A probabilistic modeling approach to hearing loss compensation
Abstract
Hearing loss is a serious and prevalent condition that is characterized by a frequency-dependent loss of sensitivity for acoustic stimuli. As a result, a tone that is audible for a normal-hearing person might not be audible for a hearing-impaired patient. The goal of a hearing aid device is to restore audibility by amplification and compressing the dynamic range of acoustic inputs to the remaining audible range of the patient. In practice, current hearing aids apply frequency- and intensity-dependent gains that aim to restore normal audibility levels for the impaired listener. The hearing aid algorithm design problem is a difficult engineering issue with many trade-offs. Each patient has her own auditory loss profile and individual preferences for processed audio signals. Yet, we cannot afford to spend intensive tuning sessions with each patient. As a result, there is a need for automating algorithm design iterations based on in-situ collected patient feedback.
Thijs van de Laar
Assistant professor
I am an assisant professor at BIASlab, where I work on artificial agents that learn to control themselves in uncertain environments. I take inspiration from physics and neuroscience, and develop theory and (software) tools that allow for efficient, real-time interaction.
