Unattended testing
The main focus of Fiorio’s research was training the algorithms in hearing aids. Training was validated using a combination of objective metrics and, where possible, by listening to audio clips output by the software. Unfortunately, we were unable to test the software for personally worn hearing aids.
“It’s difficult to test proposed algorithms on a device if you’re not part of a hearing aid company. This isn’t necessarily a good thing for researchers based in academia, but there is a gap in testing devices focused on hearing-based research. I only know of one startup that is manufacturing hearing aids on a small scale. Hearing aid companies often have their own testing equipment, but its use is mostly done in-house.”
hearing aids of the future
Given the increasing efficiency of computer chips and the power of machine learning and deep learning, it is likely that hearing aids will rely more on machine learning approaches in the future.
“These algorithms rely on efficient hardware implementations to offset computational costs, which is also a bottleneck,” Fiorio points out. “A more efficient implementation means less battery consumption.”
Fiorio believes his research in training hearing aids to recognize sounds without labeling them will be crucial. “On-the-fly learning of hearing aids during operation is the ultimate goal. In this way, any user will have access to a purpose-built device that adapts to their needs based on their personal preferences and surrounding environment. The first step to achieving this is to forget about labels and focus on the audio content itself.”
Mr. Fiorio will continue his work to revolutionize the future of hearing aids in his new role as research scientist at GN Hearing, part of the Danish company GN Store Nord.
