Capturing the songs of mice with an improved detection and classification method for ultrasonic vocalizations (BootSnap)

House mice communicate through ultrasonic vocalizations (USVs), which are above the range of human hearing (>20 kHz), and several automated methods have been developed for USV detection and classification. Here we evaluate their advantages and disadvantages in a full, systematic comparison, while also presenting a new approach. This study aims to 1) determine the most efficient USV detection tool among the existing methods, and 2) develop a classification model that is more generalizable than existing methods. In both cases, we aim to minimize the user intervention required for processing new data. We compared the performance of four detection methods in an out-of-the-box approach, pretrained DeepSqueak detector, MUPET, USVSEG, and the Automatic Mouse Ultrasound Detector (A-MUD). We also compared these methods to human visual or ‘manual’ classification (ground truth) after assessing its reliability. A-MUD and USVSEG outperformed the other methods in terms of true positive rates using default and adjusted settings, respectively, and A-MUD outperformed USVSEG when false detection rates were also considered. For automating the classification of USVs, we developed BootSnap for supervised classification, which combines bootstrapping on Gammatone Spectrograms and Convolutional Neural Networks algorithms with Snapshot ensemble learning. It successfully classified calls into 12 types, including a new class of false positives that is useful for detection refinement. BootSnap outperformed the pretrained and retrained state-of-the-art tool, and thus it is more generalizable. BootSnap is freely available for scientific use.Author summary: House mice and many other species use ultrasonic vocalizations to communicate in various contexts including social and sexual interactions. These vocalizations are increasingly investigated in research on animal communication and as a phenotype for studying the genetic basis of autism and speech disorders. Because manual methods for analyzing vocalizations are extremely time consuming, automatic tools for detection and classification are needed. We evaluated the performance of the available tools for analyzing ultrasonic vocalizations, and we compared detection tools for the first time to manual methods (“ground truth”) using recordings from wild-derived and laboratory mice. For the first time, class-wise inter-observer reliability of manual labels used for ground truth are analyzed and reported. Moreover, we developed a new classification method based on ensemble deep learning that provides more generalizability than the current state-of-the-art tool (both pretrained and retrained). Our new classification method is free for scientific use.