Classification of Familiarity Based on Cross-Correlation Features Between EEG and Music

摘要

An approach to recognize the familiarity of audlistener with music using both the electroencephalogram (EEG)udsignals and the music signal is proposed in this paper. Eightudparticipants listened to melodies produced by piano sounds asudsimple natural stimuli. We classified the familiarity of eachudparticipant using cross-correlation values between EEG and theudenvelope of the music signal as features of the support vectorudmachine (SVM) or neural network used. Here, we report thatudthe maximum classification accuracy was 100% obtained byudthe SVM. These results suggest that the familiarity of musicudcan be classified by cross-correlation values. The proposedudapproach can be used to recognize high-level brain states suchudas familiarity, preference, and emotion.ud