Automatic detection for bioacoustic research: a practical guide from and for biologists and computer scientists

Area of Science:

• Ecology
• Bioacoustics
• Computational Biology

Background:

• Passive acoustic monitoring (PAM) is increasingly used in ecological studies, generating large datasets.
• Manual analysis of extensive PAM data is becoming impractical due to data volume.
• Advances in machine learning (ML) offer potential for automated analysis of acoustic events.

Purpose of the Study:

• To review trends in bioacoustic PAM and the need for automated data analysis.
• To explore ML methods and tools for automatic acoustic event detection.
• To provide a practical guide for implementing automatic detection in bioacoustics.

Main Methods:

• Review of current literature on bioacoustic PAM and machine learning applications.
• Exploration of various ML algorithms and computational tools for acoustic data analysis.
• Development of a step-by-step guide for building automated detection pipelines.

Main Results:

• The field of automatic acoustic event detection in bioacoustics is rapidly evolving.
• Machine learning offers viable solutions for analyzing large-scale acoustic datasets.
• Bridging the expertise gap between biology and computer science is crucial for adoption.

Conclusions:

• Automated detection using ML is essential for managing growing bioacoustic data.
• This review provides foundational knowledge and practical steps for bioacousticians and computer scientists.
• Future directions include refining ML models and expanding their application in ecological research.