Communication Disorder Identification from Recorded Speech using Machine Learning Assisted Mobile Application

Communication is most important to share information, opinion, thoughts with others in an effective way. Better communication is necessary to achieve in all human activities. Communication disorders problem in children leads to poor education and less involvement in academic activities. Communication disorders people may unable to participate fully and competently in day-to-day interpersonal. Limited participation in day-to-day life activities due to their communication problem leads to associated emotional and behavioral problems. In our country, educational and vocational skills training for these persons become tough because of limited resources and funding.Though SSA schools are dedicated to special children, they have limited resources and facilities. Hence, it is very difficult to provide individual care to every child with special needs, and the children remain under sublimated care and training. A computer-based Education tool will alleviate the problems. In the proposed work, the audio is preprocessed and then preprocessed audio is passed to a convolutional neural network for extracting the features. The features are fed into a fully connected layer for the classification of communication disorder.A model is created using a convolutional neural network in deep learning. The model created using a convolutional neural network is transformed into a Tensorflow model. Tensorflow model is deployed in the android for classifying the type of communication disorder. To analyze the efficiency of the proposed system, audio is recorded from the normal and communication disorder children in android application. The proposed structure has been observed that outclasses the performance of computer-based education tools to alleviate the problem of finding the particular communication disorder types.