Advancing Auditory EEG Decoding: The ICASSP Grand Challenge, the Imperative for Open Datasets

Jonas Vanthornhout, KU Leuven

Abstract

Born out of CogHear 2022, the ICASSP "Auditory EEG Decoding Challenge" exemplifies innovative problem-solving in the neural tracking field. Participants had access to 157 hours of EEG data from 85 subjects and were tasked with either a match-mismatch task or a speech envelope reconstruction.

This presentation will overview the diverse methodologies, shared themes, novel strategies, and challenges encountered. It will also address the persistent barriers that slow the field's progression.

The emphasis will be on the significance of open datasets, their creation, and their vital role in driving our field forward. By leveraging open datasets, as demonstrated in the challenge, we aim to catalyze innovation and collaboration in neural tracking and progress towards decoding complex auditory neural responses.