Comparisons of Spectra 22 Performance in Noise with and without an Additional Noise Reduction Preprocessor

Abstract

This paper describes a two-microphone, software-programmable noise-reduction de vice that was interfaced to the Nucleus Spec tra 22 speech processor to act as a front-end noise-reduction preprocessor. The develop ment for the portable processor and the noise-reduction algorithm, more formally known as beamforming, was originally moti vated by complaints from individuals who use hearing aids. These individuals com plain about a deterioration in performance with increasing levels of background noise. Since individuals who use cochlear implants have similar complaints, it was a natural ex tension to pose the question: "What benefit, if any, would the beamforming algorithm provide to individuals who use cochlear im plants?" To arrive at an answer, the audio in terface to the noise reduction device was modified (to make it compatible to the Nu cleus Spectra 22 speech processor), and a set of precursory subject experiments were per formed. 1 The precursory studies were spe cific to the Nucleus 22 Channel Cochlear Implant and the Spectra 22 speech proces sor, both manufactured by Cochlear Corpo ration and Cochlear Limited. The noise-reduction device used in the precursory studies is known as the Alpha II and is devel oped by AudioLogic Inc. Eleven English-speaking subjects par ticipated in a series of sessions during which they were tested with their own Spectra 22 speech processor and with the Alpha II beam-forming algorithm acting to preprocess the input data to their device. The beamform ing