Combining face and voice modalities for person verification from video sequences | IEEE Conference Publication | IEEE Xplore