Use of a sigmoidal-shaped function for noise attenuation in cochlear implants.

A new noise reduction algorithm is proposed for cochlear implants that applies attenuation to the noisy envelopes inversely proportional to the estimated signal-to-noise ratio (SNR) in each channel. The performance of the proposed noise reduction algorithm is evaluated with nine Clarion CII cochlear implant patients using IEEE sentences embedded in multi-talker babble and speech-shaped noise at 0-10 dB SNR. Results indicate that the sigmoidal-shaped weighting function produces significant improvements to speech recognition compared to the subjects' daily strategy. Much of the success of the proposed noise reduction algorithm is attributed to the improved temporal envelope contrast.