Prediction of Emotion Stimulated by Music

Abstract

Existence of a mechanism for predicting musically induced emotions is of value to the music therapy domain, while being a study of considerable interest. Despite numerous studies having been conducted in this area, certain limitations as the lack of a music specific emotion model has direly affected the conclusions of those studies. This research has focused on creation of a music specific emotion model, whereas an array of experiments have been conducted to realize a fair music emotion prediction model. The best prediction model was realized to be audio feature based music emotion prediction incorporating oversampling and Random Forest algorithm.