EEG alpha power as a measure of listening effort reduction in adverse conditions

Noise levels in everyday environments are typically low enough to comprehend speech, but over extended periods of time, listeners may experience tiredness and even fatigue. We measured the neurophysiological compensation required to separate speech from noise –listening effort –using electroencephalography, for normal and AdaptDRC enhanced speech.

AdaptDRC is a noise-dependent near-end-listening-enhancement algorithm which significantly improves the intelligibilityof speech in noise and reduces subjectively rated listening effort (Rennies et al., 2018).Participants (N=27) completed a listening effort and intelligibility task using a categorical listening effort scale, in which sentences were presented at five SNRsand in two types of noise.

Subjective listening effort was significantly lower, and speech intelligibility was significantly higher for AdaptDRC speech than unprocessed speech (p<.001). Analyses of the neurophysiological data show that there is a non-linear relationship between SNR and spectral alpha power with a peak between 0-5dBSNR, and that alpha power is sensitive to noise type and speech processing.

These findings provide insight into the neurophysiological correlates of listening effort and aid the development of an objective measure of cognitive load for speech in noise.