Auditory attention decoding (AAD) through a brain-computer interface has had a flowering of developments since it was first introduced by Mesgarani and Chang (2012) using electrocorticograph recordings. AAD has been pursued for its potential application to hearing-aid design in which an attention-guided algorithm selects, from multiple competing acoustic sources, which should be enhanced for the listener and which should be suppressed. Traditionally, researchers have separated the AAD problem into two stages: reconstruction of a representation of the attended audio from neural signals, followed by determining the similarity between the candidate audio streams and the reconstruction. Here, we compare the traditional two-stage approach with a novel neural-network architecture that subsumes the explicit similarity step. We compare this new architecture against linear and non-linear (neural-network) baselines using both wet and dry electroencephalogram (EEG) systems. Our results indicate that the new architecture outperforms the baseline linear stimulus-reconstruction method, improving decoding accuracy from 66% to 81% using wet EEG and from 59% to 87% for dry EEG. Also of note was the finding that the dry EEG system can deliver comparable or even better results than the wet, despite the latter having one third as many EEG channels as the former. The 11-subject, wet-electrode AAD dataset for two competing, co-located talkers, the 11-subject, dry-electrode AAD dataset, and our software are available for further validation, experimentation, and modification.

Objectives: Hearing-protection devices (HPDs) are made available, and often are required, for industrial use as well as military training exercises and operational duties. However, these devices often are disliked, and consequently not worn, in part because they compromise situational awareness through reduced sound detection and localization performance as well as degraded speech intelligibility. In this study, we carried out a series of tests, involving normal-hearing subjects and multiple background-noise conditions, designed to evaluate the performance of four HPDs in terms of their modifications of auditory-detection thresholds, sound-localization accuracy, and speech intelligibility. In addition, we assessed their impact on listening effort to understand how the additional effort required to perceive and process auditory signals while wearing an HPD reduces available cognitive resources for other tasks. Design: Thirteen normal-hearing subjects participated in a protocol, which included auditory tasks designed to measure detection and localization performance, speech intelligibility, and cognitive load. Each participant repeated the battery of tests with unoccluded ears and four hearing protectors, two active (electronic) and two passive. The tasks were performed both in quiet and in background noise. Results: Our findings indicate that, in variable degrees, all of the tested HPDs induce performance degradation on most of the conducted tasks as compared to the open ear. Of particular note in this study is the finding of increased cognitive load or listening effort, as measured by visual reaction time, for some hearing protectors during a dual-task, which added working-memory demands to the speech-intelligibility task. Conclusions: These results indicate that situational awareness can vary greatly across the spectrum of HPDs, and that listening effort is another aspect of performance that should be considered in future studies. The increased listening effort induced by hearing protectors may lead to earlier cognitive fatigue in noisy environments. Further study is required to characterize how auditory performance is limited by the combination of hearing impairment and the use of HPDs, and how the effects of such limitations can be linked to safe and effective use of hearing protection to maximize job performance.

No abstract

Auditory attention decoding (AAD) through a brain-computer interface has had a flowering of developments since it was first introduced by Mesgarani and Chang (2012) using electrocorticograph recordings. AAD has been pursued for its potential application to hearing-aid design in which an attention-guided algorithm selects, from multiple competing acoustic sources, which should be enhanced for the listener and which should be suppressed. Traditionally, researchers have separated the AAD problem into two stages: reconstruction of a representation of the attended audio from neural signals, followed by determining the similarity between the candidate audio streams and the reconstruction. In this work, we compare the traditional two-stage approach with a novel neural-network architecture that subsumes the explicit similarity step. We compare this new architecture against linear and non-linear (neural-network) baselines using both wet and dry electroencephalogram (EEG) systems. Our results indicate that the wet and dry systems can deliver comparable results despite the latter having one third as many EEG channels as the former, and that the new architecture outperforms the baseline stimulus-reconstruction methods for both EEG modalities. The 14-subject, wet-electrode AAD dataset for two competing, co-located talkers, the 11-subject, dry-electrode AAD dataset, and our software are available to download for further validation, experimentation, and modification.

No abstract