Feature extraction using multimodal convolutional neural networks for visual speech recognition | IEEE Conference Publication | IEEE Xplore