Interaction Tool for the Deaf and Mute Using Neural Networks

This paper proposes a complete communication model which can provide aid to the deaf and mute people in a way helping them to communicate with the normal people. The model is a personalized deep learning structure which is trained according to the physically disabled person's behavior and character. Integration of the conventional neural nets like the Hidden Markov Model, the Long Short Term Memory and the WaveNet neural networks are used to create a structure for the communication model. The paper delineates hidden markov model for converting the speech to text which can be accessible to the deaf and mute person. To formulate a reply based on the text formed by speech, Long Short Term Memory Neural Networks are used. WaveNets are described in order to convert the textual reply into audio speech so as to be heard by the other person involved in communication.