Bioacoustic processing and analyses of mouse vocalizations: Current methods and future directions.

Area of Science:

• Bioacoustics
• Animal Communication
• Computational Biology

Background:

• Rodents, including house mice (Mus musculus), use sonic and ultrasonic vocalizations (USVs) for communication, but their functions remain largely unknown.
• Analyzing audio recordings of these vocalizations presents significant challenges in data processing and interpretation.

Purpose of the Study:

• To provide a comprehensive review of current bioacoustic tools for processing and analyzing mouse vocalizations.
• To identify and address challenges in the data processing pipeline, from audio recording to USV classification and sequencing.
• To propose solutions and a hierarchical framework for future research in rodent acoustic communication.

Main Methods:

• Review of conventional and alternative methods for audio file processing, including time-frequency representations and mouse-adapted techniques.
• Comparison of machine learning and signal processing approaches for automated USV detection, denoising, and feature extraction.
• Evaluation of supervised and unsupervised classification methods, alongside manual validation, for analyzing USVs and sequencing approaches.

Main Results:

• Identified critical steps in the bioacoustic data processing pipeline where errors can propagate.
• Highlighted the need for improved denoising, frequency contour tracking, and feature extraction methods.
• Emphasized the importance of reliable manual methods as a gold standard for validating automated machine learning approaches.

Conclusions:

• Current bioacoustic tools require refinement to accurately capture and analyze mouse vocalizations.
• Developing robust automated methods, validated by manual approaches, is crucial for advancing the study of rodent communication.
• Addressing technological challenges will enable deeper insights into the functions of USVs in house mice.