Identification of Musically Induced Emotion: A Machine Learning Based Approach

Potential for music to evoke emotions in individuals has been a phenomenon experienced by many. This capability of music has motivated number of research work in the area of Music Emotion Recognition (MER). In this paper, we present a comprehensive data driven mechanism which ultimately provides with a model that could predict musically induced emotion with a fair level of accuracy. Subsequent to identifying emotion classes associated with songs, classification experiments were attempted for predicting the most representative emotion of a song. Naïve Bayes, Random Forest, SVM and C4.5 decision tree algorithms were attempted in this study. Among these Random Forest with oversampling produced comparative best results.