Discussion sessions II

Discussion sessions II

Friday, June 2nd 2023, morning

I. Hardware & Acoustics

d) Model-based hearing devices: Dynamic compression & listening comfort

Session Chair: Stefan Launer (Sonova), Dirk Oetting (Hörzentrum Oldenburg gGmbH)

Short description: The not yet achieved dream of complete hearing impairment compensation with hearing devices implies that the knowledge about all individual impairment factors enables to specifically tailor a hearing aid algorithm to the individual patient. With the advent of OTC and more refined self-fitting methods the importance of deriving an efficient fit for a hearing device (that provides optimum speech perception & listening comfort) with a minimum set of previously measured audiological parameters of the client becomes even more urgent. How far have we come by now? How much can we use past and current trends in modelling hearing loss (like, e.g., loudness models if adapted to binaural loudness summation and hyperacusis, optimum detector, attenuation and distortion component, dead regions, synaptopathy, spectrotemporal receptive fields) to shape current hearing aid processing and fitting schemes? Will the dream of a knowledge- and model-driven compensation scheme with optimum amplification and comfort be replaced by ML methods that might do the job even without knowing why?


II. Psychophysics & Evaluation

e) Technology to support a hybrid model of audiologic care (in between the clinic and OTC).

Session chairs: Heike Heuermann (WSA) & Erin Picou (Vanderbilt University, Nashville Tennessee)

As we approach the age of OTC hearing aids with self-fitting mechanisms there will likely be a hybrid model of HA fitting and audiology support to bridge the gap between today’s clinic based services and the future OTC self-service model. This hybrid approach could be a pay-for-service model and it will require technology to support connections to professional customer service support. Hearing aid manufacturers could benefit from coordinated development and potential standardization of technology and protocols to support such a hybrid model of support. The HADF theme could focus on methods, technology, and potential standards to support this customer-service style method of audiology support.



III. Algorithms & Software

f) Machine learning for hearing aid algorithms (DSP)

Session chairs: Bert de Vries (GN), Jon Barker (University of Sheffield)     

Short description: Even though machine learning (ML) promises to significantly advance the classical hearing aid algorithms and functionality, some of the promises fall short when faced with the reality in hearing instrument development (availability of training data and computational as well as power ressources, real-time and latency constraints…). Where and how can we expect significant improvements by using ML methods and where might the approach fail?



Internetkoordinator (Stand: 29.05.2024)  | 
Zum Seitananfang scrollen Scroll to the top of the page