TU Berlin

Quality and Usability Lab08_06_2020_Uhrig

Inhalt des Dokuments

zur Navigation

Impact of Talker Location on Speed of Behavioral Talker Identification in a "Turn-Taking" Listening Scenario

Location:  Zoom link (Please ask Saman Zadtootaghaj for access)  

Date/Time: 08.06.2020, 14:15-15:00 

SPEAKER: Stefan Uhrig (TU Berlin)

Abstract:  This study investigates the importance of talker location for behavioral talker identification in a listening situation involving two talkers, solely one would be speaking at a time. Spoken sentences served as stimuli, which were either clean, degraded due to background noise or bandpass filtering. Stimuli were serially presented through three horizontal loudspeakers: In non-spatial reproduction, both talkers were presented through a central loudspeaker; in spatial reproduction, each talker was presented through the central or a talker-specific lateral loudspeaker. Participants identified talkers via speeded keypresses. Behavioral responses were delayed by degraded speech due to obscuring of talker voice cues. Additional talker location cues entailed faster responses for talker identification, which were still slower than responses for pure talker localization, suggesting a primary response strategy based on voice recognition. Response times for spatial reproduction of clean speech were generally increased relative to non-spatial reproduction, reflecting costs in switching spatial auditory attention that gradually diminished under degraded speech. Behavioral effects of spatialization manifested even in the absence of significant differences in subjective category ratings (speech quality, speech intelligibility, voice similarity, talker identification effort), implying a higher sensitivity of response times for detecting more subtle variation in talker identification performance.



Schnellnavigation zur Seite über Nummerneingabe