Petrini Karin, Tagliapietra Sara
University of Padova, Italy.
J Speech Lang Hear Res. 2008 Apr;51(2):485-501. doi: 10.1044/1092-4388(2008/035).
This study examined age-related differences in speaker similarity judgments, in which acoustic cues known to be important in speaker and speech recognition and identification were varied.
Four groups of listeners of 5- to 6-, 8- to 9-, 10- to 11-, and 25- to 30-year-olds were asked to judge the similarity between an original talker's speech sample and 9 versions of a second speech sample from the same talker, obtained by manipulating the pitch and speech rate.
Results showed that although pitch was used by all age groups as information to judge the similarity between the single talker speech samples, there was a significant age-related increase in the weight of speech rate attributed to similarity judgments. Furthermore, an age-related effect in accuracy was found, as was an effect of pitch level for the younger group.
The use of relevant acoustic features to discriminate a speaker changes with age, probably arising from a shift in cue weighting strategy as described by the developmental weighting strategy hypothesis (S. Nittrouer, C. Manning, & G. Meyer, 1993). Future studies should clarify whether the present findings can be extended to any pair of acoustic cues and should involve longitudinal methods to precisely detect when the strategy shift occurs.
本研究考察了说话者相似度判断中与年龄相关的差异,其中对说话者及语音识别和辨认中已知重要的声学线索进行了变化处理。
四组听众,年龄分别为5至6岁、8至9岁、10至11岁和25至30岁,被要求判断原始说话者的语音样本与来自同一说话者的第二个语音样本的9个版本之间的相似度,这9个版本是通过改变音高和语速获得的。
结果显示,虽然所有年龄组都将音高用作判断单个说话者语音样本之间相似度的信息,但在归因于相似度判断的语速权重方面,存在与年龄相关的显著增加。此外,发现了与年龄相关的准确性效应,以及较年轻组的音高水平效应。
用于区分说话者的相关声学特征的使用随年龄而变化,这可能源于发展加权策略假说(S. 尼特鲁尔、C. 曼宁和G. 迈耶,1993年)所描述的线索加权策略的转变。未来的研究应阐明本研究结果是否可扩展到任何一对声学线索,并且应采用纵向研究方法来精确检测策略转变发生的时间。