Application of machine learning in analyses of bioacoustics data from terrestrial mammals and insects: a systematic review and appraisal of methods.

Background: Bioacoustics studies sound production, transmission, and reception in animals. Machine learning (ML) has emerged as a transformative tool in bioacoustics in recent years. ML algorithms, especially deep learning models, have demonstrated the ability to process and classify vast datasets of animal vocalizations efficiently. With the rapid advancements in computational power and algorithmic innovation, ML enables scalable and automated species monitoring. The major effort lies with birds and aquatic mammals. However, challenges remain for terrestrial animals, including primates and rodents. The challenge arises from variability in recording conditions and limited annotated datasets. This study aims to systematically review recently developed ML algorithms applied to classify bioacoustics data focusing on terrestrial animals. Methods: This systematic review will select and analyze studies published between January 1, 2020, and January 31, 2025, identified through searches in Web of Science and PubMed using predefined search strategies. The review involves ASReview-aided title and abstract screening as well as manual full-text screening. Multiple reviewers will verify the final selections. We will extract data on algorithms, taxa, training and testing datasets, and performance, evaluating study quality using a developed instrument. Data synthesis: Findings will be synthesized narratively, highlighting algorithm performance across taxa, orders, and regions. Visual and tabular summaries will illustrate insights into effective ML algorithms for bioacoustics classification, accounting for dataset variability and study objectives.