Improving speech intelligibility for bilateral cochlear implant users using Weiner filters and its impact on cognitive load

While cochlear implants have improved speech understanding for patients who are profoundly deaf, performance in noisy environments is still far from that of normal hearing listeners. It has been common to suggest that performance improvements can be obtained using speech enhancement/noise reduction algorithms. These algorithms are often evaluated using theoretical measures only, or via speech intelligibility tests that measure percent correct scores. However, many of these algorithms are also known to introduce undesirable distortions which may negate the theoretical signal-to-noise ratio benefits because of an increase in cognitive load required to understand the enhanced, but distorted, speech signal. In this work, we evaluate a simple noise reduction algorithm based on Wiener filtering principles with cochlear implant users, and show that improvements in speech understanding performance can be seen with percent correct scores overall, but at the cost of increased cognitive load in some listeners.