Multivariate timeseries analysis for EEG measures of attended speech

Petteri Hyvärinen

The capability to overcome the cocktail party problem degrades with hearing impairment; focusing auditory attention at will to a sound source while ignoring distracting sounds becomes harder. This ability cannot be regained by simply providing amplification with a hearing aid to compensate for the hearing loss. Instead, more elaborate signal processing has to be applied in order to acoustically target a single sound source. While it is for example already possible to have a hearing aid tune in to a specific spatial location, the challenge is in knowing where exactly to point the focus. Automatically steering the hearing aid towards an auditory target of interest requires real-time information about the internal intentions and mental efforts of the user. Petteri Hyvärinen will be working with the COCOHA project to find out how electrophysiological measures of brain activity (EEG) could be used for choosing and boosting only those sounds from the surrounding acoustic environment that are relevant to the user.


Petteri Hyvärinen
DTU Health Tech