Human screams have been suggested to comprise a salient and readily identified call type, yet few studies have explored the degree to which people agree on what constitutes a scream, and the defining acoustic structure of screams has not been fully determined. In this study, participants listened to 75 human vocal sounds, representing both a broad acoustical range and array of emotional contexts, and classified each as to whether it was a scream or not. Participants showed substantial agreement on which sounds were considered screams, consistent with the idea of screams as a basic call type. Agreement on classifications was related to participant gender, emotion processing accuracy, and empathy. To characterize the acoustic structure of screams, we measured the stimuli on 27 acoustic parameters. Principal components analysis and generalized linear mixed modeling indicated that classification as a scream was positively correlated with 3 acoustic dimensions: one corresponding to high pitch and roughness, another corresponding to wide fundamental frequency variability and narrow interquartile range bandwidth, and a third positively correlated with peak frequency slope. Twenty-six stimuli were agreed upon by > 90% of participants to be screams, but these were not acoustically homogeneous, and others evoked mixed responses. These results suggest that while screams might represent a salient and possibly innate call type, they also exhibit perceptual and acoustic gradation, perhaps reflecting the wide range of emotions and contexts in which they occur.

中文翻译：

---

那是尖叫吗？听众协议和主要区别声学特征

已经提出人的尖叫声包括明显且容易识别的呼叫类型，但是很少有研究探索人们对构成尖叫声的共识程度，并且尚未完全确定尖叫声的定义声学结构。在这项研究中，参与者聆听了75种人的声音，这些声音既代表着广泛的听觉范围，又代表着一系列的情感语境，并对其是否尖叫声进行了分类。与会者对将哪些声音视为尖叫声达成了基本共识，这与将尖叫声作为基本通话类型的想法相一致。关于分类的协议与参与者的性别，情感处理的准确性和同理心有关。为了表征尖叫声的声学结构，我们在27个声学参数上测量了刺激。主成分分析和广义线性混合建模表明，尖叫声的分类与三个声学维度呈正相关：一个对应于高音高和粗糙度，另一个对应于宽基频可变性和狭窄的四分位数范围带宽，第三个与峰值正相关频率斜率。超过90％的参与者同意有26种刺激是尖叫，但这些刺激在听觉上并不均匀，其他刺激引起了混合反应。这些结果表明，虽然尖叫声可能代表着突出的，可能是与生俱来的呼叫类型，但它们也表现出感知和听觉上的渐变，这可能反映了它们发生的各种情感和环境。