Author: Riya O. SinghPublisher:
ISBN:
Category :
Languages : en
Pages :
Book Description
This research investigates the longstanding problem of understanding human speech perception. We aim to study speech perception and decode perceptual cues in speech by conducting psychoacoustic experiments on several subjects by presenting them with nonsense consonant-vowel (CV) syllables in various kinds of masking noise at di fferent signal-to-noise ratios (SNRs). Our research with a large number of normal hearing (NH) listeners shows that NH speech perception is deterministic and the error is essentially zero when the main perceptual feature (or event) of the utterance is audible. With the perceptual CV cues precisely known, one can predict how an average normal hearing (ANH) listener would behave in a certain type and degree of masking noise. The next major goal of the current research is to characterize hearing-impaired (HI) ears by using our knowledge of specific consonant speech cues in ANH ears, thus quantifying how the HI ears di ffer from ANH ears in their use of acoustic cues. Our analysis shows that HI ears may have poor temporal and/or frequency resolution, because of which they are unable to hear only a few consonants, yet they can hear the rest. We argue that it is necessary to measure this consonant dependence in order to design a more sensitive hearing aid fi tting technique, and no other clinical measure used currently (audiometry, average speech recognition scores, speech in noise tests) is useful in characterizing speech-loss, in HI ears. We measured 46 HI ears with our CV discrimination test using the current hearing aid amplifi cation technique NAL-R; the results show that though NAL-R improves the aver- age score, it degrades a few consonants under certain circumstances. This research also addresses the important issue of cochlear dead regions, which are places along the basilar membrane of the cochlea where the inner hair cells are degenerate. We propose a new method to diagnose dead regions based on comodulation masking release. This project extends our effort to achieve a fundamental insight into the nature of both ANH and HI speech perception, enabling the design of hearing aids that are functionally useful in high ambient noise and that help make audible the sounds that the HI ear could not hear previously, without aff ecting the sounds that they can hear.