Dolphins are among the “chattiest” creatures in the oceans. They use sound to find and catch food, as part of their courtship rituals, and just to stay in touch with other dolphins.
Many of those sounds are short clicks and high-pitched squeaks. And a team of researchers is using computers to pick out patterns of these sounds produced by different species of dolphins and whales.
The scientists are using a technique known as machine learning. They feed large amounts of data to a computer, and the program looks for patterns. Finding patterns is relatively easy for trained human experts, but it takes a long time. Computers can do it faster, but they need to learn how to find patterns buried in huge amounts of data.
In this case, the scientists used audio recorded at five sites in the Gulf of Mexico over a two-year period. The recordings included more than 50 million “clicks” produced by dolphins and whales.
The computer wasn’t given any examples to guide it. Instead, it had to sort through all the clicks to find patterns on its own. And it identified seven types of clicks. Researchers recognized one type, which is produced by one species of dolphin. Now they’re trying to tie the other types of clicks to other species.
Their goal is to develop a way to keep an “ear” on dolphins. Identifying which species are in a body of water can help biologists track changes in populations, perhaps as a result of oil spills or climate change -- all by listening to the sounds of these “chatty” creatures.