SqueakOut: Autoencoder-based segmentation of mouse ultrasonic vocalizations

Abstract

AbstractMice emit ultrasonic vocalizations (USVs) that are important for social communication. Despite great advancements in tools to detect USVs from audio files in the recent years, highly accurate segmentation of USVs from spectrograms (i.e., removing noise) remains a significant challenge. Here, we present a new dataset of12,954annotated spectrograms explicitly labeled for mouse USV segmentation. Leveraging this dataset, we developedSqueakOut, a lightweight (4.6Mparameters) fully convolutional autoencoder that achieves high accuracy in supervised segmentation of USVs from spectrograms, with aDicescore of90.22.SqueakOutcombines a MobileNetV2 backbone with skip connections and transposed convolutions to precisely segment USVs. Using stochastic data augmentation techniques and a hybrid loss function,SqueakOutlearns robust segmentation across varying recording conditions. We evaluate SqueakOut’s performance, demonstrating substantial improvements over existing methods like VocalMat (63.82Dicescore). The accurate USV segmentations enabled bySqueakOutwill facilitate novel methods for vocalization classification and more accurate analysis of mouse communication. To promote further research, we release the annotated12,954spectrogram USV segmentation dataset and theSqueakOutimplementation publicly.