niels_round
Niels Henrik Pontoppidan

Research Area Manager

> Learn more

mail@eriksholm.com

We gather data on the programme preferences of test subjects in their everyday life. This gives valuable information about their needs in specific situations.

Connecting phone and hearing aid

We dream of hearing aids that adjust itself completely according to the needs of the user. One step in that direction is to gather user data from 10 test subjects. This is the first time anyone continuously over months have collected data capturing both how the devices perceive the changing sound environments, as well the corresponding program, settings, and volume preferences of users. Running the tests over longer time and in the test subjects’ real environment rather than in a lab makes it possible to learn individual preferences that may change throughout the day depending on not only the context but also the intents or cognitive state of the user.

The experimental setup was based on a smartphone app which collects data from the hearing aid via bluetooth. The app enables the user to choose between four programs providing alternative amplification settings. Dependent on the context, the user may prefer to increase clearness to facilitate speech intelligibility, enhance soft sounds to intensify the perceived loudness, or conversely reduce the background noise. That is, some programmes may provide more spatial cues enabling the user to shift his attention to separate sounds, while others improve the signal to noise ratio by attenuating ambient sounds. Thus, the user may change programmes throughout the day according to what situation they are in.

The hearing aids used are from our international research project EVOTION, which you can read more about here: http://h2020evotion.eu/

soundscapes_fig2

Results

The illustration shows an example of how different soundscapes and contexts trigger different preferences, resulting in a programme selected by the user. P1, P2 etc. are the different programmes, while the shades of green indicate varying amounts of speech and noise in the soundscape acoustical environments.

Initial trials indicate that the test subjects preferred to change between alternative programmes rather than staying in one. Programmes offering a natural omnidirectional focus were combined with contrasting settings, offering either more spatial cues or conversely removal of noise. Empowering users to choose between contrasting programmes may increase personalization beyond the “one size fits all” average settings many users are currently given. Even when analysing only the usage behaviors of 10 test subjects it is evident that people are not average but have very different programme preferences dependent on the context.