What is it about?
This work compares ASR accuracy (Google speech-to-text) to intelligibility and clinician ratings of speech impairment. We found that ASR accuracy is less consistent than intelligibility and aligns with clinician ratings only in a general sense. ASR may be appropriate in some instances to determine dysarthria severity, but it should not be used as a one-to-one substitute for intelligibility.
Featured Image
Photo by Volodymyr Hryshchenko on Unsplash
Why is it important?
ASR systems are becoming more accurate and more easily accessible. Because they compute accuracy automatically and objectively, they may offer a more efficient approach to rating speech intelligibility and severity in people with dysarthria. Currently, ASR is being used by some clinicians and in research settings, particularly those that have collected large amounts of speech data. It is important to understand the limitations and appropriate uses of this new tool.
Perspectives
I hope this article helps clinicians and researchers who are looking for more efficient ways to evaluate speech better understand ASR as a tool, and possibly alert them to this potentially helpful measure of speech impairment. Because we found that ASR accuracy is an imperfect proxy for speech intelligibility in people with dysarthria due to ALS, this work could provide a reason for developers to improve ASR as a clinical tool.
Sarah Gutz
Harvard University
Read the Original
This page is a summary of: Validity of Off-the-Shelf Automatic Speech Recognition for Assessing Speech Intelligibility and Speech Severity in Speakers With Amyotrophic Lateral Sclerosis, Journal of Speech Language and Hearing Research, June 2022, American Speech-Language-Hearing Association (ASHA),
DOI: 10.1044/2022_jslhr-21-00589.
You can read the full text:
Contributors
The following have contributed to this page
