Sound Finder: a new software approach for localizing animals recorded with a microphone array

Acoustic localization is a powerful technique for monitoring the positions, movements and behaviours of terrestrial animals. However, its prevalence in biological studies has been constrained by hardware and software that are custom-built, expensive and difficult to use. We recently helped to relieve the hardware constraint by describing a microphone array that is affordable, portable, easy to use and commercially available. Here, we help to relieve the software constraint by developing an acoustic localization program called “Sound Finder”, which is easy to use, freely available and accurate for a variety of animals and recording conditions. It runs in the free software environment R, and in spreadsheet programs such as Microsoft Excel and the open-source software LibreOffice. In this study, we describe how Sound Finder functions, and then test its accuracy by localizing natural sounds that were broadcast through loudspeakers and re-recorded with microphone arrays. We quantify Sound Finder's accuracy by comparing its location estimates with known loudspeaker locations and with output from other localization approaches. We show that Sound Finder generates accurate location estimates for a variety of animal sounds, microphone array configurations and environmental conditions. Furthermore, Sound Finder generates an error value that allows the user to assess its accuracy. In conclusion, Sound Finder provides accurate estimates of a vocalizing animal's location. It is easy to use, requires only widespread and affordable software and is freely available in a standard form as Supplementary material to this article.

声学定位（acoustic localization）是一种用于监测陆生动物位置、活动与行为的高效技术。然而，该技术在生物研究中的普及曾受限于定制化、高成本且操作复杂的硬件与软件系统。此前我们通过一款兼具高性价比、便携性与易用性且可商用的麦克风阵列（microphone array）方案缓解了硬件层面的限制。本研究中，我们通过开发一款名为"Sound Finder"的声学定位程序，进一步破解了软件层面的瓶颈——该程序操作简便、可免费获取，且适配多种动物声源与录音场景的定位需求。该程序可在免费软件环境R，以及Microsoft Excel、开源软件LibreOffice等电子表格软件中运行。本研究中，我们首先阐释了"Sound Finder"的运行原理，随后通过将自然声源通过扬声器播放并使用麦克风阵列重新录制的方式，对其定位精度开展验证测试。我们通过将"Sound Finder"的位置估算结果与已知的扬声器布设位置，以及其他定位方法的输出结果进行对比，量化了该程序的定位精度。实验结果表明，"Sound Finder"可针对多种动物声源、麦克风阵列布设方案与环境条件生成精准的位置估算结果。此外，"Sound Finder"可生成定位误差值，便于使用者自行评估其精度表现。综上，"Sound Finder"可精准估算发声动物的位置。该程序操作简便，仅依赖普及度高且性价比出众的软件，且以标准格式作为本文的补充材料免费公开获取。