does it seem to accurately reflect voices that are perceptually similar?