
Researchers at the University of Washington School of Medicine have developed a software program that enlists artificial intelligence to decipher the ultrasonic vocalizations made by mice and rats.
The “DeepSqueak” program, described today in a study published by the journal Neuropsychopharmacology, focuses on squeaks and whistles that are well above the range of human hearing.
Software converts the audio signals into visual graphs, or sonograms, and then puts those images through the kinds of machine-vision algorithms that are used for autonomous vehicles.
“DeepSqueak uses biomimetic algorithms that learn to isolate vocalizations by being given labeled examples of vocalizations and noise,” Russell Marx, one of the study’s authors, explained in a news release.