A Study on Speech Emotion Prediction using Deep Learning Algorithm

The method of recognizing human emotion is known as emotion recognition. People's ability to recognize other people's feelings varies greatly. One of the best algorithms for predicting and managing voice and signal data is the neural network. The. wav format of voice data is used here. Speech emotion is one of the most important aspects of human emotion. Here, two techniques are used to find the emotion of that person using that data. 1) Extraction of features, 2) Prediction. In feature extraction librosa module is used to convert.wav format file to datasets. Librosa module containsMel-frequency cepstral coefficients (mfcc), chroma and mel sub module to calculate the signal. Eight distinct kinds of feelings are taken. They are depressed, joyful, relaxed, neutral, furious, afraid, disgusted, and shocked. One can extract the features from that data. Using the Sound file module to access a. wave file before function extraction. Split the dataset into two parts after extraction. There are two types of data: 1) training data and 2) test data. The performance model is predicted using the training dataset.