LATE INTEGRATION IN AUDIO-VISUAL CONTINUOUS SPEECH RECOGNITION

Using visual information in speech recognition has been an area of interest because it can signi cantly improve the speech recognition e ciency in the conditions where audio only recognition su ers due to noisy environment. In this paper, we present a new approach to combine audio and video to improve the robustness of the speech recognition system in the noisy environments. We also compare the results of the new approach with the corresponding results of the approaches proposed earlier in the literature.