Coral reefs emit a complex soundscape that can be recorded and analyzed by AI.

Researchers at the University of Exeter have trained an AI algorithm on multiple recordings of the soundscapes from coral reefs, teaching it to distinguish between sounds from healthy and degraded reefs. The AI could later listen to new reef recordings and accurately deliver an assessment of their health. The findings are invaluable for monitoring reef restoration projects. (Unite.ai)

Song of the reefs

Unknown to most people, coral reefs have an extremely complex soundscape that can give insights into the health of the coral body. Researchers dubbed these sounds the “song of the reef” and experimented with recording them and using AI for their analysis.

This was found to be a more reliable method compared to a visual examination of the reefs.

According to Ben Williams, leader author of the research, visual and acoustic surveys of reef health suffer from being highly labor-intensive and are not nearly as accurate as machine learning.

“Visual surveys are also limited by the fact that many reef creatures conceal themselves, or are active at night, while the complexity of reef sounds had made it difficult to identify reef health using individual recordings,” he continued. “Our approach to that problem was to use machine learning — to see whether a computer could learn the song of the reef.”

Reef restoration and AI

When scientists used the trained AI model on new recordings, they found it delivered accurate assessments of reef health 92% of the time.

“Our findings show that a computer can pick up patterns that are undetectable to the human ear. It can tell us faster, and more accurately, how the reef is doing,” Williams added.

Related Story: The NOAA’s S.O.S to AI and Robotics for Coral Restoration