The work proposed here is focused on the spatial aspects of understanding speech in complex environments, most importantly environments with multiple speech sources. This would be a continuation of work in our lab that has been focused on binaural hearing, with increasing attention to the abilities of hearing-impaired listeners to understand speech in these complex acoustic environments. These environments, which often include simultaneous speakers, surface re ections, and other acoustical disturbances, are particularly frustrating for hearing-impaired listeners because they are typical of group social interactions and because these diculties are often even worse for users of hearing aids or cochlear implants. The proposed studies incorporate multiple approaches to understanding how this situation is processed in both normal and impaired auditory systems, with the goal of both scientific understanding and the development of strategies for the improvement of hearing aids and cochlear implants. Modern hearing aids and cochlear implants are increasingly applied to both ears with binaurally coordinated function and provide opportunities for improving performance in these complex environments. Our proposed study integrates three categories of work: In the first category, we will continue to develop explicit, quantitative models to help us understand (and make quantitative predictions for) speech intelligibility performance of human listeners in the presence of other speech sources. This modeling can be divided into two sub-categories: (1) basic signal-processing modeling in the spirit of traditional psychophysical models and (2) physiological-mechanism modeling that explicitly includes basic representations of available neurophysiological data. Models in the first sub-category are exemplified by the short-time equalization-cancellation (STEC) model (Wan et al., 2014) that we developed and applied to available data for speech masked by other speech sources. Ongoing Work in the second sub-category relates cortical-level responses to the processing of spatially distributed targets and maskers (Dong et al., 2015a,b) and explores the implications of this processing. In the second category, we will measure abilities of normal-hearing and hearing-impaired listeners in a variety of behavioral tasks, including both speech intelligibility thresholds in environments containing interfering speech sounds and basic sensitivity tasks such as hearing thresholds, sensitivity to interaural differences, and temporal resolution. These measurements are motivated by questions raised by our previous experimental results and by the results of the modeling predictions. The basic plan is to be able to apply our models to individual listeners, with parameters chosen to describe the full span of the various experiments and their possible relationship. Finally, in the third category, we plan to extend our modeling and experiments to listeners wearing hearing aids and cochlear implants with the goal of characterizing the effects of the current processing algorithms and testing new algorithms that are inspired by the results of the experiments and modeling described above. The work in this category will be done in collaboration with colleagues in industry. Current interactions include work with Dr. Abhijit Kulkarni from Advanced Bionics and several colleagues from the hearing aid industry. We are in contact with these people through ongoing projects related to the acoustic characterization of complex environments and to the processing of current cochlear implants.
Relevance to the public health The work proposed here addresses, with an encompassing perspective, the widespread problem of listening to a speech signal of interest, in an environment that contains other sound sources, especially other speech sources. This is a particularly difficult challenge and a prevalent complaint for listeners with impaired hearing, even when they are using hearing aids or cochlear implants. The proposed study integrates modeling of both perceptual and neurophysiological data, psychophysical experimentation, and design and testing of binaural processing algorithms that could be used as pre-processors for hearing aids and cochlear implants.
Showing the most recent 10 out of 60 publications
