Classification of Music Emotion with Gaussian Mixture Model (GMM)

Abstract

Music is a link between cognition and emotion, and people are not able to share same feeling for a song. There has a need to process vast qualities of musical data. One of the operations is music emotion classification which is very popular today and an automatic extraction is needed, relating to various aspects of music. Music emotion recognition through a learning model is considered in this paper. In order to capture the salient nature of music signals features such as cepstral is applied. Classification of music signals is considered by Gaussian Mixture Model (GMM). In this approach, Thayer’s model is adopted for the description of emotions. This music mood detection approach is validated through an experimental study on a dataset containing 60 famous popular songs from English albums.