Age estimation by different listener groups

Abstract

This contribution is focused on the perception of speaker age. With forensic applications in mind, the following questions are. asked : ( 1) are trained listeners better at estimating a speaker s chronological age than untrained listeners, (2) are listeners more confident about correct estimates than incorrect ones and (3) do a speaker's smoking habits influence perceived age? The results of an experiment involving direct age estimations by experts and non-experts reveal that professionals cannot be expected to do significantly better than the untrained listeners as long as correlations between perceived age and calendar age are considered. However, phonetic training did manifest itself in greater intersubject consistency. Furthermore, smoking can be demonstrated to influence age estimation in the sense that the smokers were judged to be older than non-smokers of the same calendar age. An examination of the confidence ratings shows that expert listeners tend to be more cautious than non-experts, but confidence does not imply correctness for either group.