Speech Emotion Recognition and Features Extraction Based on NN Classifier

Abstract

Speech emotion recognition is a process of detecting and recognising various emotion present in the speech. In this paper various emotions are recognized by the system from features extraction based on NN classifier. Speech emotion Recognition (SER) performs an important role for measuring different emotions of a person by linking expressions with the basic emotions like happy, anger, fear, sad, surprise etc. The proposed speech emotion recognition system includes the following process-speech emotion detection, features extraction and feature selection and then finally classification. The system includes Training and Testing of dataset. In training part all the samples of the database will be trained using Neural Network classifier that will classify the dataset based on feature extraction and labelling and then the signals will be tested. Median filter will be used to remove the noise and pass signal without noise.

Keywords: RNN, DWT, NN classifiers, NN training, Speech emotion recognition (SER), Features Extraction.