Virtual speech therapy system: Speech intervention application for children with autism spectrum disorder
With technological advancement, the integration of digital technology into education has become possible. Nowadays, parents are more open to adopting and using technology like software applications ("apps") to enrich teaching and learning processes for the benefit of their children. Based...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English |
| Published: |
2023
|
| Subjects: | |
| Online Access: | https://eprints.ums.edu.my/id/eprint/40555/1/24%20PAGES.pdf https://eprints.ums.edu.my/id/eprint/40555/2/FULLTEXT.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | With technological advancement, the integration of digital technology into education has become possible. Nowadays, parents are more open to adopting and using technology like software applications ("apps") to enrich teaching and learning processes for the benefit of their children. Based on research, there is an increase in the use of apps in early education. Apps are also being utilized as intervention tools for children with Autism Spectrum Disorder (ASD). Most apps are solely to serve as education resources rather than a virtual therapy system, lacking the evidence-based standard teaching: the Applied Behavioural Analysis (ABA). Hence, this research aims to develop a Virtual Speech Therapy System (VSTS), incorporating the ABA strategy, particularly Discrete Trial Training (DTT), for children with ASD. Two key objectives were set for the VSTS to be realized: (1) To transform and implement the conventional DTT into a software application, and (2) To develop a speech recognizer implementing the Mel-frequency Cepstrum Coefficients (MFCC) and Hidden Markov Model (HMM). In achieving the first objective, the practicality of the coding method and design of the DTT-based lesson module is assessed and investigated. Next, a speech recognizer is developed, trained, and tested using 2904 speech samples. The speech recognizer is developed using the Hidden Markov Model (HMM) that utilizes 39 Mel Frequency Cepstral Coefficient (MFCC) features. The speech recognizer achieves 95.2% and 95.0% training and testing accuracies for English speeches, respectively. For Malay speeches, the recognition performance is 94.38% for training and 93.75% for testing, respectively. This research contributes to developing a speech intervention application that employs the ABA technique, namely the DTT. This intervention application is not intended to replace therapists but to enhance learning outcomes. |
|---|