Recognition of Useful Music for Emotion Enhancement Based on Dimensional Model

Abstract

Human emotions can be evoked and recognized in many ways. The use of music stimuli is one of the simple practices to evoke particular emotions in human beings. Features extracted from physiological (EEG) signals can be used to easily recognize these emotions. The physiological signals have an unbiased nature and the results based on these signals are highly trusted, that is why the use of these signals has made advancements in the emotion recognition systems. This paper presents an overview of the procedure to identify the changes in emotions evoked by music. 25 subjects participated in the study. The study is conducted using two types of music, preferred music and relaxing music. The preferred music is selected by the participants according to their choice and hence it belongs to seven different genres of the music. Relaxing music is composed of alpha binaural beats. EEG data is collected from all the subjects before listening to music, after listening to their preferred music, and relaxing music. The data is preprocessed for artifact removal. Alpha and beta bands are extracted from the clean EEG signals. Two dimensional model based on arousal and valence is selected to recognize the changes in the emotional state. The changes in emotions, induced by both music, are measured using the arousal and valence values. These changes are compared with their baseline emotion level (the emotion level before listening to music). The results show that the level of arousal and valence increased after listening to both music. Hence it may imply that listening to music is helpful in shifting the emotions into positive state. Preferred music and relaxing music increased the arousal level by 0.8% and 30% respectively. Whereas, the increase in valence was 90% and 177% after listening to preferred music and relaxing music respectively. The arousal and valence level produced by relaxing music, was more helpful in evoking positive emotions as compared to the preferred music.