A comparison of connectionist models of music recognition and human performance