01-05-2017, 04:08 PM
Automatic recognition of the emotional states of human speech is a current research topic with a wide range. In this paper we have tried to recognize and classify the emotion of speech from three linguistic databases, namely databases of emotions from Berlin, Japan and Thailand. Fundamental Frequency (F0), Energy, Zero Crossing Rate (ZCR), Linear Predictive Coding (LPC) and Mel Frequency Cepstral Coefficient (MFCC) of shortwave signals are extensively investigated. In this sense, Vector Support Machines (SVM) are used as the classification model. Empirical experimentation shows that the combined characteristics of F0, Energy and MFCC provide the highest accuracy in all databases provided using the linear core. It gives a classification accuracy of 89.80%, 93.57% and 98.00% for the emotion databases of Berlin, Japan and Thailand, respectively.
