Effect of reverberation on spatial unmasking for nearby speech sources

Abstract

Individualized HRTFs were measured in a moderately reverberant room (T60=450 ms) for sources directly in front of and to the right of a listener for both near (15 cm) and far (1 m) distances. The full HRTFs (including reverberation) and pseudo-anechoic HRTFs (time windowing out the reverberation) were used to simulate a speech target and a speech-shaped noise masker over headphones. Speech reception thresholds were measured adaptively, varying the target level while keeping the masker level constant at the better ear. Thresholds were measured for both left and right monaural signals as well as for binaural signals. Results show the magnitude of spatial unmasking that can arise for sources very close to the head, where large interaural level differences (ILDs) arise, and determine the degree to which spatial unmasking is due to better ear and binaural effects. These results are compared to previous anechoic results simulating sources near a listener in which large ILDs appear to degrade binaural performance below predicted better-ear performance. Comparisons between pseudo-anechoic and realistic reverberation conditions address the degree to which reverberation interferes with spatial unmasking. [Work supported by a grant from the Air Force Office of Scientific Research.]