![\"\"](\"https:\/\/images.newscientist.com\/wp-content\/uploads\/2024\/09\/17202147\/SEI_222070861.jpg\")
<\/div>A Bryde\u2019s whale surfacing for air<\/p>\n
NOAA Fisheries<\/p>\n<\/div>\n<\/figcaption><\/figure>\n<\/p>\n
A mysterious sound emitted from the depths of the Pacific Ocean has finally been identified as a Bryde\u2019s whale. Now, artificial intelligence is helping researchers track the elusive whale species responsible for the call.<\/p>\n
The puzzle began in 2014 when researchers recorded a sound resembling a moan followed by metallic pings over the Mariana Trench. \u201cYour average person would not think that it was made by an animal \u2013 they would think it was some ship or the [US] Navy,\u201d says Ann Allen<\/a> at the National Oceanic and Atmospheric Administration (NOAA).<\/p>\n Years later, additional recordings of the sound, which researchers call a biotwang, were eventually linked to sightings of Bryde\u2019s whales (Balaenoptera brydei<\/em>) at the nearby Mariana Islands.<\/p>\n Bryde\u2019s whales are large baleen whales that filter huge amounts of krill and fish from the water, but they are rarely seen because they are nearly always on the move. Allen saw an opportunity to track migrating Bryde\u2019s whales by finding similar biotwang sounds in more than 180,000 hours of underwater recordings from NOAA\u2019s network of hydrophones mounted on the Pacific seafloor.<\/p>\n \u201cIt\u2019s important any time you can discover a new call type for any species of animal that occurs most of the time out of sight of people because it allows you to use passive acoustic monitoring to detect their presence,\u201d says Caroline Casey<\/a> at the University of California, Santa Cruz.<\/p>\n <\/p>\n Given the impracticality of manually sifting through the audio recordings, Allen teamed up with Google researchers who had been developing an AI model to automatically identify vocalisations of multiple whale species. Google trained its AI to categorise different patterns in spectrogram images, which convert sound to a visual form.<\/p>\n