The overarching goal of this project is a quantitative, model-based survey and optimisation of the acoustic processing quality of hearing aids, as perceived subjectively by affected individuals in real-life situations. A direct cooperative group of model developers, algorithm developers, hardware (or field test prototype) designers and clinicians aims to optimise the interior representation of noises in a deficient hearing system by model-controlled algorithm adjustments ("model in the loop", meaning that model parts are integrated in the algorithm) with regard to three criteria:
- Good consistency with the intact system (model-based reconstruction)
- Good consistency with the interior representation in (linear) improved external SNR ("clean" or "dry" in extreme cases) in situations with external ambient noises (model-based ambient noise cancelling)
- Good consistency with the changed objective (based on cognitive factors, experience/acclimatisation or limitations of hearing aid technology)
The work will be examining the hierarchy of the internal representation on three levels:
- Auditory nerve (meaning the degree of compensation of cochlear and retrocochlear failures)
- Output of the binaural display (meaning the degree of compensation of binaural failures)
- Cognitive/conscious level (for example, corresponding to the "optimum detector" with PEMO), which, among other things, can be modified by cognitive factors (for example, relative weighting of pauses or frequency bands, expectations of sound quality, acclimatisation...).
As part of the model, the quality of reproduction (reconstruction + ambient noise reduction) will be assessed and examined/developed by pitch/quality models (at several model levels) in relevant tests with test persons.
The interplay of model, algorithm, laboratory and field tests and their possible production as commercial hearing aids will be carried over from the previous project and expanded with the aim of objectively evaluating the quality of the hearing aids.