Publication

Abstract : Lip-reading depends much on the context and the knowledge of language as it does on visual clues and is notoriously challenging. Human lip-reading is a tedious task that requires not only the knowledge about the language but also the visual evidences to approximately or appropriately predict the spoken words. Even the speech recognition in the noisy environment can be made possible through the lip-reading technique. The accuracy achieved through manual lip reading is 40% which can be improvised using AI deep learning techniques. To improve the accuracy of predicting phrases from the low-resolution videos. Enhanced video articulation (EVA) is an articulatory lip-reading technique determining the labial movements of a person. The face of the person is recognised and is segmented for the labia. The process counts on neural networks, AI algorithms comprising many simple computing components coupled together that learn and process information in a way similar to the human brain. As a result, we infer that the deep learning architectures perform similarly to the traditional methods but will improve the word recognition rates.