Abstract:
It is shown how recognition performance in automated speech preception can be significantly improved by additional lipreading, so called speech-reading. It is shown on an extension of an existing state-of-the-art speech recognition system, a modular multi-state time-delay neural network (MS-TDNN). The acoustic and visual speech data are preclassified in two separate front-end phoneme TDNNs and combined to acoustic-visual hypotheses for the dynamic time warping algorithm. This is shown on a connected word recognition problem, the letter spelling task. With speech-reading the error rate can be reduced up to half of the error rate of pure acoustic recognition.
copied = false, 2000);
">
@conference{Bregler-1993-15947,
author = {C. Bregler And S. Manke And H. Hild And Alex Waibel},
title = {Bimodal sensor integration on the example of "speechreading"},
booktitle = {Proceedings of IEEE International Conference on Neural Networks (ICNN '93)},
year = {1993},
month = {March},
volume = {2},
pages = {667 - 671},
}
author = {C. Bregler And S. Manke And H. Hild And Alex Waibel},
title = {Bimodal sensor integration on the example of "speechreading"},
booktitle = {Proceedings of IEEE International Conference on Neural Networks (ICNN '93)},
year = {1993},
month = {March},
volume = {2},
pages = {667 - 671},
}