Validating a model to detect infant crying from naturalistic audio.
Validating a model to detect infant crying from naturalistic audio.
复制标题
DOI：
10.3758/s13428-022-01961-x
复制
发表时间：
2023-09
影响因子：
5.4
通讯作者：
de Barbaro, Kaya
中科院分区：
文献类型：
--
作者：
Micheletti, Megan;Yao, Xuewen;Johnson, Mckensey;de Barbaro, Kaya
人类婴儿的哭声演变为一种信号，以引起父母的照顾，并积极影响婴儿的行为，以及照顾者的互动。近几十年来，自动化的哭泣检测算法变得越来越流行，虽然存在一些模型，但它们还没有在一整天的自然主义音频记录中进行彻底的评估。在这里，我们通过在对发展研究人员很重要的评估场景中测试一种新的深度学习哭泣检测模型来验证它。我们还评估了深度学习模型相对于LENA的哭泣分类器的性能，LENA的哭泣分类器是量化儿童哭泣的最常用商业软件系统之一。总的来说，我们发现深度学习和LENA模型的输出都显示出对婴儿哭声的人类注释的收敛有效性。然而，相对于LENA，深度学习模型在所有测试时间尺度（24小时，1小时和5分钟）上具有更高的准确性指标（召回，F1，kappa）和与人类注释的更强相关性。平均而言，相对于人类注释和深度学习模型，LENA每24小时低估婴儿哭泣50分钟。此外，两种自动模型检测到的每日婴儿哭泣时间低于文献中父母报告的估计值。我们提供建议和解决方案，以利用自动化算法来检测家中的婴儿哭声，并使我们的训练数据和模型代码开源和公开。
Human infant crying evolved as a signal to elicit parental care and actively influences caregiving behaviors as well as infant-caregiver interactions. Automated cry detection algorithms have become more popular in recent decades, and while some models exist, they have not been evaluated thoroughly on daylong naturalistic audio recordings. Here, we validate a novel deep learning cry detection model by testing it in assessment scenarios important to developmental researchers. We also evaluate the deep learning model’s performance relative to LENA’s cry classifier, one of the most commonly-used commercial software systems for quantifying child crying. Broadly, we found that both deep learning and LENA model outputs showed convergent validity with human annotations of infant crying. However, the deep learning model had substantially higher accuracy metrics (recall, F1, kappa) and stronger correlations with human annotations at all timescales tested (24 hours, 1 hour, and 5 minutes) relative to LENA. On average, LENA underestimated infant crying by 50 minutes every 24 hours relative to human annotations and the deep learning model. Additionally, daily infant crying times detected by both automated models were lower than parent-report estimates in the literature. We provide recommendations and solutions for leveraging automated algorithms to detect infant crying in the home and make our training data and model code open source and publicly available.
登录
查看更多内容
影响因子：
5.2
作者：
Petzoldt, Johanna;Wittchen, Hans-Ulrich;Martini, Julia
通讯作者：
Martini, Julia
影响因子：
4.6
作者：
RICHTERS, J;PELLEGRINI, D
通讯作者：
PELLEGRINI, D
影响因子：
2.6
作者：
Stifter, Cynthia A.;Spinrad, Tracy L.
通讯作者：
Spinrad, Tracy L.
影响因子：
2.6
作者：
Gilkerson, Jill;Richards, Jeffrey A.;Paul, Terrance D.
通讯作者：
Paul, Terrance D.
影响因子：
2.1
作者：
Barr, Ronald G.;Fairbrother, Nicole;Brant, Rollin
通讯作者：
Brant, Rollin