Identify fish by sound: scientists can now recognize species from their sounds

Fishermen have long insisted that fish make audible noises beneath the waves, a claim often met with skepticism from the scientific community. A new study out of British Columbia gives those seasoned anglers more than just stories — it shows that fish noises can be reliably matched to individual species using modern acoustic tools.

Researchers combined underwater sound arrays with synchronized video to pinpoint which reef fish were making which calls. The work, led by scientists at the University of Victoria and published in the Journal of Fish Biology, opens the door to non-invasive ways to monitor marine life and to decode an underwater soundscape that’s been largely ignored.

How scientists pinpointed fish calls on rocky reefs

To locate and identify the sources of underwater sounds, the team deployed an acoustic localization array in Barkley Sound. This setup uses multiple hydrophones to triangulate where a sound originates — similar to how GPS uses signals from several satellites. By running this system alongside time-synced video, researchers could visually confirm which fish produced a given noise.

The approach paired precise acoustic tracking with direct observation, eliminating much of the uncertainty that has plagued past attempts to link sound to species.

Which reef species were heard — and which were recorded for the first time

The study captured more than 1,000 distinct fish sounds and successfully matched calls to eight rocky-reef species. Among the identifications were several rockfish types and other common reef dwellers.

• Multiple rockfish species, including newly documented calls from canary rockfish and vermillion rockfish.
• Lingcod, known for their presence on rocky bottoms.
• Pile perch and kelp greenling — smaller, vocal reef fish often overlooked in acoustic surveys.

Recording calls from canary and vermillion rockfish was a notable first — the study officially adds those species to the roster of sound-producing marine animals.

Machine learning reveals species-specific sound signatures

After collecting raw audio, the team turned to a machine learning model to sort and classify the calls. The algorithm examined 47 different acoustic features — elements like pitch, duration, and temporal patterns — to build distinct profiles for each species.

The model achieved classification accuracies as high as 88 percent in some comparisons, even when distinguishing between closely related fish. That level of performance demonstrates that species produce reliably different acoustic signatures that computers can learn to recognize.

What features and behaviors the sounds reflect

• Call pitch and duration often correlated with body size — smaller fish tended to emit higher-frequency sounds.
• Some sound types lined up with observed behaviors such as courtship displays and aggressive interactions.
• Other vocalizations occurred during feeding attempts or escape responses, suggesting a range of functional contexts for fish sounds.

Why underwater listening could transform conservation and monitoring

Traditional population surveys rely on nets, traps, or divers — methods that can be disruptive, costly, and limited in duration. Acoustic monitoring offers an alternative: long-term, low-impact surveys that can operate day and night.

If researchers can consistently estimate fish size, location, and abundance just from audio, it would provide a powerful new tool for marine managers. Continuous sound recording could reveal seasonal trends, population shifts, or the impact of human activity without removing animals from their habitats.

Practical next steps: counting fish by sound and hearing regional accents

Looking ahead, the team plans to refine techniques to count individuals by analyzing call rates and to test whether fish populations in different areas have distinct acoustic signatures — in other words, do fish from separate reefs sound different?

• Develop algorithms that translate call frequency and pattern into reliable abundance estimates.
• Compare recordings across sites to detect possible regional “accents” or dialects among the same species.
• Integrate acoustic monitoring with other remote-sensing tools to build richer population models.

Proving that fish produce identifiable and repeatable calls is a first step toward a quieter revolution in marine science — one that listens more and disturbs less.

You might also like:

• Dolphin health revealed by seawater eDNA sampling in a conservation first
• Night dive reveals new fish-anemone relationship
• Oyster reef restoration off Portsmouth becomes England’s largest with 200+ volunteers
• Seeds hear rain: MIT study finds sound waves speed germination
• Why this mysterious 45-kg fish found on a beach has scientists intrigued

William Anderson

William Anderson is a multimedia producer specializing in videos, podcasts, and interactive galleries. With five years of immersive content creation, he turns information into a rich audio‑visual experience. His storytelling skills draw you directly into the heart of every story, on any platform.