John W. Hawks
Kent State Univ., School of Speech Pathology and Audiology, Kent, OH 44242
Marios S. Fourakis
Ohio State Univ., Columbus, OH 43210
Margaret W. Skinner
Timothy A. Holden
L. K. Holden
Washington Univ. School of Medicine, St. Louis, MO 63110
The aim of the experiment reported here was to examine the responses to synthetic vowels of the processor of the Nucleus Cochlear Implant System, using the MPEAK and SPEAK speech coding strategies. In previous work with natural vowels (Skinner et al., submitted to Ear Hear.), it was found that second formant information was better transmitted using the SPEAK strategy, while first formant information was better transmitted using the MPEAK strategy. Examination of the output of the SPEAK processor showed activation of multiple adjacent electrodes resulting in less distinct spectral cues than with MPEAK. This prompted the present investigation, which uses synthetic vowels with varying formant bandwidths. Four synthetic vowels, which had been identified consistently by normal hearing subjects as one of [(small capital eye), (cursive beta), U, (inverted vee)] were used as anchor points for the creation of continua in which the bandwidths of the first and second formants were systematically and independently decreased in steps of 10% to a minimum of 50% of the formant bandwidths of the anchor stimuli. Results from the processors and identifications from subjects using the two processing stategies will be presented and compared. [Work supported by NIH.]