Assessing and reducing listening effort of listening to speech in adverse conditions

Background: Normally hearing listeners successfully compensate for speech distortions in everyday environments, but can become fatigued as a result. AdaptDRC is a near-end-listening-enhancement algorithm that alters speech signals for playback, dependent on environmental noise, for improved intelligibility (Schepker et al., 2013). Aim: In this electroencephalography (EEG) study I measured neurophysiological correlates of listening effort (LE) by comparing unprocessed speech to AdaptDRC-processed speech. Method: I recorded EEG while normally hearing participants (N=27) listened to unprocessed or AdaptDRC-processed sentences in noise, then rated the subjective listening effort. I also measured speech intelligibility, hearing and cognitive abilities. Results: For intelligible speech, subjective LE decreases with increasing SNR and is lower for AdaptDRC speech than unprocessed speech. Spectral alpha power (8-12Hz) analyses suggest that peak cognitive effort occurs at 0dB SNR. Spectral alpha also increases with task duration, indicating an association with fatigue. Conclusions: This experiment provides insight into the neurophysiological correlates of effortful listening in adverse conditions, and the benefits of near-end-listening-enhancement technology.