Executive Summary : | One of the most important challenges in speech recognition is the ability to focus on a specific human voice while ignoring the other voices (e.g., at a party). Younger adults with normal hearing (YNH) have a remarkable ability to segregate and understand a target person's speech in the presence of multiple speakers. Whereas older adults with normal hearing (ONH) face a lot of difficulty in segregating and concentrating on a target person's voice. One of the fundamental limitations to the success of speech recognition models is the current lack of understanding of signal processing (‘neural processing') underlying reduced speech identification scores for ONH subjects. There are anatomical and physiological changes in the auditory system, especially in the inner ear and auditory-nerve (AN) fibers due to an increase in age. More specifically, with age, endocochlear potential (EP) is reduced and there is a loss of AN fibers, referred to as cochlear synaptopathy (Cs). The extent to which these changes contribute to the lower identification scores for ONH subjects is not fully understood. Concurrent vowel identification experiments are often studied to understand the multiple talkers' situation. The difference in fundamental frequency (F0) between speakers is the dominant cue to segregate the target speaker from the remaining speakers. There is only one signal-processing-based computational model in the literature that had incorporated the changes in the inner ear and AN fibers to capture the age effect on concurrent-vowel scores across F0 differences. The modeling study suggested that a reduced ability to segregate two vowels, based on F0 difference, resulted in lower identification scores for ONH subjects. However, to obtain a more precise underlying cause behind the reduced concurrent-vowel scores across F0 differences for ONH subjects, the neural representations of F0 coding of both vowels are required that can directly test the hypothesis that a reduced F0-based segregation ability resulted in poor identification scores. Thus, this proposal aims to compare the Electroencephaloghy (EEG)-based representations of F0 coding of the concurrent vowels between YNH and ONH subjects to address the reduced identification scores across F0 differences for elderly adults. Finally, identify the relationship between behavioral, computational, and EEG-based neural studies for reduced identification scores across F0 differences for ONH subjects. As this proposal addresses the challenges faced by ONH listeners in speech recognition, the findings will be more relevant and might open up new innovative research methodologies for understanding the age effects in a more realistic complex listening environment. |