Emotion Regulation Music Recommendation Based on Feature Selection

Chinese traditional music has been proved to be effective in emotion regulation for thousands of years. Five different groups of Chinese traditional music which have been proved can regulate different emotions (Angry, Depressed, Feverish, Desperate, Sorrowful) in the literature. 54 audios features are extracted by using the Librosa library for each music group. Five features are manually selected using histogram analysis which show significant difference between the five groups of music. Combined with KNN, SVM and Deep forest classification algorithms, the five manually selected audio features are shown to have better classification performance than traditional feature selection algorithms, like PCA and LDA. We hypothesize that these five significant audio features may be the underlying basis why so such music can effectively perform emotion regulation. Based on this classification models, prototype emotion regulation music recommendation interface (TJ-ERMR) was built that can be used for music therapy. In the future, we will use this classification model to find more music to expand the initial repertoire of our music recommendation system.