Electronic Thesis and Dissertation Repository


Doctor of Philosophy


Health and Rehabilitation Sciences


Ewan Macpherson, Margaret Cheesman


Previous studies provide evidence that listening experience in a particular reverberant environment improves speech intelligibility and localization performance in that environment. Such studies, however, are few, and there is little knowledge of the underlying mechanisms. The experiments presented in this thesis explored the effect of reverberation context, in particular, the similarity in interaural coherence within a context, on listeners' performance in sound localization, speech perception in a spatially separated noise, spatial release from speech-on-speech masking, and target location identification in a multi-talker configuration.

All experiments were conducted in simulated reverberant environments created with a loudspeaker array in an anechoic chamber. The reflections comprising the reverberation in each environment had the same temporal and relative amplitude patterns, but varied in their lateral spread, which affected the interaural coherence of reverberated stimuli. The effect of reverberation context was examined by comparing performance in two reverberation contexts, mixed and fixed. In the mixed context, the reverberation environment applied to each stimulus varied trial-by-trial, whereas in the fixed context, the reverberation environment was held constant within a block of trials.

In Experiment I (absolute judgement of sound location), variability in azimuth judgments was lower in the fixed than in the mixed context, suggesting that sound localization depended not only on the cues presented in isolated trials. In Experiment II, the intelligibility of speech in a spatially separated noise was found to be similar in both reverberation contexts. That result contrasts with other studies, and suggests that the fixed context did not assist listeners in compensating for degraded interaural coherence. In Experiment III, speech intelligibility in multi-talker configurations was found to be better in the fixed context, but only when the talkers were separated. That is, the fixed context improved spatial release from masking. However, in the presence of speech maskers, consistent reverberation did not improve the localizability of the target talker in a three-alternative location-identification task. Those results suggest that in multi-talker situations, consistent coherence may not improve target localizability, but rather that consistent context may facilitate the buildup of spatial selective attention.