Might we soon understand sperm whale speak? | The Excerpt

On a special episode (first released on June 6, 2024) of The Excerpt podcast: These gentle giants of the ocean have long been a source of fascination for us humans. And while researchers have made huge gains in fostering a greater understanding of these majestic creatures in recent decades, the use of AI may have just cracked open the mother lode of all puzzles: thier language. Could this lead to one day the possibility of talking with whales? David Gruber, the founder & president of Project CETI, an interdisciplinary scientific and conservation project aimed at listening to and translating the communication of sperm whales, joins The Excerpt to share this exciting new development.

Hit play on the player below to hear the podcast and follow along with the transcript beneath it. This transcript was automatically generated, and then edited for clarity in its current form. There may be some differences between the audio and the text.

Podcasts: True crime, in-depth interviews and more USA TODAY podcasts right here

Dana Taylor:

That clicking noise you're hearing, is the sound of sperm whales.

Hello and welcome to The Excerpt, I'm Dana Taylor. Today is Thursday, June 6th, 2024, and this is a special episode of The Excerpt. The gentle giants of the ocean have long been a source of fascination for us humans, and while researchers have made huge gains in fostering greater understanding of these majestic creatures in recent decades, the use of AI may have just cracked open the mother load of all puzzles, their language. Might this lead to one day the possibility of talking with whales?

To learn more, I'm joined now by one of the scientists behind this latest development. David Gruber is the founder and president of Project CETI, which stands for Cetacean Translation Initiative, an interdisciplinary scientific and conservation project aimed at listening to and translating the communication of sperm whales. David, thank you so much for joining me.

David Gruber:

My pleasure. Thanks for inviting me.

Dana Taylor:

Let's start with the $64,000 question. What are the chances that one day we could communicate directly with sperm whales?

David Gruber:

Ooh, that is an interesting question. I guess the idea is that we're more interested in understanding the communication, and of course that opens up the possibility, but we already are communicating with many animals like dogs and cats. So there's a ton of communication that's already happening, it's about really using digital bioacoustics and artificial intelligence to look at it another level.

It's currently being described as the new telescope or microscope. Just how the telescope aided us in being able to look out into the cosmos or the microscope allowed us to see inside of cells, by kind of listening with the help of advanced machine learning, we could hear things that maybe the human ear hadn't heard before.

Dana Taylor:

So, I won't ask you to speak to me in sperm whale terms right now or share something with me that you learned from studying these whales. I do want you to talk to me about this promising development and how it came about. What technologies were involved here?

David Gruber:

I'm fortunate and really honored to be the lead of Project CETI, which is the Cetacean Translation Initiative. We're now over 50 scientists coming from about eight different disciplines that have joined together really with the intent of listening to and working to translate the communication system of another animal, and in our case it's sperm whales.

So we came together in 2020, and one of the key things that we're really doing is we're building the work of The Dominica Sperm Whale Project, which for 15 years had been working in Dominica with about ... there there's groups of probably around 200 to 400, mainly female sperm whales that live in matrilineal societies. In the course of the last 15 years, they began to kind of see things like dialects that are emerging, almost like accents among the different groups of whales, so that gave us some foundation to build this study.

This study is actually not really using big data, it's using a data set of about 9,000 codas, and almost everybody on the team has not really worked with whales before, and they're coming at it with a new set of lens. So this lens was really coming at it from a natural language processing lens, and they were able to kind of see some of the core building blocks and fundamentals that are going into the sperm whales' communication system.

Dana Taylor:

What is a coda? You said there are about 9,000 of them.

David Gruber:

Codas are ... if you listen to whales, they make these stereotypical sounds. In Dominica, the ones that they use is called the 1-1-3, so it's like. That is a coda, and you'll hear them kind of repeat this over and over again. I would try to do it with my mouth, but I don't want to embarrass myself with trying to make whale clicks.

So that is what a coda, and if you're underwater with them, you'll just hear this amazing clicking sounds. But that'll be the one thing that really the human ear will pick out, you'll really notice the 1-1-3, because the Dominican sperm whales are saying that quite often.

Dana Taylor:

And how do codas make up the language?

David Gruber:

Well, so now with this paper, they thought there was about 20 different types of these types of codas in the Caribbean, but they're looking at more at the how quickly. So you might just hear, click, click, click, click, click, or it might be, click, click, click, click, click, and taking things of musical elements of tempo and rubato.

Then sometimes, every now and then if you're listening, you'll hear just an extra click at the end, like click, click, click, click, click, click, and that would be an ornament. So in this paper [inaudible 00:05:20] lead with our colleagues from MIT CSAIL, they were able to categorize these into what's known as a sperm whale phonetic alphabet, which opens up the possibility for much more communication than we previously thought.

Dana Taylor:

How has this new language model deepened your understanding of how they communicate?

David Gruber:

I think honestly, one of the real concerns, or as we embarked on this project, we're putting so much investment onto one animal onto the sperm whale. There's so many reasons why we're so deeply invested in the sperm whale, and it's an animal of superlatives with the largest brain that we know of, live in these very complicated family structures. One of the worries would be is that, well, what if their communication system is rather boring or there's not much to it? That they're just saying the same things over and over again, and I think what we're starting to see now is there's a lot more complexity.

As we begin this kind of further ascent to continue to study them and continue to learn more, we're essentially baby sperm whales at the moment, and we're trying to learn the communication system from the family from the ground up. This paper again was based on about 9,000 codas, as we begin to get nine million or nine billion, we'll really be able to get much more resolution into their communication system.

Dana Taylor:

David, are there categories of messages you've been able to decipher? Are there messages of danger and messages of caring, for example?

David Gruber:

That's a really good question. We're not yet into that. That's going to be coming out soon, and we're going to have an in to some of the messages. This paper really didn't get into what they're saying. It more got into the possibilities and the foundation, but you're getting to the idea of the first things that we'll be able to get to would be things that are quite evident. Like for instance, their diving behavior. They're communicating, there's clicks going, and then there might be an event where they're diving where we would be able to see that information because we would have information on the whale. We might have a tag on it that would have the depth on it. We would know which whale was around that. So those would be the first types of events that we'd be able to find. We have another paper coming out.

It's looking at using unsupervised machine translation, and there was two for understanding sperm whale communication. And there was two real takeaways from that, is that the more, well, one, it was that if their communication system is complex, these tools that we're seeing now, these tools, like large language models that you're seeing in ChatGPT, they'll be better at translating if the communication system is more complex. But the other thing would be how much of our world and their world are in common? So those would also be the other parts that would be easier to translate. So as we continue on this journey, parts where our world and their world intersect will be some of the other things that we'll be looking at. And it'll get really interesting when we get to the points of like, where is it where our world and the sperm whale world has almost nothing in common?

Dana Taylor:

Well, this is a revolutionary breakthrough. It's not the first time scientists have studied sperm whale communication. What's the history here?

David Gruber:

Even getting to when humans even knew that sperm whales were making sounds is, there's a paper in 1957, so just really thinking of it's not been that long that we even understood that sperm whales were vocalizing. So the fact that we're here in 2024 with this sperm whale phonetic alphabet and really understanding the core foundation, it's actually moved quite quick in a way. And I think what we're really excited about is just the possibilities. We've seen so much in the last few years of how advanced machine learning and artificial intelligence has really transformed humans and how humans are communicating. So it's now beginning to apply some of these new tools that we had to the non-human, which really get me excited.

Dana Taylor:

Do you think this development and understanding will help the conservation movement?

David Gruber:

Yeah, we hope so. I mean, we think about that all the time. After almost every meeting, we reflect back on how does this work assist whales, or how is this to the benefit of whales? And unfortunately, in the last year, we lost our key science advisor, Roger Payne and his work in the 1960s showed that whales sing. Along with Roger and Katie Payne, really with their musical background in a similar way that this study is done, by listening to humpback whales through their musical ears, they were able to decipher the whale song, which has now essentially led to this album and a paper in science on songs of the humpback whale. And that essentially became the anthem of the Save the Whales movement, which is considered one of the most successful environmental movements of all time.

It saved several species from extinction, and it stopped large scale Whaling led to the Marine Mammal Protection Act. So really the hope here is that, that was just by knowing that they sing. Imagine if we could really know what they're saying, how would that benefit the whales? Would it lead to additional protections? Would it allow us to understand how our human generated noises are affecting their habitat, and be able to really understand that on a deeper level. So again, we kind of feel like we're building a nice foundation now, which we could build these future studies upon. But the hope is that this is all in the benefit and the service of whales.

Dana Taylor:

It's a truly fascinating development David. What's next for you and your team? Where do you go from here?

David Gruber:

Thanks, Dana. Well, we've got a lot of work. Again, we were so fortunate, in last July we witnessed a sperm whale birth, and we've been working on this nonstop, developing new tools to analyze that. There were some new types of codas that we heard during this event. So that is something that's really taking our interest. Yeah, we're beginning the development of tools. How are we listening to the whales? We have a core whale listening station in Dominica. There are tags that go on the whales. Our group there just came up with the Harvard Micro-robotics lab had invented a new type of suction cup that's bio-inspired by sucker fish that delicately stays on the whales. Yeah, there's just so much to do. And in a project like this, there's really not a store that you can go to and buy deciphering sperm whale communication tools. So a lot of these things are, were being custom-made for this project.

Dana Taylor:

Well, it's all incredible. Thank you so much for being on The Excerpt, David.

David Gruber:

My pleasure. Thanks for having me Dana.

Dana Taylor:

Thanks to our senior producer Shannon Rae Green for production assistance. Our executive producer is Laura Beatty. Let us know what you think of this episode by sending a note to [email protected]. Thanks for listening, I'm Dana Taylor. Taylor Wilson will be back tomorrow morning with another episode of The Excerpt.