UVic Researchers Use AI to Decode Unique Fish Sounds

Biologists from the University of Victoria have made significant strides in understanding the sounds made by various fish species. Their research reveals that even closely related species produce distinct sounds, which can be differentiated using advanced technology. By employing passive acoustics, researchers identified unique auditory signatures for eight fish species native to Vancouver Island.

The team developed a machine learning model capable of predicting fish species based on their sounds with an impressive accuracy rate of 88 percent. According to Darienne Lancaster, a PhD student at UVic and the lead researcher, “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 way ornithologists use bird songs to identify specific bird species.

Unique Sounds and Their Implications

The findings highlight the diversity of fish communication. For instance, the black rockfish produces a long, growling sound reminiscent of a frog’s croak, while the quillback rockfish emits a series of short knocks and grunts. Lancaster elaborated, “It has been exciting to see how many different species of fish make sounds and the behaviours that go along with these calls.”

The research indicates that some sounds serve specific purposes. The quillback rockfish makes rapid grunting noises when threatened, suggesting a defensive mechanism, while the copper rockfish produces knocking sounds while pursuing prey on the ocean floor.

Innovative Research Techniques

To capture these fish sounds, Lancaster utilized a method known as passive acoustic monitoring. This process involves collecting underwater audio and video data through a sound localization array designed by Xavier Mouy, a former UVic PhD student and collaborator on the project. By analyzing sound characteristics, researchers were able to discern subtle differences in how various species communicate.

The AI model utilized a comprehensive set of 47 different sound features, including duration and frequency, to identify the small variations in sounds that distinguish each species. The techniques developed in this research hold potential for scientists globally, enabling them to decode fish calls in different environments.

Funding for this project was provided by the Natural Sciences and Engineering Research Council of Canada and Fisheries and Oceans Canada, underscoring the importance of this research in understanding marine life and ecosystems.

As the study progresses, the potential applications of this technology could enhance biodiversity conservation efforts and improve our understanding of aquatic environments. The ability to identify fish by their sounds could lead to more informed management practices in fisheries and marine habitats.