Music player based on emotion recognition of voice signals

In this paper, a smart music system is designed by recognizing the emotion using voice speech signal as an input. The objective of the speech emotion recognition (SER) system is to determine the state of emotion of a human being's voice. This study recognizes five emotions-anger, anxiety, boredom, happiness and sadness. The important aspects in implementing this SER system includes the speech processing using the Berlin emotional database, then extracting suitable features and selecting appropriate pattern recognition or classifier methods to identify the emotional states. Once the emotion of the speech is recognized, the system platform automatically selects a piece of music as a cheer up strategy from the database of song playlist stored. The analysis results show that this SER system implemented over five emotions provides successful emotional classification performance of 76.31% using GMM model and an overall better accuracy of 81.57% with SVM model.