Patrick Zurek - US grants

The ultimate goal of this research is the development of sensory aids that sample the acoustic environment at more than one point in space (multimicrophone aids) to help hearing-impaired subjects function more effectively in complex environments containing interference and reverberation. The more immediate goal is the creation of multimicrophone systems that enhance monaural speech reception in such environments. This research, combined with research on binaural interaction in impaired listeners, will provide solid background for the development of multimicrophone aids to assist impaired listeners who have significant hearing in either one or two ears. To enhance monaural speech reception, we envision a microphone array that resolves the incoming signals into simultaneous directional channels, followed by a directional coding operation that transforms these signals in such a way that resolution is preserved at the perceptual level after the signals are summed for presentation to a single ear. Proposed work on the microphone array focuses on the creation of a single directional channel and reduction of interference from sources directionally distinct from the target source. The interference reduction schemes to be studied draw heavily from previous work on antenna theory and binaural hearing and include both fixed and adaptive, and linear and nonlinear, signal- processing algorithms. Based on positive results already obtained in this program, a subset of these schemes will be implemented in wearable devices for field testing. Proposed work on directional coding includes study of the coding effectiveness of differential filtering, variations in voice quality, and extrapolations of these variations by signal processing. Also, for use as a comparative reference, the coding effectiveness of variations in binaural lateralization (the natural coding method) will be examined. Many of our results on the reduction of interference will be applicable to cochlear-implant aids and sensory-substitution aids as well as conventional acoustic aids.

The ultimate goal of this research is the development of sensory aids that sample the acoustic environment at more than one point in space (multimicrophone aids) to help people with hearing impairments function more effectively in complex environments containing interference and reverberation. In particular, the proposed work is aimed at the development of multimicrophone systems that enhance the reception of a target speech source by reducing interference from sources directionally distinct from that of the target. The interference-reduction schemes to be studied are outgrowths of past work on both fixed directional, and adaptive noise-cancelling, microphone arrays. Much of the proposed work involves further algorithmic development of these interference-reduction systems, leading to robust, maximally-directive head-worn fixed arrays and to improved performance of adaptive arrays in reverberant and dynamically-changing environments. The multimicrophone systems under development are also being designed to allow user control of the 'look' direction, as well s to convey information about the background, i.e., the presence and location of non- target sources. Proposed methods for incorporating this background information include beam scanning and source selection. binaural, as well as monaural, output systems are under study for improved displays of target and background signals. Past evaluations of the physical performance of these systems, and prospects for further improvement, are promising. Little is known, however, about the use and benefit of directional and source-cancelling systems in everyday environments. Therefore, the proposed work is aimed at adapting algorithms for implementation in flexible real-time processors operating in conjunction with microphone arrays. These prototype hearing aids will be worn and evaluated in realistic acoustic conditions that incorporate reverberation and multiple interference sources, with natural temporal variations. Both speech intelligibility and the ability to monitor the acoustic environment will be assessed. Progress in this work will be directly applicable to similar problems of interference reduction for cochlear implants, sensory-substitution aids, and speech-input devices for the handicapped.

The objective of this contract is to design, develop, and evaluate in laboratory-based and field-trial studies, innovative speech processing strategies for hearing aids. The focus will be on new and creative approaches, although it may include the evaluation of recently developed strategies using existing technologies such as multiple microphone arrays, automatic signal processing technologies (e.g., fixed-frequency or level- dependent frequency response, compression amplification), and programmable hearing aids employing digital processing. Wearable speech processors will be developed for use in field trials.