Abstract
Perception of consonants for the House 3M single-channel cochlear implant using a two-channel signal processing model was investigated. The input signal was split into a low frequency band and a high frequency band with the cutoff frequencies of the bands adjusted in nine conditions. The conditions ranged from overlapping bands that yielded a flat spectrum to bands that removed a considerable portion of the midrange frequencies. The stimuli for this study were 60-item vowel-consonant-vowel lists with male and female talkers that were presented through direct electrical connection to the implant wearer. The stimuli were recorded on tape in quiet and noise backgrounds to measure interactions among speaker gender, background, and filtering scheme. The four patients in this study were experienced implant wearers. The results were analyzed both in terms of percent correct consonant and in terms of percent correct consonant category. Three of the filter conditions show better performance than the patient's own processor. Interestingly, the best conditions have large portions of the midfrequency regions attenuated.
Get full access to this article
View all access options for this article.