Design and Performance of the Low-Power Noise Reduction Algorithm of the Med-El Sonnet 2™ Cochlear Implant Audio Processor

Speech perception in noise is still a difficult task for cochlear implant (CI) users. They need a significantly higher signal-to-noise ratio (SNR) compared to normal hearing individuals for a comparable speech perception. Furthermore, the listening effort in speech-in-noise scenarios is much higher for CI users leading to listener fatigue. Noise reduction (NR) methods using one or more microphones are thus important means to push or even enable speech perception for CI users in noisy environments. In this paper the performance of a very-low-complexity single channel NR algorithm implemented in commercially available CI audio processors (Sonnet 2™ and Rondo 3™ from Med-El GesmbH) is disussed. Even though only a maximum of twelve subbands are used, thus the frequency resolution is very low, a significant improvement of speech perception in noise is achieved, as well as a significantly lower listening effort is reported, as subjective results from adult CI users demonstrate.