A Comparative Analysis of Different Approach for Basic Emotions Recognition from Speech

Abstract

Human-Computer Interaction is one of the most progressive technology of recent times, and speech emotion recognition is one of the factors working behind this technology. The communication between two humans via verbal expressions is a tough process when this process is about to be implemented on robots or any system. If the robot or system cannot recognize the emotion that it receives from the speech, it will be a disastrous situation. This article shows some Machine Learning and Deep Learning approaches for recognizing emotions from speech. In this paper, many algorithms were compared to bring out a better model to perform the task more precisely. Algorithms like Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), Deep Neural Network (DNN), and Convolutional Neural Network (CNN) were used here. Among these algorithms, the CNN model has performed well by giving an accuracy of more than 85%. Moreover, this customized CNN model has also outperformed all the models that were applied to the Emo-DB dataset.