A Short History of Bioacoustics
Edited by Sam Suren, RFCx Staff | Published on 01/JUL/2026 - 10:36 A.M EST
How a submarine-hunting surveillance network, built for Cold War military purposes, turned out to be one of the most consequential whale-research instruments ever assembled, almost entirely by accident.
Credits for cover photo go in here.
Sometime in the 1950s, engineers monitoring a classified U.S. Navy hydrophone network built to track Soviet submarines kept picking up a low, mournful sound they couldn't identify. They nicknamed it the "Jezebel Monster," after the project's codename. It took years to work out what the monster actually was: blue and fin whales, calling to each other across vast stretches of open ocean through a deep ocean layer that channels low-frequency sound with very little loss of energy, a phenomenon discovered by the Woods Hole scientists Maurice Ewing and J. Worzel in 1944.
The Navy's listening stations weren't built with whales in mind. They existed, at enormous expense, to detect the faint mechanical signature of enemy submarines. But at one of those secret stations off Bermuda, an engineer named Frank Watlington had also recorded something even stranger than the Jezebel Monster: hours of humpback whales singing in long, structured, repeating phrases. Watlington had made the recordings years earlier (accounts differ on exactly when, somewhere between the late 1950s and 1964), and in 1966 he passed them to a young zoologist named Roger Payne, who heard immediately what the Navy's own sonar technicians had missed. It wasn't noise. It was song. Payne and a colleague, Scott McVay, published their analysis in Science in 1971. A year earlier, Payne had already turned a selection of the recordings into an album, Songs of the Humpback Whale, which went on to become, according to Guinness World Records, the best-selling nature album ever made, with more than 125,000 copies sold. In January 1979, National Geographic pressed 10.5 million flexi-disc copies of humpback song into a single issue of the magazine - reportedly the largest single vinyl pressing in recording history. The songs later flew into interstellar space aboard the Voyager Golden Record, and they're widely credited with helping build the public support behind the 1972 U.S. Marine Mammal Protection Act and the International Whaling Commission's 1986 moratorium on commercial whaling. That accident is, in miniature, the whole plot of bioacoustics - the study of how animals make, transmit, and perceive sound. For more than two centuries, the field has advanced in the same two steps, repeated with different hardware each time: build a machine that can hear something humans can't, then discover that the natural world had been talking the whole time.
Credits for cover photo go in here.
Discovering Sounds Beyond Human Hearing
The roots of bioacoustics stretch back to 1793, when Italian priest and physiologist Lazzaro Spallanzani carried out experiments in which he deliberately damaged the eyes of an owl and several bats in an effort to understand how they navigated without sight. The owl became helpless in the dark, but the bats continued flying with remarkable accuracy, even avoiding bells strung on wires across a room. When he blocked the bats’ ears, however, they began crashing into objects. Unable to explain what he had observed, Spallanzani reluctantly proposed that bats might navigate using an unknown “sixth sense.” The eminent French anatomist Georges Cuvier disagreed, confidently claiming that bats simply had an unusually refined sense of touch - a mistaken explanation that went largely unchallenged for the next 145 years.
The mystery wasn't solved until 1938, when a Harvard undergraduate named Donald Griffin brought a cage of bats into the laboratory of the physicist G.W. Pierce, who had built a device capable of detecting sound at frequencies far above human hearing. The bats Spallanzani had described as flying in silence turned out to be anything but quiet: they were screaming constantly, just at a pitch no human ear could register. Working with Robert Galambos, Griffin confirmed that bats navigate by echolocation, emitting ultrasonic calls and reading the echoes that bounce back, and coined that term in 1944. Even then, the science proceeded cautiously: the pair's first paper on the subject, published in 1938, stopped short of connecting the ultrasonic calls to navigation at all, noting only that the sounds had been detected. It's a small, telling detail: even after building a machine that can finally hear the answer, it can take a while to believe it.
A related instrument reshaped the field just as much, though it wasn't designed with animals in mind at all. During the Second World War, Bell Labs engineers developed the sound spectrograph as classified military technology, used to analyze the acoustic signatures of submarines and help decode enemy voice transmissions. Declassified after the war and published in 1946–47 as Visible Speech, the device translated any sound into a two-dimensional image plotting frequency against time. Commercialized as the "Sono-Graph" by Kay Elemetrics, it gave ornithologists their first way to see birdsong on paper, and in one form or another, the spectrogram remains the basic unit of bioacoustic analysis today, whether the subject is a sparrow or a sperm whale.
Credits for cover photo go in here.
How Animals Talk: Dances, Rumbles, and Warning Calls
Photo: prarie dogs and predators
Not every breakthrough depended on new technology. Sometimes, it came from careful observation and the courage to suggest an unusual idea.
In the first half of the 20th century, the Austrian zoologist Karl von Frisch documented that honeybees communicate the direction and distance of food to their hive-mates through a "waggle dance," a figure-eight routine performed at 13 to 15 wags per second and accompanied by roughly 280 Hz sounds produced by the bee's wings. The idea that an insect brain could manage something resembling symbolic communication struck many scientists as far-fetched, but von Frisch was eventually vindicated enough to win a share of the 1973 Nobel Prize in Physiology or Medicine. That didn't end the argument: the entomologist Adrian Wenner spent much of the late 1960s insisting that bees actually navigate by smell rather than by watching each other dance, in a dispute researchers later nicknamed the "Bee Battles."
Elephants delivered an equally startling discovery, and this one began with a hunch rather than an instrument. In 1984, the bioacoustician Katy Payne, who had already spent fifteen years studying whale song, was standing near the elephant enclosure at Portland's Washington Park Zoo when she felt what she later described as a throbbing in the air, like the lowest note of a church organ. On the flight home, she guessed that elephants might be communicating below the range of human hearing. She was right: elephants produce rumbling calls between 14 and 35 Hz, under the roughly 20 Hz floor of human perception, and field recordings have since shown these infrasonic calls can travel up to 10 kilometers in good conditions, letting separated herds coordinate movement or warn each other of danger across distances no human ear could ever bridge.
A more contested case involves Gunnison's prairie dogs. Research led by the biologist Con Slobodchikoff has found that the animals' alarm calls vary depending on the specific predator approaching, distinguishing a coyote from a hawk from a domestic dog, and, more controversially, may even encode details about a human intruder's size, shape, and clothing color, all within about a tenth of a second. Playback experiments suggest prairie dogs do respond appropriately to different calls. Whether any of this amounts to language, though, remains genuinely disputed among scientists, and the more striking claims deserve some skepticism until they've been independently replicated.
Credits for cover photo go in here.
The machine-learning era
Hydrophone photo?
If the 20th century's breakthroughs came from analog instruments, the 21st century's are increasingly coming from machine learning applied to enormous archives of recorded sound.
The most ambitious example is Project CETI, an interdisciplinary nonprofit using hydrophone arrays, biologging tags, and machine learning to analyze the clicking patterns, or "codas," that sperm whales use to communicate — a field that barely existed a lifetime ago, since researchers didn't even know sperm whales made sound until the 1950s. In a 2024 study in Nature Communications, a team led by Sharma analyzed roughly 9,000 codas recorded off Dominica and identified 156 distinct patterns, built from a combination of rhythm and tempo layered with more subtle variations the researchers termed "rubato" and "ornamentation." That's a far larger repertoire than the 21 coda types previously assumed to make up the sperm whale's entire vocabulary. A separate study involving UC Berkeley researchers found structures in the codas resembling vowels and diphthongs. None of this amounts to translation, and CETI's own researchers have been careful to describe their findings as statistical structure rather than meaning. Whether sperm whales are saying anything in particular remains unknown.
Machine learning is also showing up in more tangible conservation work. Healthy coral reefs are loud, full of fish grunts, purrs, and the crackle of snapping shrimp, and coral and fish larvae appear to use that soundscape to decide where to settle. In a 2019 study in Nature Communications, researchers led by Gordon found that playing recordings of a healthy reef over underwater speakers at a degraded reef roughly doubled fish abundance and increased species richness by about 50 percent. A 2024 follow-up in Royal Society Open Science, led by Aoki, found that one coral species, Porites astreoides, settled at 1.7 times the rate (and up to seven times, in the best cases) at acoustically enriched sites compared with silent ones.
Plants, it turns out, have something to say too. In a 2023 study in Cell, researchers at Tel Aviv University recorded tomato and tobacco plants emitting ultrasonic clicks, not unlike the sound of popping bubble wrap, when they were dehydrated or physically cut. Stressed tomato plants popped roughly 35 times an hour, in a frequency range of 20 to 100 kilohertz, detectable from three to five meters away by anything capable of hearing that high. Mice and moths, among others, likely can. Humans, whose hearing tops out around 20 kilohertz, had no way of knowing any of this until someone finally pointed a classifier at the problem.
Credits for cover photo go in here.
The Future of Bioacoustics
Rfcx photo
The newest chapter of this story is also the most explicitly applied — less about discovering that something makes sound, and more about using that sound to act. At Rainforest Connection, we deploy solar-powered acoustic sensors, initially built from recycled smartphones and nicknamed "Guardian" devices, through forest canopies, using machine-learning models to identify the specific sound of a chainsaw or a gunshot in real time and alert rangers before illegal logging can do much damage.
“Add quote from Topher on bioacoustics future”
The underlying approach is grounded in genuine acoustic-monitoring research. But as with most conservation technology backed by nonprofits and vendors competing for funding, the specific performance figures — trees saved, carbon avoided — are worth treating as promotional claims until they're verified independently.
That caution feels like the right note to end on, because it's really been true of bioacoustics since Spallanzani first plugged a bat's ears. Roger Payne recognized song in recordings that a Navy engineer had simply logged as strange underwater sound. Griffin and Galambos initially hedged on what their own data meant. The question of whether bees, prairie dogs, or sperm whales are "communicating" in any rich sense is still unresolved, decades or centuries after each was first proposed. Each new listening device — the ultrasonic detector, the spectrograph, the hydrophone, and now the machine-learning model — hasn't just picked up new sounds. It has quietly redrawn the boundary of which creatures we consider worth listening to in the first place. More than two hundred years after a blinded bat first suggested that something might be there, we're still finding out how much we'd been missing.
References.
Meadowlands Environmental Research & Restoration Institute and New Jersey Sports and Exposition Authority (December 2024) https://njmc.s3.us-east-2.amazonaws.com/MRRI/pdfs/FinalReport_Dec2024.pdf
Michael J. Turso (2024) Using Passive Acoustic Monitoring to Assess the Distribution of a Rare Frog in the NJ Meadowlands Urban Naturalist, No. 73
Bloustein School of Planning & Public Policy (2023), Rising Tides, ArcGISStoryMaps
Arbimon (finish reference)
Stay in the loop.
Get more stories like this from the frontlines of conservation.
We respect your privacy. We also hate spam.
Listen to a live preview.
Download the app to access our free library of rainforest live streams and species calls.
