Foundation Models for Bioacoustics -- a Comparative Review

TL;DR

This review evaluates large-scale pretrained bioacoustic models, analyzing their transferability, training data, and architecture, and presents empirical results comparing their performance on biodiversity monitoring benchmarks.

Contribution

It provides a systematic overview and empirical comparison of bioacoustic foundation models, highlighting their transferability and guiding model selection for new tasks.

Findings

Perch 2.0 achieves highest BirdSet score and strong linear probing results.

BirdMAE excels among probing strategies on BirdSet.

Self-supervised models trained on AudioSet outperform specialized bird sound models.

Abstract

Automated bioacoustic analysis is essential for biodiversity monitoring and conservation, requiring advanced deep learning models that can adapt to diverse bioacoustic tasks. This article presents a comprehensive review of large-scale pretrained bioacoustic foundation models and systematically investigates their transferability across multiple bioacoustic classification tasks. We overview bioacoustic representation learning by analysing pretraining data sources and benchmarks. On this basis, we review bioacoustic foundation models, dissecting the models' training data, preprocessing, augmentations, architecture, and training paradigm. Additionally, we conduct an extensive empirical study of selected models on the BEANS and BirdSet benchmarks, evaluating generalisability under linear and attentive probing. Our experimental analysis reveals that Perch~2.0 achieves the highest BirdSet score (restricted evaluation) and the strongest linear probing result on BEANS, building on diverse multi-taxa supervised pretraining; that BirdMAE is the best model among probing-based strategies on BirdSet and second on BEANS after BEATs$_{NLM}$, the encoder of NatureLM-audio; that attentive probing is beneficial to extract the full performance of transformer-based models; and that general-purpose audio models trained with self-supervised learning on AudioSet outperform many specialised bird sound models on BEANS when evaluated with attentive probing. These findings provide valuable guidance for practitioners selecting appropriate models to adapt them to new bioacoustic classification tasks via probing.