Automated detection of 50-kHz ultrasonic vocalizations using template matching in XBAT

Abstract

Background Ultrasonic vocalizations (USVs) have been utilized to infer animals' affective states in multiple research paradigms including animal models of drug abuse, depression, fear or anxiety disorders, Parkinson's disease, and in studying neural substrates of reward processing. Currently, the analysis of USV data is performed manually, and thus time consuming. New Method The goal of the present study was to develop a method for automated USV recognition using a ‘template detection’ procedure for vocalizations in the 50-kHz range (35-80 kHz). The detector is designed to run within XBAT, a MATLAB graphical user interface and extensible bioacoustics tool developed at Cornell University. Results Results show that this method is capable of detecting >90% of emitted USVs and that time spent collecting data by experimenters is greatly reduced. Comparison with Existing Methods Currently, no viable and publicly available methods exist for the automated detection of USVs. The present method, in combination with the XBAT environment is ideal for the USV community as it allows others to 1) detect USVs within a user-friendly environment, 2) make improvements to the detector and disseminate and 3) develop new tools for analysis within the MATLAB environment. Conclusions The present detector provides an open-source, accurate method for the detection of 50-kHz USVs. Ongoing research will extend the current method for use in the 22-kHz frequency range of ultrasonic vocalizations. Moreover, collaborative efforts among USV researchers might enhance the capabilities of the current detector via changes to the templates and the development of new programs for analysis.

Automated acoustic analysis of 50-kHz ultrasonic vocalizations using template matching and contour analysis

“…Finally, this emphasizes the assertion that a subset of files should always be scored manually in order to produce and report inter-rater reliability between manual and automated scoring. 2 Our future goal is to further automate USV detection and analysis to remove human review and, therefore, significantly increase speed of analysis and improve repeatability and reliability. The method presented here adds a powerful post hoc/postdetection acoustic analysis method that both increases analysis speed and keeps the data in the MATLAB environment for continued analysis.…”

“…2 Briefly, USVs were detected in XBAT, a MATLAB based graphical user interface and extensible bioacoustics tool developed by the Bioacoustics Research Program at Cornell University. A library of USVs between 35 and 80 kHz were used as templates for the detection of USVs in recorded ".wav" files using spectrogram correlation.…”

“…Ultrasonic Vocalizations (USVs) are a useful tool for inferring affective states in the rat and have been incorporated in research paradigms modeling important human conditions such as drug abuse, [1][2][3] depression, fear, and anxiety disorders, 4 Parkinson disease, 5 and aging. 6 Moreover, USVs are a useful tool for examining the neurobiological substrates of reward.…”

“…Previously, we developed and validated a novel tool for automatically detecting 50-kHz USVs using a template matching procedure. 2 The goal of the present study was to develop and validate a method for measuring acoustic parameters of USVs following the automatic template detection. Parameters included the minimum, maximum, and mean fundamental frequencies of the entire USV, bandwidth, duration, the minimum, maximum, and mean derivative, number of frequency modulations per second, number of frequency peaks, and number frequency peaks per second.…”

Ethological concepts enhance the translational value of animal models

“…This is a very time consuming process as the number of calls can go from 500 to 1000 different elements in just one recording of 10 min. Recently, the first attempts to automate the analysis of ultrasonic sounds have been reported (Barker et al, 2014;Reno et al, 2013), which hopefully will initiate the further technical development of these automated tools.…”