SHINE: Sequential Hierarchical Integration Network for EEG and MEG
This work addresses speech decoding from brain signals for cognitive neuroscience, but it is incremental as it builds on existing methods and competition benchmarks.
The paper tackled the problem of detecting speech from MEG signals in the LibriBrain Competition 2025, achieving F1-macro scores of 0.9155 and 0.9184 on test sets using ensemble methods with their proposed SHINE network.
How natural speech is represented in the brain constitutes a major challenge for cognitive neuroscience, with cortical envelope-following responses playing a central role in speech decoding. This paper presents our approach to the Speech Detection task in the LibriBrain Competition 2025, utilizing over 50 hours of magnetoencephalography (MEG) signals from a single participant listening to LibriVox audiobooks. We introduce the proposed Sequential Hierarchical Integration Network for EEG and MEG (SHINE) to reconstruct the binary speech-silence sequences from MEG signals. In the Extended Track, we further incorporated auxiliary reconstructions of speech envelopes and Mel spectrograms to enhance training. Ensemble methods combining SHINE with baselines (BrainMagic, AWavNet, ConvConcatNet) achieved F1-macro scores of 0.9155 (Standard Track) and 0.9184 (Extended Track) on the leaderboard test set.