neuro-steered hearing device

In this preprint, we propose a multiscale Gaussian mixture model that jointly estimates HMM emission parameters across multiple window lengths, improving HMM-based post-processing for selective auditory attention decoding.

In this article, presented at EUSIPCO2026, we present an alternative post-processing method to hidden Markov models for auditory attention decoding algorithms, based on the Markov switching model.

In this preprint, we show that ignoring temporal dependence can substantially overestimate the performance of auditory attention decoding algorithms. We propose several measures to address these issues, including rIMI, a metric that quantifies the rate of new information gained.

In this preprint, we present a proof of concept of a wireless EEG sensor network using two around-ear EEG nodes for selective auditory attention decoding.

In this article, presented at ICASSP 2026, we present efficient unsupervised training algorithms for EEG-based auditory attention decoding that have no initialization bias.

In this preprint, we show that real-time auditory neurofeedback in a closed-loop selective auditory attention decoding paradigm can degrade decoding accuracy and user experience, rather than improve it, within a single session.

In this article, we present analysis of a conversation tracking paradigm with 2 or 3 conversations during selective auditory attention decoding from EEG.

In this article, we present a novel dataset and accompanying analysis enabling the first direct comparison between scalp, around-ear, and in-ear electroencephalography (EEG) for neural selective auditory attention decoding to speech.

In this article, presented at EUSIPCO 2025, we present a novel method to model the performance curve of EEG-based AAD algorithms starting from a single decision window length.

I gave a 15-min talk at EUSIPCO2025 in Isola delle Femmine, Palermo about our work on performance modeling for correlation-based algorithms for selective auditory attention decoding.