Asynchronous stream modeling for large vocabulary audio-visual speech recognition | IEEE Conference Publication | IEEE Xplore