Automatic Genre Classification of Latin Music Using Ensemble of Classifiers

This paper presents a novel approach to the task of automatic music genre classification which is based on ensemble learning. Feature vectors are extracted from three 30-second music segments from the beginning, middle and end of each music piece. Individual classifiers are trained to account for each music segment. During classification, the output provided by each classifier is combined with the aim of improving music genre classification accuracy. Experiments carried out on a dataset containing 600 music samples from two Latin genres (Tango and Salsa) have shown that for the task of automatic music genre classification, the features extracted from the middle and end music segments provide better results than using the beginning music segment. Furthermore, the proposed ensemble method provides better accuracy than using single classifiers and any individual segment.