The role of reverberation in release from masking due to spatial separation of sources in speech recognition

Arbogast et al. [ARO Mtg. (2002)] found a large release from masking obtained by spatial separation of a target talker and competing speech masker. Both stimuli were sentences from the Coordinate Response Measure corpus [Bolia et al., J. Acoust. Soc. Am. (2000)] processed by extracting the envelopes of 15 narrow frequency bands and using the envelopes to modulate carrier tones at the center of each band. By playing nonoverlapping subsets (6–8) of bands from signal and masker they minimized the energetic component while maximizing the informational component of masking. This study extends that work to determine the interaction between reverberation, masker type, and spatial release from masking. Stimuli were processed and presented as above. The target sentence was played at 0-deg azimuth while the masker sentence was played at 0 or 90-deg azimuth. Noise–masker controls were also tested. The listening environment was an IAC booth having dimensions of 12 ft×13 ft. Acoustic extremes were achieved using PlexiglasTM (highly reflective) or foam (highly absorptive). The results indicated that the amount of masking and the spatial release from masking depend both on the characteristics of the room and masker type. Discussion will center on the acoustic and perceptual factors affecting performance. [Work supported by NIH/NIDCD.]