Perceptual evaluation of noise reduction in hearing aids

Difficulty to understand speech in noisy situations is the number-one complaint of hearing aid users. Hearing aid manufacturers take measures against the problem of speech in noise by implementing signal-processing algorithms that should reduce background noise. The most widely applied measure against noise is single-microphone noise reduction, which has the task to estimate from a single input signal whether it does contain noise or not and to reduce the hearing aid gain accordingly, without affecting speech if present.
Given that there are about 650 000 hearing aids in use in the Netherlands of which the vast majority contains single-microphone noise reduction, it is remarkable that there is only little knowledge on the implementation and effects of noise reduction. Clinicians who are responsible for prescribing and fitting the hearing aids have no insight in the consequences of activating noise reduction in a hearing aid. Thus, for thousands of hearing aid users it is not sure whether their hearing aid is chosen and fitted optimally to compensate for the wide-spread problem of reduced speech perception in noise.
This thesis describes several studies that were conducted to learn more about the effects of different single-microphone noise-reduction algorithms on perceptual outcomes, such as speech intelligibility, listening effort, and personal preference. The major part of the thesis comprises studies investigating noise-reduction implementations from commercial hearing aids. In two additional studies state-of-the-art noise-reduction algorithms from literature were used to deepen our knowledge on some specific aspects of noise reduction.