Contribution
Additivity of Segregation Cues and Build-up of Stream Segregation for Speech-in-Speech
* Presenting author
Abstract:
Differences in gender, i.e., differences in f0 and vocal-tract length, and spatial location, help listeners understand speech in the presence of distractor talkers. It is unclear, however, how these cues add up perceptually and how effects build up over time.20 participants completed a speech-in-speech masking experiment using OLSA sentences. Participants reported either one or four words of the target sentence uttered by a male talker at 0° azimuth. Two simultaneous masker talkers were either male, female, or male with the f0 shifted to match the female talker, and either co-located or spatially separated at ±30° azimuth. Speech-reception thresholds (SRTs) were determined by adaptively adjusting the signal-to-noise ratio of the target and masker talkers. Additionally, d’s were calculated at the average SRT across conditions and used to test the additivity of segregation cues by optimal-observer prediction. Finally, build-up was studied by comparing masking release across target words.Results suggest a sub-additive combination of cues, but each added cue helps to separate the target from the maskers by a similar amount. SRTs were lower for single-word reports. For multiple-word reports, build-up of masking release only occurred from the first to the second of the four target words, independent of segregation cue.