Gustafson G E, Green J A, Tomic T
Dev Psychobiol. 1984 May;17(3):311-24. doi: 10.1002/dev.420170310.
The acoustic features that differentiate the cries of individual human infants were examined in this study. A recognition task, performed by 400 nonparent adults, was used to classify twenty 30-day-old infants as easy or difficult to recognize on the sole basis of their tape-recorded cries. The cries of easy- vs difficult-to-recognize infants were then compared on measures of duration, fundamental frequency, peak frequency in the spectrum, signal-to-noise ratios, and energy in selected frequency bands. The results indicated that each of these measures differentiated the cries of easy- vs difficult-to-recognize infants. Such redundancy should make the cry robust to degradations of selected acoustic parameters, thus enhancing its ability to carry information about individual identity across distances.
本研究对区分人类个体婴儿哭声的声学特征进行了考察。一项由400名非父母成年人执行的识别任务,被用于仅根据20名30日龄婴儿的录音哭声,将他们分类为容易识别或难以识别。然后,在持续时间、基频、频谱中的峰值频率、信噪比以及选定频段的能量等指标上,对容易识别与难以识别婴儿的哭声进行了比较。结果表明,这些指标中的每一项都能区分容易识别与难以识别婴儿的哭声。这种冗余应使哭声对选定声学参数的退化具有鲁棒性,从而增强其在远距离传递个体身份信息的能力。