Silent Speaker: A Lip-reading Model Using Deep Learning
Date
2020
Authors
Student ID
Research Supervisor
Editor
Journal Title
Volume
Issue
Journal Title
Journal ISSN
Volume Title
Abstract
Silent Speaker is an applied model of human-computer interaction. This model can be applied in various vital areas, such as crime fighting and helping the hearing-impaired. It consists of one subject: Speech Recognition. This project is done to recognize speech without the presence or support of any auditory signal. So far, a lot of research has been done on lip-reading in English, French, Chinese, and many other languages. However, little research has been done to recognize speech from a silent video in the Bengali language. This thesis work provides a new approach to detecting some words in the Bengali language using deep learning. We want to classify some words using the dataset we created for this project. We track the distance between the inner and outer lip extract features for LSTM and create our model that can classify the word. Our final accuracy is 43%, and in the future, we want to increase our dataset size and modify our model to produce more accuracy.
Description
Keywords
Citation
Department Name
Electrical and Computer Engineering
Publisher
North South University