Emotion Recognition Using Speech Processing

Since many years ago, emotion identification from voice processing has been a significant factor in applications involving human-machine interfaces. In the psychological well-being of humans, emotions are pertinent. It serves as an agent for communicating one's viewpoint or mental condition to others. Speech Emotion Recognition (SER) is a manner of detecting the speaker's emotional state from the speech signal. Any computer system with limited processing resources may be programmed to sense or generate the few universal feelings, like Neutral, Anger, Happiness, and Sadness as needed. The following characteristics are retrieved in this work: Mel-frequency cepstral coefficients (MFCC), Chromogram, Mel scaled spectrogram, Spectral contrast, and Tonal Centroid. The emotion in this study is regarded using a deep neural network. The categorization of the speech in the output layer is done using Softmax. We used the combined 1440 audio recordings from 24 individuals in the RAVDASS speech dataset for training. Speech DNN has a 96% accuracy rate for emotion identification, which is greater than other algorithms like KNN, LDA, and SMO. Automatic emotion detection from human speech is becoming increasingly prevalent these days as it boosts interactions between computers and humans.