Decoding Speech Envelopes from Electroencephalogram with a Contrastive Pearson Correlation Coefficient Loss

TL;DR

This paper introduces a contrastive Pearson correlation coefficient loss for EEG-based speech envelope reconstruction, enhancing auditory attention decoding accuracy by explicitly maximizing the difference between attended and unattended PCCs.

Contribution

It proposes a novel contrastive PCC loss function that improves speech envelope decoding from EEG signals, outperforming traditional methods focused solely on attended PCC maximization.

Findings

Improves envelope separability and AAD accuracy across datasets.

Reveals dataset- and architecture-dependent failure cases.

Enhances DNN-based speech envelope reconstruction performance.

Abstract

Recent advances in reconstructing speech envelopes from Electroencephalogram (EEG) signals have enabled continuous auditory attention decoding (AAD) in multi-speaker environments. Most Deep Neural Network (DNN)-based envelope reconstruction models are trained to maximize the Pearson correlation coefficients (PCC) between the attended envelope and the reconstructed envelope (attended PCC). While the difference between the attended PCC and the unattended PCC plays an essential role in auditory attention decoding, existing methods often focus on maximizing the attended PCC. We therefore propose a contrastive PCC loss which represents the difference between the attended PCC and the unattended PCC. The proposed approach is evaluated on three public EEG AAD datasets using four DNN architectures. Across many settings, the proposed objective improves envelope separability and AAD accuracy, while also revealing dataset- and architecture-dependent failure cases.