Optimizing Frequency Channels in Cochlear Implants

Business/Education/Speech and Hearing Science (The Ohio State University Denman Undergraduate Research Forum)Cochlear implants (CIs) are devices used by individuals with hearing loss to improve communication through the use of an electrode array that directly stimulates the auditory nerve. Existing signal processing strategies utilize a logarithmic frequency-to-electrode allocation, mimicking the representation of frequencies along the basilar membrane (high frequencies at the base and low frequencies at the apex). These strategies support some degree of open-set speech recognition for CI users; however, average speech recognition remains well below what normal-hearing adults are capable of. To enhance speech recognition in adult CI users, this study examined one promising alternative to the standard logarithmic frequency-to-electrode allocation maps. The frequency-to-electrode allocation maps were modified to provide more refined representations of the first two (and most important) vowel formant frequencies (energy peaks in vowels that are critical to speech perception). Twelve participants were tested using two different CI maps: one based on existing clinical frequency-to-electrode allocation strategies (Standard) and one designed to improve the resolution of the first two formants, which should especially enhance vowel recognition (Speech). Alternating between these maps, participants listened to and repeated three kinds of stimulus materials: (1) highly meaningful five-word sentences, (2) syntactically correct but not meaningful four-word sentences, and (3) phonetically balanced consonant-vowel-consonant words in isolation. Analyses revealed that some participants benefitted from the Speech strategy. Moreover, an improvement in vowel recognition in words strongly predicted an improvement in recognition of words in sentences. These findings suggest that optimizing the representation of the first two formants enhances speech recognition for CI users. Future efforts should focus on better representing this speech-specific information in modern-day signal processing strategies.Academic Major: Speech and Hearing Scienc