Original title: An End-to-End Workflow for Fin Whale Song Detection, Note Characterization, and Localization with Distributed Acoustic Sensing
Dídac Diego-Tortosa, Miriam Romagosa, Arantza Ugalde, Hugo Latorre +3 · August 2026
Tracking fin whales across huge areas of the ocean is challenging. To solve this, researchers built an automated system that uses undersea fiber-optic telecom cables as giant underwater listening devices, a method called distributed acoustic sensing. The system spots fin whale calls by identifying distinct sound spikes across the cable channels. It then groups those sound hits into single calls using spatial clustering, combines related groups, and filters out bad signals with mathematical curve fitting. Next, the tool measures sound features like timing and energy to tell call types apart. Finally, by comparing when the sounds hit different points along the cable, it calculates where the whales are swimming. The team showed that this workflow separates overlapping whale calls, identifies specific call patterns, and successfully tracks the animals as they move.
Imagine you are at a stadium full of people. You close your eyes and clap. The sound arrives at each person's ears at slightly different times, and you could figure out where you are by comparing when everyone heard it. Now imagine the whole stadium is tiled with thousands of listening devices, but many of them are unreliable and hear random noise. You only keep the claps that line up in a wave pattern. This paper does exactly that: it uses the ocean cable as thousands of ears, throws away false detections, and then deduces the whale's position from the tiny time differences along the cable.
Fin whales are hard to monitor—they roam vast oceans and their calls are low-frequency. This workflow uses submarine internet cables as thousands of microphones, turning existing infrastructure into a whale-tracking network. It lets scientists detect, classify, and locate whale calls automatically, which could help protect species and study migration patterns without expensive equipment.
The abstract notes the workflow was tested on two specific cables in the Strait of Gibraltar and western Alboran Sea, so performance may differ in other environments. Evaluation used only six fin whale songs, a small manually annotated dataset. The workflow is designed for fin whale notes and may not directly transfer to other species without adaptation. Localization accuracy is not quantified beyond stating a grid-search procedure is used.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
An End-to-End Workflow for Fin Whale Song Detection, Note Characterization, and Localization with Distributed Acoustic Sensing — Dídac Diego-Tortosa, Miriam Romagosa, Arantza Ugalde et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.