Improving acoustic modeling using audio-visual speech | IEEE Conference Publication | IEEE Xplore