Emotion Recognition from German Speech Using Feature Selection Based Decision Support System

The speech signal plays a vital role in human communication. From last very years emotion characterization from speech has gained popularity in domain of signal and speech processing. In this article we propose a decision support system for emotion recognition from German speech signal database. A set of 535 speech signals with 7 emotion labels are investigated using prominent feature descriptors viz, Mel frequency coefficients, signal energy, pitch , zero crossing rate and formant feature numeric’s to generate a set of 37 features for each speech signal respectively. Decision support system is modelled in feature space using Support Vector Machine and resultant classification accuracy is further enhanced using Information Theory based feature selection measures to achieve the highest test classification accuracy of 93 percent respectively.