A Stuttering Degree Diagnosis Tool Based on a Neural Network Model Trained with Multimodal Data

Ronald Xu

Vol. 2 Issue 1. p. 9- 12 (2020)

https://doi.org/10.36838/v2i1.2

Download the Article Here.

 

ABSTRACT: Stuttering causes people of all ages to have low self-esteem and trouble maintaining social relationships. Since the treatments for different degrees of stuttering are distinct, a crucial step in treating stuttering is an accurate diagnosis. Compared to existing clinical practices, the research outcome presented here is a low-cost and convenient method to diagnose stuttering based on the hypothesis that erratic facial movements and abnormal verbal speech strongly correlate with the degree of stuttering. The correlation presented in the hypothesis was modeled using a neural network trained via data from open repositories on audio recognition and facial recognition. A user-friendly software system was designed and programmed to automate the processes of video recording and processing, facial recognition, audio recognition, and stuttering degree identification. It is capable of detecting the occurrence of stuttering prior to the stutter through facial recognition, neural network, and stuttering degree quantification technologies. To the best of the author’s  knowledge, the innovative tool developed based on this research is the first of its kind designed to assist people who stutter.
KEYWORDS: Neural Network; Facial Recognition; Voice and Speech Recognition; Stuttering Degree Diagnosis