An Intelligent System for Audio Emotion Recognition

Abstract

Emotion from audio signal processing involves analysis of many vocal parts. With the analysis of these vocal parts emotion can be extracted. Many feature extraction exists for determining emotions. Not all these features are equally important for determining emotions. In order to improve this feature selection methods are used to optimize the results. In this paper feature selection methods are used based on various acoustic features. The features are further classified by different classification scheme. Based on the obtained result it is clear that proposed method reduces feature dimension and improves classification accuracy.