Lip reading is a method of processing the
shape and movement of lips, recognizing and predicting
the speech pattern and translating the speech to text. It is
a method, usually used by the hearing impaired, to
understand speakers when auditory information is
unavailable and when the idea of learning a new
language is difficult. Computerized lip reading services
use image processing for recognition and classification
that are widely implemented in various applications.
There are many challenges involved in this process, like
coarticulation, homophones, etc. Deep learning using
Long-Short Term Memory is a way to help solve the
issue, in conjunction with facial feature extraction to
optimize the process. Color imaging combined with
depth sensing helps in additional improvement to the
accuracy of the classifier. And a Facial Expression
Recognition algorithm to identify face values, using these
algorithms the program detects for specific regions of the
face and tracks their movement.
Keywords : Long-Short Term Memory, Facial Expression Recognition, Face Values, Color Imaging, Deep Learning.