Analysis on Lip Reading Techniques and Image Concatenation Concepts

Lip Reading is the method that visualizes the lip movement of a speaker and identifies the spoken speech. Recent developments in the field of computer science help to automate the challenging task of lip-reading. The system consists of three major stages that include detecting or localizing human faces and lips, feature extraction, and recognition of text. The accuracy of systems depends mainly on the lip localization and robustness of the extracted features. The mouth region of a face shows most of the relevant visual speech information for a VSR system. The system is useful for various application domains. In this study, phases contributing to the lip-reading task, methods used, applications, challenges, and datasets available for this task are presented along with the implementation of certain stages using the MIRACL-VC1 dataset.