UVic Biologists Harness AI to Decode Distinct Fish Sounds

Researchers at the University of Victoria have made significant strides in understanding fish communication by using artificial intelligence (AI) to identify and differentiate the sounds made by various fish species. Their findings reveal that even closely related fish produce unique and identifiable sounds, allowing for a new method of species identification.

Using a technique known as passive acoustic monitoring, the team successfully identified distinct sounds from eight fish species native to Vancouver Island. Their innovative approach involved developing a machine learning model capable of predicting which sounds correspond to specific species with an impressive accuracy rate of 88 percent.

Darienne Lancaster, a PhD student at UVic and the lead researcher on this project, expressed excitement about the discovery. In a recent news release, she stated, “We knew previously that many fish were making sounds in the wild, but we didn’t know which sounds belonged to which species, or if it was possible to tell these sounds apart.” This breakthrough parallels the established practice of using bird songs to identify different bird species.

The research team uncovered fascinating sound patterns among the fish. For instance, the black rockfish emits a long, growling sound reminiscent of a frog’s croak, while the quillback rockfish produces a series of short knocks and grunts. Lancaster noted that these sounds are not merely random; they can indicate specific behaviours. “Some fish, like the quillback rockfish, make rapid grunting sounds when they’re being chased by other fish, so it’s likely a defensive mechanism,” she explained.

In contrast, other species, like the copper rockfish, create knocking sounds when pursuing prey along the ocean floor. Such behaviours highlight the complex communication system among fish, which is often overlooked.

To gather the necessary audio data, Lancaster utilized a sound localization array developed by Xavier Mouy, a former PhD student at UVic and collaborator on the project. This system captures underwater audio and video, enabling researchers to analyze sound characteristics and distinguish between various species’ calls. The AI model employed a comprehensive set of 47 different sound features, including duration and frequency, to detect subtle differences in the calls of each species.

The implications of this research extend beyond Vancouver Island. The techniques developed by Lancaster can be adapted by scientists globally to decode fish calls in various marine environments. This advancement could enhance our understanding of marine ecosystems and the roles that different species play within them.

Funding for this groundbreaking research came from the Natural Sciences and Engineering Research Council of Canada and Fisheries and Oceans Canada. As scientists continue to explore the aquatic soundscape, the potential for new discoveries regarding fish behaviour and communication is vast. By leveraging technology like AI, researchers are paving the way for a deeper understanding of the underwater world.