The goal of this research is to develop a practical, wearable auditory device that acts as an adjunct to speechreading for profoundly hearing- impaired listeners. A limited number of important features that are not available visually will be extracted, and presented in acoustic cues that are matched to the residual auditory channel of the individual user. The primary motivation is not to compress all possible speed information into a hearing system with limited information handling characteristics, but rather to base the amount of information presented on the particular capabilities of that system. One speech characteristic to be extracted is fundamental frequency; this will be presented as a low frequency sine wave/ In addition, the energy of a high frequency consonant band will be vocoded onto a low frequency carrier. These two cues carry both prosodic and segmental information. Prototypes of wearable devices will be constructed, and extensive training on four hearing-impaired subjects will be conducted and monitored. Evaluations of speechreading improvement will be made using the Connected Discourse Tracking measurement technique.
