PhD defence J.G. (Jan) Dingemanse

A cochlear implant (CI) is a surgically implanted device that converts sound into electrical signals that stimulate the auditory nerve in the inner ear. It is a valuable treatment for people with severe to profound sensorineural hearing loss. In post-lingually deafened adults a CI improves auditory functioning and speech perception in a quiet environment, although maximum speech understanding, expressed as percentage of correctly recognized speech elements, can vary greatly from person to person. In daily life, there are often background sounds that hinder speech perception. Speech perception in background noise is a challenge for CI recipients or is not even possible. This limited speech perception originates from the fact that the auditory information that passes the CI and the auditory nerve is less detailed than in normal-hearing listeners. As a consequence, top-down processing of the incoming auditory signal is required to recognize words by filling in the gaps in the incoming auditory signal. Linguistic and cognitive processes are involved in this top-down processing. Listening to limited auditory information may be effortful. First, this thesis investigated how speech perception of CI users is influenced by background noise, using three outcome measures: speech recognition in noise, noise tolerance and listening effort. A speech test consisting of everyday sentences was used to determine the scores of a CI user on these outcome measures. Contemporary sound processors of CI systems incorporate a single-microphone noise reduction algorithm with the aim of improving the speech-in-noise perception. A second topic of this thesis was therefore the evaluation of the effect of clinically available single-microphone noise reduction algorithms on speech perception in noise using the three above-mentioned outcome measures. Third, the relative influence of bottom-up auditory speech characteristics in the incoming signal and cognitive top-down processing on speech perception in noise was investigated. The relationship of the mentioned outcome measures with the amount of bottom-up information in the incoming signal was studied with a spectral resolution test. The effect of the top-down processing on the outcome measures was investigated with a test for working memory capacity. Furthermore, a model was used that models how contextual information present within a sentence is used to be able to correctly understand speech elements that are not properly recognized in the bottom-up information. The fourth element in this thesis concerns the question of how the speech perception in noise can best be investigated with the existing Dutch speech material that consists of everyday sentences. The measurement methods and various measurement properties of the outcome measures, when used in the group of CI users, were investigated.

Due to the lockdown, the PhD defences will not take place publicly in the usual way. A live stream link has been provided to candidate. The ceremony will begin exactly at 13:00.