From Data to Emotions: Affective Computing in Voice Emotion Detection

Emotion recognition and analysis seeks to automate the complex language of human emotions. Understanding human behavior, increasing communication, and improving different parts of our lives all depend on our capacity to effectively identify and analyze emotions. The study of emotions brings together computer technology, artificial intelligence, and psychology. The accessibility of appropriate emotional databases, the discovery of the pertinent feature vector, and the selection of appropriate classifiers are just a few of the difficulties faced by voice emotion recognition systems and had a major challenge in front of researchers. This chapter critically analyzes the literature on emotion recognition through speech, in terms of audio databases, audio features, conventional ML (machine learning), and deep learning algorithms. Also, proposes a methodology for voice emotion recognition using PCA (principal component analysis) and CNN (convolutional neural network) algorithms for voice emotion recognition to increase recognition rates. This chapter has identified popular speech databases and associated accuracy levels attained by deep learning (DL) techniques.