Listening to Seabirds with AI
On Stora Karlso, RISE researchers are using AI and Adaptive Change-Point Detection to analyze guillemot calls, turning vast audio recordings into ecological insights about seabird behavior and marine health.
On Stora Karlso, the Auk Lab rises like an artificial cliff above the Baltic Sea. Each summer, thousands of guillemots crowd its ledges, filling the air with a constant chorus of calls. For years, researchers have watched their every move through cameras and sensors. Now, with help from RISE, they are listening too.
The Challenge of Analyzing Bird Calls
Eighteen microphones capture the colony’s nonstop chatter, including courtship songs, feeding calls, and warning cries, creating an ocean of sound far too vast for humans to analyze alone. To turn noise into knowledge, RISE researchers have introduced Adaptive Change-Point Detection, a method that helps experts train AI models that pinpoint when important vocal events occur.
Continuous Improvement Through Feedback
With each round of human feedback, the system improves, revealing patterns that would otherwise remain buried in hours of recordings.
Why It Matters
These insights matter. As deep-diving top predators, guillemots mirror the health of the sea itself. Changes in their behaviour offer early clues about shifting fish stocks, warming waters, and the pressures of a changing climate.
Broader Applications
By linking sound with video and environmental data, ecologists can now explore entirely new questions about communication, stress, and survival. And because the tools are open source, this work may soon help protect vulnerable ecosystems far beyond Stora Karlso.
