Music Genre Classification using Machine Learning A Comparative Study

Abstract

To classify songs into different genres, music researchers have used many different techniques.  However, most current approaches rely heavily on feature extraction and subsequent analysis of the  extracted features. Deep learning approaches have become increasingly popular, but a comparison  between these methods and the five traditional machine  learning algorithms was still needed to give a more  accurate representation of how effective they were.  Several experiments were run on GTZAN dataset, and obtained promising results with about 66% accuracy.