Advanced Model Implementation to Recognize Emotion Based Speech with Machine Learning

Abstract

Emotions are essential in developing  interpersonal relationships. Emotions make emphasizing  with others’ problems easy and leads to better  communication without misunderstandings. Humans  possess the natural ability of understanding others’  emotions from their speech, hand gestures, facial  expressions etc and react accordingly but, it is impossible  for machines to extract and understand emotions unless  they are trained to do so. Speech Emotion Recognition is  one step towards it, SER uses ML algorithms to forecast  the emotion behind a speech. The features which include  MEL, MFCC, and Chroma of a set of audio parts are  extracted using python libraries and are used to build the  ML model. An MLP (Multi-Layer Perceptron) is used  which will be mapping the features along with the sound  file and predicts the emotion. The project details more  about the development and deployment of the model. A  technique known as "Speech Emotion Recognition" could  identify emotional characteristics in speech signals by  computer and contrasts and analysis the characteristics  parameters and the emotional change acquired. In current  market, speech emotion recognition was emerging  crossing field of artificial Intelligence.