Speech recognition with a CIS strategy for the ineraid multichannel cochlear implant.

Abstract

Wilson et al. proposed a new sound-processing strategy for multichannel cochlear implants, the Continuous Interleaved Sampling (CIS) strategy. Their study was performed on seven American patients, selected for their excellent performance with the Ineraid multichannel cochlear implant, and involved refined testing of several parameter modifications of the CIS strategy during a 1-week period. At the end of the week, the CIS strategy produced large improvements in speech-recognition tests for all subjects. To evaluate the generality of this promising result, the goal of this study was to assess whether similar improvements of performance could be observed in a typical population of Ineraid users and implemented as a clinical protocol. Therefore we designed one unique, predetermined CIS processor that could be temporarily fitted to the patients in < 2 h, and we evaluated speech recognition with consonant-and vowel-identification tests in a group of patients with performances ranging from star to almost chance levels and speaking six different native languages. Scores of vowel and consonant identifications obtained with this predetermined CIS processor and with the standard processor of the Ineraid system were compared in 15 Ineraid users. Fourteen of 15 patients had significantly better scores of consonant identification with the new CIS strategy. The group mean scores of vowel identification with either strategy were not statistically different. In agreement with these observations, most patients immediately reported that the CIS strategy sounded subjectively "more clear" for real-time speech recognition. It is now possible to implement a CIS speech-coding strategy as a standard clinical procedure to improve speech-recognition performances of all Ineraid users.