Özet:

Very successful results have been obtained in areas such as computer vision and voice recognition when applying deep learning methods. Technologies that facilitate the lives of people have been developed as a result of the successes of deep learning within these areas. One of these technologies is voice recognition devices. Research has shown that these devices do not give good results in noisy environments; although, they do give good results in silent environments. With deep learning methods, voice recognition in noisy environments can be achieved using visual signals. Thanks to computerized vision, the success of voice recognition devices can be increased with the analysis of human lips in order to determine what the speaker is saying. In this study, lip-reading studies using deep learning methods published between 2017 and 2020 were examined and data sets were introduced. As a result of the study, it is seen that CNN and LSTM architectures are used more intensively in lip-reading studies, hybrid models are preferred more and the success rates are increasing day by day. In this context, it is seen that technologies that can be used in line with the need can be developed by conducting more academic studies on lip reading.

Anahtar Kelimeler: