Binaural-cue weighting in sound localization with open-fit hearing aids and in simulated reverberation

Abstract

Interaural time (ITD) and level differences (ILD) are susceptible to distortion by multipath acoustics due to reverberation, echoes, and potentially with open-fit, behind-the-ear (BTE) hearing aids, which pose an additional delay between acoustic and processed sound (~2-5ms). Here, ten young, normal hearing listeners localized broadband and narrowband (500, 4000, 500+4000 Hz) noise. Listeners were fit with linear amplification and evaluated in three aided conditions: unaided, open-fit, and occluded BTE coupling. Sounds were presented from 64 loudspeakers covering 360º azimuth in an anechoic chamber. Twenty-three target locations were evaluated, spanning ±61º. Sounds were presented in three virtual acoustic environments: anechoic, single virtual wall, and simulated 10m X 10m room. The virtual wall was 80% reflective (α=0.2) and oriented parallel to the listener’s forward gaze, at a distance of 5m to the right. The room condition simulated four virtual walls (α=0.5), 5m to the left and right, 6.67m in front, and 3.33m behind the listening position. Performance was assessed across conditions by measuring localization gain (slope), localization variance, localization error, and front-back confusion rate. Results showed greater variability in the simulated room than anechoic conditions, particularly for aided conditions (occluded > open > unaided). Aided listening compressed localization gain at 4000 Hz and expanded gain at 500 Hz. Errors were generally reduced when 500 and 4000 Hz were played simultaneously. Additionally, in-ear acoustic recordings of broadband noise were made in all aided and room conditions. Binaural cross-correlation and intensity-difference calculation were used to estimate frequency-specific ITD and ILD, respectively, from the recordings. Consistent with previous research, ITD became erratic and ILD diminished in simulated rooms compared to anechoic. Effects of hearing aid venting were less noticeable. Finally, results were quantified in the form of binaural-cue weighting (ITD/ILD “trading ratio”) on the basis of measured interaural cue values extracted from in-ear recordings obtained for each listener and condition.