In-situ Personalization of Signal Processing Systems


Despite a decade of steady industrial and academic investment in signal processing technology for hearing aids, about 20% of patients remain unhappy with the sound quality of their instruments. As it turns out, we can design a fine hearing aid for a given person in any given situation, but we don’t have enough upfront information to design a great solution for an unknown client who lives in a volatile acoustic world. Therefore, hearing aids must be both adaptable to environmental context and personalizable by the patient. In this lecture I will address design challenges of adaptable hearing aids, and more generally of signal processing systems, which can be in situ personalized by end users. I will discuss a few salient properties of a successful adaptive and personalized signal processing system, namely the brain. In the end, biological information processing provides an inspiring reference for engineers, whether one is designing hearing aids, personalized lighting systems or any other intelligent ambient agent.

Inaugural lecture report