Automatic mouse ultrasound detector (A-MUD): A new tool for processing rodent vocalizations

Area of Science:

• Animal Behavior
• Bioacoustics
• Computational Biology

Background:

• House mice (Mus musculus) produce complex ultrasonic vocalizations (USVs) during social and sexual interactions.
• USVs exhibit song-like features, including syllable types arranged in sequential patterns.
• Manual analysis of USV data is labor-intensive, time-consuming, and prone to subjectivity.

Purpose of the Study:

• To develop an automated algorithm for detecting mouse ultrasonic vocalizations (USVs).
• To improve the efficiency and objectivity of USV data processing.
• To systematically compare the performance of the developed algorithm against manual analysis and commercial software.

Main Methods:

• Development of the Automatic Mouse Ultrasound Detector (A-MUD) algorithm.
• Implementation of A-MUD within the free STx acoustic software (S_TOOLS-STx version 4.2.2).
• Evaluation of A-MUD's error rates using manually segmented USV files as a gold standard.
• Comparative analysis of A-MUD's performance against a commercially available detection program.

Main Results:

• A-MUD demonstrated a significant improvement in processing efficiency, being 4-12 times faster than manual segmentation.
• A-MUD exhibited lower error rates compared to commercial software, with significantly more true positives and fewer false positives/negatives.
• The primary errors generated by A-MUD were false negatives, indicating a high specificity.

Conclusions:

• A-MUD provides a highly efficient and accurate automated method for detecting mouse ultrasonic vocalizations.
• This study represents the first systematic comparison of error rates for automatic USV detection methods.
• A-MUD and its future versions will be made available to the scientific community to advance bioacoustic research.