From Raw Recordings to Objective Behavioral Data

Python signal-processing and statistical analysis for a published mouse communication study. Five-minute audio-video recordings from 20 mice yielded 6,018 ultrasonic vocalizations, classified into 10 syllable types and aligned with four scored behaviors. The workflow combines synchronized capture, pattern extraction, chi-square and Fisher's exact tests, and distribution-similarity measures. Results include 4,489 versus 1,529 calls in treated and control groups. The deck links the public code repository and the 2024 European Journal of Neuroscience paper, on which I am first author.