How to Speak Whale: Inside Project CETI's Effort to Decode Sperm Whale Clicks With AI

Nothing on this page shows what a coda means.
The feature itself cautions that its visuals illustrate the research process, not decoded whale language.
Mark

Why does Dominica matter so much to this project?

Mimi

Sperm whales live there year-round, which is rare. Researchers can keep observing the same animals and listening to their exchanges rather than catching a brief passing visit.

Mark

What does the team actually have out there?

Mimi

Petri lists three boats, drones carrying tags, hydrophones in the water, and analysis pipelines looking for structure in the data.

Luke

Structure isn't meaning, though. Has anyone decoded anything?

Mimi

Not according to this piece. The feature even says nothing on the page shows what a coda means.

Mark

So what's being paired with what?

Mimi

Whale clicks are paired with dive data, so patterns in sound can be compared against what the animals are doing.

Luke

And the click timings in the demo? Are they expert-labeled?

Mimi

No. They were detected automatically from a Project CETI recording, not hand-annotated.

Mark

What's Petri's own view of the limits?

Mimi

He says such complicated events probe the limits of what we can ever hope to translate.

Luke

That's the line to hold onto. The scale, more than 50 researchers, is confirmed. Whether translation is possible is still open.

  • Sperm whales, up to 60 feet long, are being followed through the waters off Dominica, where they live year-round and can be watched over time.
  • Three boats, drone-delivered tags, and hydrophones in the water feed a demanding stream of acoustic data that researchers must sort and analyze.
  • Machine learning and network science are being pressed to find repeating structure in whale clicks and link it to what the animals do during their dives.
  • The work is still in the tools-and-patterns stage, and no one can yet say what any whale click means, leaving the promise of translation open and unproven.

On a rainforested Caribbean island where volcanic peaks meet deep blue water, a physicist has come to study one of the largest creatures that has ever lived, not to find dinosaurs but to listen to a living mind that speaks in clicks. Giovanni Petri and his colleagues in Project CETI are pairing the animals' sounds with records of their dives, hoping that structure in the noise will reveal something of how sperm whales talk to one another. Humanity has long wondered whether other species have language, and this effort stands at the threshold of that old question, still short of an answer. The patterns they hear may yet prove to be a true language, or they may mark the edge of what human understanding can ever carry across.

Giovanni Petri describes the feeling of arriving on Dominica for the first time as stepping into Jurassic Park. The island, known as Nature Island, is wrapped in rainforest, volcanic mountains, waterfalls, and reefs that drop into deep water. Petri, a professor at Northeastern University London, first came in August 2022, and he was not there for dinosaurs. He came for the sperm whale, the largest predator alive today, which can reach 60 feet and weigh as much as 57 tons. Dominica is one of only a few places where these whales stay all year, making it possible to watch and listen to them over time.

Petri is a lead researcher on Project CETI, a non-profit founded in 2020 that aims to use artificial intelligence, machine learning, and whale-friendly robotics to record and eventually decode sperm whale communication. With more than 50 researchers at institutions in the United States and beyond, it is described as the largest interspecies communication effort in history. On the island, the team runs three boats, drone-carried tags, hydrophones, and analysis pipelines, pairing whale clicks with dive data and using network science to examine how the animals converse.

The accompanying interactive feature shows how such sounds might one day be read. It draws on press-kit recordings of a whale named Pinchy and on a 2024 Nature Communications dataset of Dominican sperm whale codas, and its visuals are schematic, with stylized models. The feature is clear that nothing on the page reveals what any coda means.

Petri is candid about the limits. Some of these exchanges, he suggests, may test the boundaries of what can ever be translated. The team has built tools that find patterns, but whether those patterns amount to language is still unproven. The question to watch is whether the structure in the clicks can be matched to what the whales are actually doing beneath the surface.

Giovanni Petri says arriving on the Caribbean island of Dominica for the first time feels a little like stepping into Jurassic Park. Petri is a professor at the Network Science Institute at Northeastern University London. He first came to the island in August 2022. Dominica is known as Nature Island, with thick rainforest, volcanic mountains, dramatic waterfalls, and coral reefs that drop away into deep blue water. He wasn't there to study dinosaurs. He was there for the largest predator alive today, the sperm whale.

Sperm whales can grow to 60 feet and weigh as much as 57 tons. They range across the world's oceans, from the equator to the cold seas around both poles. Dominica is one of only a handful of places where they live year-round, which gives researchers a chance to watch their behavior and listen to their exchanges over time.

Petri is one of the lead researchers for Project CETI, a non-profit scientific initiative. Its aim is to use advanced artificial intelligence, machine learning, and whale-friendly robotics to record the acoustic communication of sperm whales and, ultimately, decode it. The project was founded in 2020 and is described as the largest effort at interspecies communication in history. It includes more than 50 researchers at leading institutions in the United States and elsewhere.

In Dominica, the team has built a full research platform: they observe whales, engineer listening devices, and develop new ways to analyze what they collect. Petri describes the setup plainly. There are three boats, drones carrying tags, hydrophones in the water, and whole pipelines of analysis trying to find structure in the data. The approach pairs the whales' clicks with data about their dives, using new sensors and network science methods to look at how the animals converse.

The source piece, an interactive feature from Northeastern Global News, walks through how the sounds could be turned into something understandable. It includes recordings of a whale named Pinchy from the Project CETI press kit. The click timings shown were detected automatically from that same recording rather than annotated by hand. The coda rhythms and whale-to-whale exchange come from a Dominica sperm whale coda dataset published by Sharma and colleagues in Nature Communications in 2024. The visualizations are schematic, and the 3D whales and equipment are stylized models. The feature states explicitly that nothing on the page shows what a coda means.

Petri himself is cautious about the limits. Such complicated events, he said, probe the limits of what we can ever hope to translate. The project has built the tools to find patterns in whale sound, but translating them into language remains unproven. What to watch is whether the structure found in the click data can be tied to what the whales are actually doing underwater.

We have three boats, drones carrying tags, hydrophones in the water and whole pipelines of analysis trying to find structure in the data.
— Giovanni Petri, Northeastern University London
Such complicated events probe the limits of what we can ever hope to translate.
— Giovanni Petri
Nous contacter FAQ