Automatic music mood recognition using Russell’s twodimensional valence-arousal space from audio and lyrical data as classified using SVM and Naïve Bayes

Abstract

Automatic music mood recognition is still a new field of research that is gaining attention in the last decade. This study created a system that predicts which of the four quadrants of the valence-arousal space the song belongs to. The system used support-vector machine (SVM) for audio features while Naïve Bayes was used for lyrical features. audio classification achieved a high accuracy for arousal while lyrics classification achieved a high accuracy for valence.