Score: 0

Disentangling the effects of peripheral hearing loss and higher-level processes on speech intelligibility in older adults

Published: October 29, 2025 | arXiv ID: 2510.25235v2

By: Toshio Irino, Ayako Yamamoto, Fuki Miyazaki

Potential Business Impact:

Helps understand why some older people hear better.

Business Areas:
Natural Language Processing Artificial Intelligence, Data and Analytics, Software

This paper introduces a novel approach to disentangle the effects of peripheral hearing loss (HL) and higher-level processes on speech intelligibility (SI). We conducted an SI experiment with 15 young normal-hearing (YNH) listeners using stimuli processed by the WHIS simulator to emulate the hearing loss profile of a specific older adult (OA) from a previous study involving 14 OA participants. Speech-in-noise materials were presented either with ideal ratio mask (IRM) enhancement or in an unprocessed form. Results showed that the target OA achieved higher SI scores than the average YNH listener, suggesting that the OA's higher-level processes may perform more effectively than those of younger listeners. To examine the characteristics of the remaining OAs, we employed the GESI objective intelligibility measure to predict SI performance. GESI provided reasonably accurate predictions for both YNH and OA listeners. Using parameters estimated from the YNH experiment, we predicted SI scores for the 14 OA participants. The results revealed substantial variability: several OAs achieved higher SI scores than the average YNH listener, while one OA scored lower. These differences likely reflect individual variations in the efficiency of higher-level processing. Overall, these findings demonstrate that WHIS and GESI enable contrastive experiments between YNH and OA listeners, independent of hearing level, and offer a framework for investigating the role of higher-level processes in older adults on an individual basis.

Country of Origin
🇯🇵 Japan

Page Count
27 pages

Category
Electrical Engineering and Systems Science:
Audio and Speech Processing