Press

FishEye Collaborative is announcing a major advance in marine biodiversity monitoring: the first public dataset of natural, species-specific fish sounds recorded directly in the wild. Conservation technologist and marine bioacoustics researcher Matt Duggan will present the findings today at the Acoustical Society of America’s Spring Meeting in New Orleans.

The study, recently released as a preprint on bioRxiv, introduces a scalable approach to underwater sound-source identification that combines spatial audio processing with 360° video. The system—called the UPAC-360—enables researchers to attribute fish sounds to individual species on crowded coral reefs with high confidence, something that has not previously been possible at scale.

“We’ve known for decades that coral reefs are full of sound, but we’ve never had the tools to tell who was making the sound,” said Duggan. “With this method, we can now link specific fish to specific sounds in real-world reef conditions.”

Over the course of four days of fieldwork off the coast of Curaçao, FishEye’s team identified and documented the natural sounds of 40 reef-associated fish species—none of which had publicly available in situ recordings prior to this effort. These verified recordings are now accessible via FishEye’s open reference library, enabling researchers, marine managers, and machine learning developers to improve the accuracy and utility of passive acoustic monitoring (PAM) data.

The approach has immediate applications in monitoring species presence, spawning activity, and ecological changes in marine protected areas. By creating a validated foundation for acoustic machine learning models, FishEye’s dataset helps bridge a longstanding gap between raw soundscape recordings and actionable biodiversity insights.

The study and recordings are available at:
 Preprint – bioRxiv
 Sound Library – FishEyeCollaborative.org/library