The most common complaint associated with hearing loss is understanding speech, especially in adverse listening situations. While some of this difficulty directly relates to loss in hearing sensitivity, the extent of the speech-processing deficits almost always exceeds expectations based on detection thresholds as represented in the audiogram. The primary thesis of the proposed research is that these extended speech deficits relate to difficulty maintaining proper speech-signal coherence. When speech is processed through the frequency- selective channels of the human auditory system, the information can be represented in terms of fluctuation of the temporal envelope (E) and temporal fine structure (TFS) as frequency modulation (FM). The specific hypotheses of the proposed study are that for speech, (1) the role of FM relates to maintaining speech coherence to enhance streaming across acoustic elements and aid segregation of the signal from competing sources, and (2) that deficits in FM processing are in part directly responsible for speech difficulties associated with hearing loss and aging. These hypotheses will be tested in experiments of speech perception manipulating stimulus TFS cues and in psychoacoustic discrimination measures with signal parameters chosen to approximate speech characteristics. Incorporating stimulus uncertainty-a fundamental aspect of natural speech-all proposed psychoacoustic measures determine ability to discriminate among low-rate patterns of stochastic FM. Proposed experiments determine the basis of the ability to discriminate among stochastic patterns of low-rate FM, distinguishing contribution of E cues derived from TFS variation. Time-dependent fine-structure filtering and temporal interruption are used to separate these effects in measures of speech intelligibility. Results will be used to evaluate potential factors underlying the speech benefit obtained when combining acoustic and electrical stimulation. Using condition parameter values determined by results from initial work, proposed research will measure the ability of both elderly and hearing-impaired human listeners to discriminate stochastic patterns of FM relevant to speech perception. FM discrimination and speech intelligibility will be measured for four different subject groups: two with normal hearing grouped as either young or elderly, and two with a mild-to-moderate hearing loss, again either young or elderly. To evaluate the potential for FM discrimination contributing to signal segregation, conditions will be run in quiet and in the presence of a variety of maskers that represent different types of interference that listeners confront when attending to speech. Based on results, clinically feasible psychoacoustic measures will be developed to indicate the extent of the speech-processing deficits associated with hearing loss and aging, and aid in the early diagnosis of cochlear hearing loss. Results obtained from clinical populations should suggest signal-processing strategies for use in prosthetic devices that may help remediate speech deficits.
The project will provide a basis for understanding the manner by which low-rate frequency modulation (FM) contributes to speech perception in adverse listening environments, and the extent to which the speech-processing deficits of the elderly and listeners with hearing loss relate to impaired processing of this modulation. The long-term goal of this work is to develop a clinically feasible psychoacoustic measure of FM processing ability to assess, diagnosis, and aid in the remediation of difficulties with speech perception.
