01whole.pdf (2.24 MB)
Assessing spatial hearing using laboratory-based real-world environments
thesisposted on 2022-03-28, 11:42 authored by Tobias Weller
Spatial hearing is a crucial aspect of the way we perceive our environment. It helps us to orientate ourselves within an acoustic scene, follow a conversation in a noisy background, avoid potential dangers or simply enjoy the aesthetics of sound, e.g. in music. Hearing impaired listeners often report diffculties with spatial hearing, and current hearing devices frequently do not improve those deficits but rather disturb the cues that the auditory system uses to extract spatial information from the sound reaching our ears. To adequately predict the effect of digital hearing devices on the spatial perception of sound, tests are needed that reliably reproduce acoustic environments that listeners encounter in their daily life. This thesis investigated some of the challenges that are met when spatial hearing is assessed in laboratory-based environments which aim to recreate such realistic scenarios. The first three studies investigated the factors affecting audibility in a reverberant multitalker background. First, masked thresholds for the detection of a short speech stimulus were measured in normal hearing (NH) and hearing impaired (HI) listeners. Then, the effect of uncertainty about the location of the target sound source on those thresholds was examined. The results showed that audibility in the analyzed environment depended heavily on the location of the target source. Moreover, masked thresholds were generally higher in the HI group and varied less over target location than in the NH group. An analysis with an auditory detection model applied to predict the measured data suggested that this behavior is linked to the dominance of high frequency components of the target signal and their interaction with the head shadow, reverberation and hearing loss.The effect of spatial uncertainty on audibility was found to be negligible. In the fourth study, a new method to assess spatial hearing in complex environments was presented. Instead of single-source localization, the task in this study was the identification and localization of multiple concurrent talkers in a simulated reverberant environment. The results suggested that the applied method provided a reliable measure of the listeners' ability to localize the direction of the talkers within the horizontal plane, but not to measure their distance. This limitation was linked to the considered stimuli, which did not allow an unambiguous identification of the individual sources. The results of this thesis demonstrate that it is important to control target audibility when assessing spatial hearing in complex environments. This is best done with an auditory detection model, but the model developed here needs to be further improved before it can be used to reliably predict masked thresholds for individual hearing impaired listeners. Finally, it is suggested that a multi-talker localization task similar to the one presented here may be a useful tool to analyze the effects of hearing loss and hearing devices on spatial hearing with a high ecological validity.