Multimodal learning using 3D audio-visual data for audio-visual speech recognition | IEEE Conference Publication | IEEE Xplore