EEG-gebaseerde auditieve aandachtsdecodering voor neuraal gestuurde hoorapparaten ter verbetering van spraakverstaanbaarheid

People with hearing impairment often have difficulties understanding speech in noisy environments, leading to social isolation and generally decreased quality of life. Hearing aids and cochlear implants partly solve this problem through signal processing algorithms that enhance the target speaker and suppress other noise sources. However, in a scenario where multiple speakers talk simultaneously, a fundamental problem appears: how does an algorithm decide which speaker the listener is actually attending to? This can be solved by decoding the attended speaker from measured brain signals (using EEG), and accordingly enhancing the attended speaker. The performance of such auditory attention decoding (AAD) algorithms needs to be further evaluated in realistic scenarios for future application in neuro-steered hearing aids. In this PhD, we investigate (1) how speaker- and listener-related factors influence AAD performance, (2) whether decoding auditory attention at the conversation level, rather than at the level of individual speakers, provides a more realistic and feasible paradigm in complex acoustic scenarios, and (3) whether AAD-based gain control improves speech intelligibility compared to state-of-the-art hearing aids.