预印本分析220项哺乳动物生物声学研究：发声类型识别方法差异大、原始音频开放率不足2%
Current state of bioacoustics research of vocal repertoires
一项预印本分析了197个哺乳动物物种、220项研究中用于识别发声 repertoire 的方法，发现个体叫声特征提取、声谱图特征分析和叫声类型聚类方法存在很大差异，且对新型声学分析与聚类方法的采用率较低。研究还指出，所纳入研究中提供原始音频数据集的不足2%。作者据此呼吁该领域加强方法标准化与数据透明度。该研究为预印本，尚未完成同行评审。
Knowledge of the unique call types produced by a species has historically been an important part of ecological research. However, the definition of a call type is not straightforward, and across species and studies different methods are employed to study call types and vocal repertories. Here, we analysed the methodology used to identify the vocal repertoires of 197 mammalian species across 220 studies. We found a large variability in the features extract from individual calls, methods to analyse spectrographic features, and methods to cluster call types. We further see a low uptake of novel methods for acoustic analysis and clustering. Lastly, we identify a lack of open data across the included studies, with less than 2% of studies providing the raw audio dataset underlying their paper. Based on these results, we call for increased standardization of methods and data transparency across the field.
中文翻译还没有生成，先显示原文。
来源：bioRxiv · Animal Behavior and Cognition · biorxiv.org