VoICE: A semi-automated pipeline for standardizing vocal analysis across models

Area of Science:

• Neuroscience
• Bioacoustics
• Computational Biology

Background:

• Vocal communication in animal models offers insights into speech and communication disorders.
• Current vocal analysis methods lack standardization, hindering cross-laboratory and cross-species comparisons.

Purpose of the Study:

• To introduce VoICE (Vocal Inventory Clustering Engine), a novel approach for standardizing and analyzing animal vocalizations.
• To demonstrate VoICE's utility in identifying vocal repertoire changes in different species and genetic models.

Main Methods:

• VoICE groups vocal elements by creating a high-dimensionality dataset based on spectral similarity.
• Hierarchical clustering and automated algorithm pruning are used to define distinct vocalization types.
• The method was applied to birdsong and a mouse model for neurodevelopmental disorders.

Main Results:

• VoICE accurately captured acoustic property deterioration in birdsong post-deafening, including altered sequencing.
• Analysis of mice lacking the Cntnap2 gene revealed a reduced vocal repertoire.
• The study highlights VoICE's capability to standardize vocalization analyses across diverse species.

Conclusions:

• VoICE provides a standardized, automated method for analyzing animal vocalizations.
• This tool facilitates cross-species and cross-laboratory comparisons, advancing research in vocal communication and related disorders.
• VoICE is a valuable asset for the scientific community studying the neurogenetic basis of communication.