Speech emotion recognition using kNN classifier

Abstract

There are many methods that are commonly used by researches in order to recognize speech of human emotion. In this paper, Surrey audio-visual expressed emotion (SAVEE) database are used for the analysis. The Mel-Frequency Cepstrum Coefficient (MFCC), Linear Predictive Coefficients (LPC), Linear Prediction Cepstral Coefficients (LPCC) and Weighted Linear Prediction Cepstral Coefficients (WLPCC) are extracted from these speech signals. Principle component analysis (PCA) and linear discriminant analysis (LDA) is method used to selected features. Seven typical emotions for SAVEE classified with k-nearest neighbour (kNN).