A Neural Network Based Approach for Recognition of Basic Emotions from Speech

Recognition of emotions from the speech is one of the most researched topics now a days in the field of signal processing and human machine interaction system. Unlike humans, machines have a lacking ability to perceive and show emotions. However, human-computer interaction can be improved with emotions recognition automation system by reducing the need for human access. In this research paper, a method for speech emotion recognition is presented using Neural Network (NN) with Mel Frequency Cepstral Coefficients (MFCC) features. Berlin Emotional Speech dataset (EmoDB) is used here for classification purposes. From our research, we have found an average of 73.40% accuracy with the highest accuracy of 82.35%.