Supervised vs Self-Supervised Representation Learning for Fin Whale Vocalization Detection

Haodong Zhang, Adam Chareyre, Shuwen Ge, Randall Balestriero, Hervé Glotin

Keywords: Self-Supervised Learning, Bioacoustics, Passive Acoustic Monitoring, Fin Whale Vocalizations, Transfer Learning, Noise Robustness

TL;DR: Self-supervised learning reduces annotation needs and improves cross-site robustness for fin-whale detection, offering a scalable tool for passive acoustic monitoring.

Abstract: Fin whales produce low-frequency vocalizations critical for monitoring but are often masked by anthropogenic noise. While supervised detectors perform well, they require costly labels and degrade under noise or data scarcity. We present the first application of self-supervised learning (SSL) to fin-whale detection, combining contrastive predictive coding with an amplitude-aware encoder. Across datasets collected in the Norwegian Sea and the Mediterranean Sea, SSL models outperform supervised Transformers in low-label and low SNR regimes and transfer effectively across regions. Embedding visualizations further show robust class separability. These results highlight SSL as a scalable approach for passive acoustic monitoring, reducing annotation needs and paving the way for scalable, label-efficient acoustic monitoring across diverse marine habitats.

Submission Number: 41