Hybrid Particle Swarm Optimization-Gravitational Search Algorithm based Deep Belief Network: Speech Emotion Recognition

One of the most important research areas is the Speech Emotion Recognition (SER) technique, which is applied in many fields such as speech processing and human-computer interaction. In general, it is mainly concentrated on using the techniques of machine learning in order to predict the precise emotional category from speech. In affective computing as well as the human-computer interaction, the developed applications of SER are very effective that are considered as the important module of the computer's next-generation system. It is due to the automatic service provisions are granted by the natural human-machine interface that requires an improved approval of user emotional conditions. Hence, this work proposes a novel SER model which integrated both emotion and gender recognition. Various features are extracted and that is fed for the emotions classifications. In this work, the Deep Belief Network (DBN) is exploited. At last, performance analysis of the developed technique is seen that better accuracy rate while comparing with the conventional models. This work proposes a novel technique for the SER model which helps both emotion as well as gender recognition. Here, the proposed Hybrid Particle Swarm Optimization (PSO) and Gravitational Search Algorithm (GSA) algorithm are introduced to identify the optimal weight of the DBN technique.