In realistic cocktail party listening scenarios where a target talker faces a listener while masker talkers typically face other directions, listeners may receive less masker energy at extended high frequencies (EHFs; 8 kHz) due to directional propagation of EHFs. This leads to unmasking of target EHF cues, which could help listeners identify the target talker and/or segregate target speech. Children’s greater sensitivity at EHFs and reliance on greater signal bandwidth could make EHF cues especially useful for children. Motivated by previous work on children’s use of EHF cues for speech recognition, we tested adult and school-age child listeners in a multi-talker scene where a target talker faced the listener and two spatially separated masker talkers did not. Listeners were asked to (1) identify the spatial location of a target talker facing them, or (2) detect a change in the target talker’s head orientation. Stimuli were either full-bandwidth or low-pass filtered at 8 kHz to limit EHF access. Performance on both tasks was better, and response times were quicker, with larger masker talker head orientation angles and with access to EHF cues. Differences between child and adult performance will be discussed. Work supported in part by NIH Grant DC019745.
Ananthanarayana et al. (Wed,) studied this question.