Coffey, K. R., Marx, R. G., & Neumaier, J. F. (2019). DeepSqueak: A deep learning-based system for detection and analysis of ultrasonic vocalizations. Neuropsychopharmacology. doi:10.1038/s41386-018-0303-6
DeepSqueak is a fully graphical MATLAB package for detecting and classifying rodent ultrasonic vocalizations (USVs). DeepSqueak is engineered to allow non-experts easy entry into USV detection and analysis.
DeepSqueak provides a fully automated pipeline for USV detection, classification, and analysis:
State-of-the-art regional object detection neural networks (Faster-RCNN) dramatically reduces the false positive rate to facilitate reliable USV analysis in standard experimental conditions.
Create unsupervised k-means clustering models from vocalization contours and trainable neural networks for supervised call classification.
Output call statistics in spreadsheets for further analysis.
Image by Alice Gray
DeepSqueak 2.0 was designed and tested with MATLAB 2017b, 2018a, and 2018b.
To run DeepSqueak, navigate to the main DeepSqueak folder in MATLAB, and type "DeepSqueak" into the command line. DeepSqueak will add itself to the MATLAB path after running.
Copyright © 2018 by Russell Marx & Kevin Coffey. All Rights Reserved.