cs.SDNov 4, 2026

Drowning in Noise: On Signal Saliency in Self-Supervised Learning for Ocean Acoustics

Authors: Shikhar Bharadwaj, Samuele Cornell, Shinji Watanabe, Malcolm Slaney

Organizations: Carnegie Mellon University, USA · Stanford, USA

Abstract

Self-supervised learning (SSL) for audio typically benefits from larger pre-training corpora, but this relationship assumes that training data carries meaningful learning signal. In ocean acoustics, this assumption is often violated: passive recordings are dominated by waves, wind, and silence, with biological vocalizations occupying a small fraction of the total duration. We show that this overrepresentation of noise causes SSL models to allocate representational capacity to acoustically trivial structure, degrading downstream performance. Using BEATs as a testbed, we demonstrate that pre-training on 30 hours of salient recordings (segments containing animal vocalizations) outperforms 1,240 hours of unfiltered data on a 31-class marine mammal classification benchmark. Under linear probing, the manually curated salient set achieves markedly better accuracy than the full corpus, despite using 41 times less data. Automatic saliency-based filtering improves over unfiltered pre-training, but still falls well short of the manually curated salient set. Our findings suggest that, in domains where informative content is temporally sparse, data curation matters more than data scale for SSL pre-training.

Figures & tables

Explore similar work

CardsList
  1. MarineSet: Large-Scale Underwater Acoustic Data Curation and evaluation

    Oct 21, 2026Hilde I. Hummel, Sandjai Bhulai, Rob D. van der Mei +1Underwater AcousticsTraining Data Curation

  2. Dolph2Vec: Self-Supervised Representations of Dolphin Vocalizations

    Jun 10, 2026Chiara Semenzin, Faadil Mustun, Roberto Dessi +5Audio Representation LearningAudio Understanding

  3. A Self-Supervised Approach for Minimal-Annotation Hydroacoustic Data Exploration

    Jul 7, 2026Pierre-Yves Raumer, Axel Marmoret, Dorian Cazau +6Audio Representation LearningSelf-Supervised Learning