Perception of talker height and sex from children’s voices

Abstract

Several recent studies have investigated the ability of listeners to judge the height of adult talkers from their speech. Here, we examine the perception of talker height from a sample of /hVd/ syllables spoken by children ranging in age from 5 through 18 years, with 3 talkers of each sex at each age level. Twenty college students judged the height and sex of the talker for each trial. Perceptual estimates of height were moderately correlated with actual height (for males, r = 0.78; for females, r = 0.67). Consistent with our recent findings concerning the perception of talker age [Barreda & Assmann, JASA-EL 143(5), EL361, 2018], listeners' judgments of talker height appear to incorporate assumptions about talker sex and the relationships between speech acoustics and veridical height for male and female talkers of different heights. For example, when older girls were misidentified as boys, their height was likely to be underestimated. This behavior results in predictable error patterns in height judgments from speech: in situations where talker sex is misidentified, listeners apply the “wrong” acoustic model to estimate talker height, leading to correlated error patterns between apparent height and sex.