A comparison of human and machine in speaker recognition