Application of Automatic Speech Recognition in Communication Sciences and Disorders

Abstract

Communication Sciences and Disorders encompass scientific understanding of communication in both typically-developing and communication-disordered population. During this process Speech language pathologists should incorporate new strategies and technologies for diagnosis and rehabilitation. One such Technology is Systematic Analysis of Language Transcripts (SALT) that manages the process of eliciting, transcribing, and analyzing language samples. However, the process seems to be time consuming and sophisticated. In the field of Speech Language Pathology automatic detection of the speech would enable the clinicians for quick decision making. Hence, a speaker adapted automatic speech recognition tool was developed using Sphinx and incorporated Structured Query Language (SQL). The participants in the control group included 30 subjects (15 males and 15 females) within the age range of 18-30 years. The clinical group included 15 participants with 3 clinical conditions such as stuttering, voice disorder, and dysarthria. Participants were instructed to read the passage into the microphone connected to the computer with the designed tool in an acoustic treated room. The data was further analyzed for the automatic speech recognition score. Word recognition accuracy was found to be around 75%. The study concluded that the tool can be further refined and can assist Speech Language pathologists in clinical practice.