Talker/listener interactions in speech on speech masking: Effects of age and sex
Many factors determine performance when perceiving speech in the background of other talkers. We have recently found two examples of interactions between the characteristics of the listener and those of the target/masker talkers in such tasks. In both studies, we used a modified version of the Coordinate Response Measure to measure Speech Reception Thresholds (SRTs) from both adults and primary-school-aged children. Sentences were of the form ‘Show the [animal] where the [colour] [digit] is’ with 6 animals, 6 colours and 8 digits. The listener’s task was always to report the coloured number from the target sentence (containing ‘dog’). Speech maskers had the same form as the targets but with a different animal, colour and number.
In the first study, we investigated the impact of adult female and child interfering talkers (an 11-year old female) on the ability to perceive target speech from the same two talkers. SRTs were measured in four conditions, varying factorially target age (adult or child) and masker type (speech from the opposite age talker, or speech-spectrum-shaped noise matched to the target). Performance for children with the noise maskers was roughly similar to that of adults, with the adult talker more intelligible. Performance for children with speech maskers was, however, the same whether a child was supposed to attend to the child or adult, whereas for the adults, the adult target led again to better performance. It seems likely that this interaction arises from an attentional effect, with children more distracted by another child’s voice than by an adult’s (preferring fun to authority!).
In a second study, we used male and female adult target talkers in the presence of two simultaneous other talkers, one male and one female. We found a small, but robust interaction between the sex of the listener and the sex of the target, with listeners performing slightly better with talkers of their own sex. Note that this interaction cannot result from listeners preferring voices similar to their own, because target talkers were always adults, and the effect was as strong in children as in adults.
We are currently investigating the generality of this interaction by determining whether it occurs in the presence of ‘uninteresting’ maskers like speech-shaped noise, or whether it requires informational maskers like other speech.