Modeling and Optimization of Signals Using Machine Learning Techniques

18Human Emotion Recognition Intelligence System Using Machine Learning

Bhakthi P. Alva^*, Krishma Bopanna N., Prajwal S., Varun A. Naik and Lahari Vaidya

Department of Electronics and Communications Engineering, Sahyadri College of Engineering & Management, Mangaluru, India

Abstract

The human voice is versatile and displays substantial emotional variances, allowing for a better understanding of human behaviour. This model is proposed to benefit the visually impaired population to successfully engage with people and socialize by recognizing their emotions. A speech emotion recognition (SER) system is now being developed, which is based on various classification models and feature extraction algorithms. Mel-frequency cepstrum coefficient (MFCC) characteristics are obtained from voice signals and these are utilized to train some classifiers. Feature selection (FS) has been used to obtain a relevant feature subset. A wide range of machine learning (ML) methods have been used for the emotion classification challenge. To initiate, the K-nearest neighbors (KNN) algorithm is used to detect seven distinct emotions. Their results will be compared to support vector machine (SVM) approaches and are commonly used in the domain of emotion recognition for voice signals. The Toronto emotional speech set (TESS) and knowledge extraction based on evolutionary learning (KEEL) databases comprise the experimental dataset. With the objective of more accurately identifying speech percepts based on ...