British Columbia

Hey Google, help save the whales: Engineers developing AI to recognize calls of endangered orcas

Engineers at Google have been teaching a machine to recognize orca sounds to help scientists locate where the whales are in the Salish Sea.

Aim is to have a listening network that can detect killer whales and alert marine traffic in the area

In this Jan. 18, 2014, file photo, endangered orcas from J pod swim in Puget Sound west of Seattle, as seen from a federal research vessel that has been tracking the whales. (Elaine Thompson/The Associated Press)

Engineers at Google have been working on an artificial-intelligence model aimed at helping orcas in the Salish Sea. 

The project began after the U.S. National Oceanic and Atmospheric Administration contacted the tech giant for help trying to find humpback whales in a very large volume of underwater recordings, explained Google software engineer Matt Harvey.

The group decided to focus on the southern resident killer whale population and develop a system to identify them by their calls, so that traffic in the area can then be alerted to their presence. 

These whales, which are listed as endangered under the Canadian Species at Risk Act, frequent the area of the Salish Sea and feed primarily on chinook salmon. 

Working with Fisheries and Oceans Canada (DFO), Google's engineers used about 1,800 hours of underwater audio, labelled and collected from different whales, to train a neural network to classify the audio. 

"The exact technology is called a convolutional neural network," Harvey said, adding it is similar to the one the company uses to distinguish the content of different kinds of images.

So far the team has been able to distinguish different kinds of whales from each other, but they don't have enough source data to distinguish a resident killer whale from a transient one.

"We're going to need to source that data to continue to refine this," he said. 

Continuously listening hydrophones — underwater listening devices — are already set up in different locations in the Salish Sea and they are sophisticated enough to send alerts back to the Marine Mammal Unit at the DFO when an orca is nearby in real time. 

Harvey says there are many possible applications of this alert system, like helping the team locate and assist a sick animal or diverting whales to a safer area in case of a petroleum spill or discharge.

"One of the primary threats that the whales are facing are entanglement and just difficulty foraging due to vessel noise," he added.  "So they can use this [alert system] to advise traffic or the vessels themselves that orcas are in this location, consider slowing down or consider change of course."

The southern resident killer whale population has struggled in recent years. Three members from the group died last summer, and another is currently missing and feared dead

Listen to the interview on All Points West here:

Matt Harvey, software engineer with Google AI, discusses new technology being used by Fisheries and Oceans Canada in an effort to better protect killer whales off the southern coast of B.C. 6:34

For more on the future of the southern resident killer whales, listen to our CBC British Columbia podcast Killers: J pod on the brink

With files from All Points West


To encourage thoughtful and respectful conversations, first and last names will appear with each submission to CBC/Radio-Canada's online communities (except in children and youth-oriented communities). Pseudonyms will no longer be permitted.

By submitting a comment, you accept that CBC has the right to reproduce and publish that comment in whole or in part, in any manner CBC chooses. Please note that CBC does not endorse the opinions expressed in comments. Comments on this story are moderated according to our Submission Guidelines. Comments are welcome while open. We reserve the right to close comments at any time.