A Strategic Approach for Robust Dysarthric Speech Recognition

The development of a system to recognize the speeches of standard speakers has been in practice for many decades. Research development is still progressing to implement a strategy to identify the speeches uttered by people with hearing impairment/Autism spectrum disorder/dysarthria. This work includ...

Full description

Saved in:
Bibliographic Details
Published in:Wireless personal communications 2024-02, Vol.134 (4), p.2315-2346
Main Authors: Revathi, A., Sasikaladevi, N., Arunprasanth, D., Amirtharajan, Rengarajan
Format: Article
Language:English
Subjects:
Accuracy
Autism
Communications Engineering
Computer Communication Networks
Dysarthria
Engineering
Intelligibility
Networks
Signal,Image and Speech Processing
Speech processing
Speech recognition
Speeches
Voice recognition
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The development of a system to recognize the speeches of standard speakers has been in practice for many decades. Research development is still progressing to implement a strategy to identify the speeches uttered by people with hearing impairment/Autism spectrum disorder/dysarthria. This work includes various speech enhancement techniques to increase the intelligibility of spoken utterances. This system uses perceptual features and different modelling techniques for developing a dysarthric speech recognition system. Perceptual features are extracted from raw speeches, and intelligibility-enhanced spoken utterances and models are created. The design features extracted from the test utterances are given to the models, and based on the classifier used, the test utterance is identified to be associated with the model. An Implementation of speech enhancement techniques would facilitate better accuracy. Decision-level fusion classification on integrating features, models, and speech enhancement techniques has provided overall accuracy of 81% for recognizing isolated digits spoken by a few dysarthric speakers. Better accuracy can be ensured for the database containing more utterances from many dysarthric speakers. This system would help caretakers understand the speeches uttered by persons affected with dysarthria to provide the necessary assistance.
ISSN:0929-6212
1572-834X
DOI:10.1007/s11277-024-11029-y