A new feature driven cochlear implant speech processing strategy

Our study focuses on the development of a new feature driven speech-processing strategy for cochlear implant system. In each cycle of stimulation of the cochlea, an electrode, corresponding to second formant frequency was chosen among 14 basilar electrodes. On the base of voiced/unvoiced decision of the speech, an electrode corresponding to first formant frequency was selected for stimulation among 6 apex electrodes. Additionally 4-6 electrodes were chosen for stimulation by maxima energy criteria. Speech intelligibility tests on multi syllable Japanese words within normal hearing listeners by acoustic simulation were provided to evaluate performance of the proposed strategy.