Factors for speech intelligibility in the symmetric and anti-symmetric speech-masker conditions
OA Version
Citation
Abstract
The goal of this work is to better understand why it is easier for some than others to understand an in-front target talker presented amid two spatially-distributed interfering talkers. Speech reception thresholds (SRTs) of young listeners (normal-hearing and hearing-impaired) were measured, with the maskers arranged in the following spatial configurations: 1) "colocated", both in-front, 2) "symmetric", at ± 60° azimuth, and 3) "anti-symmetric", both at either +60°, or -60°, azimuth. SRTs were measured both binaurally and monaurally, with monaural stimuli presented to the "better-ear." Binaural benefits were calculated as the decrease in threshold when the second ear was added to the better-ear. Sensitivity to interaural time differences, interaural level differences and interaural cross-correlation were measured in the same listeners, using narrow-band noises centered at a low frequency (500Hz) and at a high frequency (2 or 4kHz). Large inter-subject differences were found for both listener groups among both interaural difference sensitivity thresholds and binaural benefits for the spatially-separated conditions. No correlations, for either of the listener groups, were observed between symmetric binaural benefit and any of the interaural difference sensitivity measures. Anti-symmetric binaural benefit significantly correlated with interaural difference sensitivity for many of the tasks, both for the normal-hearing group and among the set of hearing-impaired listeners with severe-amd-less hearing-loss. SRTs were predicted using the short-time Equalization-Cancellation (EC) model, in which two types of parameters were varied: first, the standard deviations of time-and intensity-jitters at the inputs of the binaural processing mechanism (analogous to changes in interaural sensitivity) and second, the temporal window in the binaural processing mechanism (affecting benefit from short-time binaural advantages). EC-window length was most important for predicting changes in symmetric SRTs. Changes in jitter standard deviations were most important for predicting changes in anti-symmetric SRTs. These results suggest some key differences in the mechanisms related to individual differences for the two speechmasker conditions. Both interaural difference sensitivity and a mechanism suggested by the importance of the EC-window length might be important for the symmetric condition, while interaural difference sensitivity is a key factor for the anti-symmetric condition.
Description
Thesis (Ph.D.)--Boston University