precision, recall, and F-score values show that the proposed system gives improved accuracy for Emo-DB. General TermsPattern Recognition

Machine Learning Algorithms for Detection and Classifications of Emotions in Contact Center Applications

“…The following parameters were selected for the speech signal: f 0 (fundamental frequency), MFCC (Mel Frequency Cepstral Coefficients), jitter and shimmer. It was determined that the first 13 MFCC coefficients describing the frequency parameters of speech would be used for emotion recognition because they contain most of the information regarding the emotion to be recognized [ 77 ]. The fundamental frequency f 0 , on the other hand, contains information about the pitch of the voice, and therefore allows us to take, e.g., gender into account, without the need for additional determination.…”