Whales talk to each other. An Indian woman at MIT hopes to crack their ‘codas’ with AI - Telegraph India

Marina: Welcome to Deep Voices — I'm so glad you're here. I'm Marina, a marine biologist who has spent more years than I can count absolutely enchanted by whales and dolphins, and this podcast is my way of sharing that enchantment with you. Every episode we dive into the songs, the cultures, the family bonds, and the big beautiful science trying to understand what these animals are actually saying to each other. Marina: Before we dive in, a warm thank you to today's sponsors — Ada, who helps businesses integrate A I into their workflows in ways that actually make sense; Ago Consulting — that's "Ahh-Go" — silicon development experts working across everything from A I to full system-on-chip design; and Zen Semiconductor, an umbrella company with a whole portfolio of businesses across the silicon industry. These folks make Deep Voices possible, and I'm grateful for every one of them. Marina: Today we have a genuinely thrilling lineup. We're going to meet a researcher at MIT who is using A I to crack open the secret code of sperm whale clicks. We're going to talk about what birdsong and whale song have in common with human language — and why that discovery is a bigger deal than it might sound. And we're going to sit with a story about humpback whales doing something so strange, so unexpected, that scientists are still scratching their heads trying to explain it. Let's go. Marina: Okay, so — thread one. Voices of the deep. And I want to start by painting you a picture, because I think it earns everything that comes after. Imagine you're underwater in the Caribbean, somewhere dark and impossibly blue, and a sperm whale — an animal that can be sixty feet long and weigh fifty tons — surfaces near your boat. And then it dives. And as it goes down, it produces this sound. Click. Click-click. Click. It's almost like morse code. Crisp, rhythmic, precise. And here's the thing — that pattern isn't random. That whale is talking. Marina: Sperm whales communicate using what researchers call codas — these short, patterned sequences of clicks. Different combinations, different rhythms, different numbers of clicks. And for a long time, scientists suspected these codas carried real meaning, that they were something like a language, or at least a sophisticated communication system. But proving it, and decoding it — that's been the hard part. Marina: Now, a study published in the journal Nature Communications in 2024 brought us the closest we've ever been. Researchers from Project CETI — that stands for Cetacean Translation Initiative — recorded and analyzed thousands of sperm whale codas off the island of Dominica. And what they found was extraordinary. The codas aren't just a small fixed vocabulary. They have what linguists call combinatorial structure. The whales vary the tempo, the rhythm, the ornamentation — these little extra clicks — and the rubato, which is basically a kind of expressive timing. And when you combine those variables, you get a combinatorial space that could potentially encode thousands of distinct units of meaning. Marina: Think about that for a second. These animals aren't just exchanging simple signals. They may be operating with something that has genuine complexity — layers of structure nested inside each other. The researchers compared it to a phonetic alphabet. You don't just have sounds, you have a system for combining sounds in ways that multiply your expressive range enormously. Marina: And into this picture steps someone I find genuinely inspiring. A researcher with roots in India, working at MIT, who is applying the tools of machine learning and natural language processing — the same kind of A I that helps computers understand human speech — to sperm whale codas. The idea is to train A I models on large libraries of recorded whale communication, and look for patterns, regularities, structures that the human ear and human analysis might miss. It's painstaking, it requires enormous amounts of data, but the early results are giving researchers real hope that we are, slowly, beginning to read the grammar of a whale. Marina: And there's something I love about this moment in science — the fact that A I, which we built to understand ourselves, might turn out to be our best tool for understanding minds that are completely unlike ours. That feels significant. That feels like something. Marina: Now — thread two. Culture and decoding. And I want to bring in a finding that genuinely stopped me in my tracks when I came across it. Researchers studying the acoustic structure of birdsong and whale song have found that both share patterns with human language that go deeper than anyone expected. I'm talking about patterns at a statistical level — the way sounds are arranged, the way sequences are structured. These similarities point to what linguists call Zipf's law and other statistical regularities that show up in human language universally. Marina: Zipf's law, if you haven't heard of it, is this fascinating pattern where in any language, the most common word appears roughly twice as often as the second most common word, three times as often as the third, and so on. It's a power law. And it shows up in English, in Mandarin, in ancient Sumerian. It shows up in birdsong. And it shows up in whale song. What that suggests — and scientists are careful here, they're not saying whales have language the way we do — but what it suggests is that the pressure to communicate efficiently, to encode information in structured, layered ways, may produce similar solutions across very different species. Marina: And that connects beautifully to what we know about whale culture. Because here's the thing — whales don't just communicate, they transmit culture. Humpback whale song is one of the most documented examples of cultural transmission in the non-human animal world. Males sing these extraordinarily complex songs — songs that last for hours, that evolve over months and years, and that spread across ocean basins. A new phrase emerges in the songs of whales off Australia, and within a couple of years, whales in French Polynesia are singing it too. That is culture. That is learning. That is innovation spreading through a community. Marina: And it's not just humpbacks. Sperm whales live in what researchers call clans — groups that share distinct dialects, distinct coda repertoires that are passed down through generations. A sperm whale calf learns to click not by instinct alone, but by listening to its family. Its grandmother. Its aunts. There is something deeply moving about that. Marina: Project CETI is trying to place non-invasive hydrophones and biologging tags on sperm whales in Dominica to record not just their sounds but the social context around those sounds — who is speaking, who is listening, what is happening in the group. Because meaning doesn't live in the sound alone. Meaning lives in context. And once you have context alongside sound, alongside A I that can detect patterns across thousands of hours of recordings, you have a genuine shot at translation. Marina: I also want to pause here and acknowledge a question that scientists and ethicists are genuinely wrestling with right now. If we crack the code — if A I gives us a window into what dolphins and whales are actually communicating — should we try to talk back? There are researchers who think the answer is yes, eventually, cautiously. There are others who worry about the disruption that could cause — to the animals, to their social structures, to the trust that wildlife researchers have built over decades. It is not a simple question. And I think the fact that we're asking it says something good about where science is heading. Marina: Alright — thread three. And this one is pure wonder. Let me tell you about humpback whales doing something that has left marine biologists genuinely baffled. In recent years, researchers have documented something called "gaping" behavior in humpback whales. These are animals opening their mouths — sometimes very wide — in contexts that don't seem to fit the usual explanations. Humpbacks do open their mouths to feed, of course, that's a well-known behavior. But this is different. This gaping is happening in social situations, during interactions between whales, sometimes at the surface, sometimes during what looks like play or socializing. Marina: And scientists don't know what it means. Is it a form of communication? Is it a threat display? Is it something to do with sensory information — humpbacks have sensory hair follicles on their rostrums, on their chins, and some researchers think gaping might expose those structures in ways that allow the whale to gather information about its environment or about another animal close by. Or is it something else entirely — a behavior we don't have a category for yet? Marina: What I love about this story is what it reveals about how much is still unknown. Humpback whales have been studied intensively for decades. They are probably the most-watched, most-filmed, most-recorded large whale on the planet. And they are still showing us behaviors we cannot explain. That's not a failure of science. That's an invitation. Marina: And it connects to a broader truth about cetaceans that I come back to again and again. These animals have brains that are, in some ways, remarkably like ours — with a highly developed limbic system, which is the part of the brain associated with emotion and social bonding, and with spindle cells, which are a type of neuron associated with self-awareness and empathy in humans. They have lives that last decades. They have friendships. They grieve. They play. They innovate. And they are doing all of this in an environment — the ocean — that is so different from ours that even imagining their sensory world requires a real leap of imagination. Marina: There is a documented case that has always stayed with me — a humpback whale off the coast of California that appeared to be entangled in crab pot lines, hundreds of pounds of rope wrapped around her flippers and her tail. A team of divers went in and spent about an hour cutting her free, a dangerous, painstaking process. And when the last line was cut, she didn't just swim away. She circled back. She approached each diver individually. She nudged them. Some described it as nuzzling. And then she was gone. Now, we can't know what was in her mind. Science asks us to be careful about that. But I will tell you — the divers who were there that day were not careful. They were certain she was saying thank you. And I find I don't have the heart to argue with them. Marina: That's our episode. Three threads, one ocean, and a whole world of minds still waiting to be understood. What moves me most — and I hope it comes through — is that we are living in this extraordinary window of time when the tools to actually listen, really listen, to what whales and dolphins are saying to each other are finally within reach. A I built to understand human language, turned toward the songs of sperm whales. Statistical tools that reveal the deep grammar of humpback song. Tags and hydrophones that capture not just sound but social context. We are so close to something. Marina: Deep Voices, brought to you by Ada, Ago Consulting, and Zen Semiconductor. If this episode made you feel something — if you found yourself leaning in — please subscribe, leave a review, and share it with someone who loves the ocean or is just starting to. Every listener you bring in is another voice for these animals. And check the show notes for links to the research and stories we talked about today. Until next time — stay curious, stay kind, and keep listening.