SPEECH AND PATTERN RECOGNITION FOR EMOTION CLASSIFICATION

Abstract

Human speech itself is a very special feature that is used for communication and expression of feelings. Speech analysis is an interesting and developing field for researchers. Physiologists and scholars from around the world are experimenting with speech as a marker for the detection of human mental physiognomies and diseases. Through speech analysis we can identify different human emotions and depressions. In our work we build a speech emotion detection system using convolutional neural network (CNN). Mel-Frequency Cepstral Co-efficient (MFCC) was used for feature extraction and speech recognition. The results showed high accuracies which were overwhelming for the start. Further, we plan to implement different other aspects of machine learning and gender recognition in speech emotion recognition and aid people with difficulty in understanding emotions.