A deep learning assisted web application tool for diagnosing communication disorder in children

Abstract

Communication is an important way of expressing one's thoughts to others. Many people are suffering from communication disorders like stammering and stuttering. Communicative disorders are mainly divided into three types namely language disorders, speech production disorders, and oral motor/swallowing/feeding disorders. Identifying the problem earlier and giving intervention at the right time is very important for children to improve their personal and academic lives. It will be better for the children to recover from the disorder if the problem is identified earlier. Disorder identification is a complex decision-making process that requires expertise in the medical field. The proposed framework gets the voice input through a web application. Later, deep learning technology predicts the speech disorder of the recorded voice. The deep learning-assisted tool developed in the proposed framework is to identify if the person has a stuttering problem, dysarthria, or whether the person can talk normally. Hence, the proposed method alleviates the problem of identifying speech disorder types by recording the person's voice through a web application and it achieves the accuracy of 95%.