Application of Music Emotion Recognition for Design of Audio Content in Terms of Affective Gaming

Abstract

Music is a medium of communicating human emotion. In fact, a research area for studying on musicality features and variations of perceived emotions in accordance has been unfolded due to this proven fact. It is referred to as Music Emotion Recognition (MER). Various researches are being conducted on determining the relationships among music structural features and perceived emotions. This research in fact, will be focusing on a practical implementation based on MER. The paper interestingly, describes on deploying a basic prototype of applying MER when playing car racing video games. This is performed with the aim of increasing the user interactivity of game players and improving the state of current video game industry to “Affective Gaming” state. The emotions experienced during racing gameplay was analyzed through player annotations. Results depicted that players frequently experience various degrees of Arousal values when playing video games. Hence, Music structural features that contribute to Arousal factor of emotions were selected. Music features are to be extracted from Electronic/Rock music genre and a machine learning model is to be built to predict emotions that will be perceived. The emotions utilized to classify music are “Happy”, “Sad”, “Anger”, “Tension” and stored in a database. This will be connected to the integrated software that will play music accordingly afterwards. The integrated software consists of the classified database, the game engine and a hardware component that captures EEG signals and analyze according to Arousal & Valence dimensions of emotions.