AI Discovers Lions Have Two Distinct Roars, Revolutionizing Wildlife Conservation Monitoring

Lions have a second roar that no one noticed until now

A closer listen to lion calls may help map where the big cat is under threat The thunderous roar of the MGM lion that has opened Hollywood films for nearly a century has conditioned us to hear the big cat's call as a blunt declaration: a booming blast announcing power and presence. But the real soundscape of a lion pride is far more intricate than that cinematic caricature, researchers report November 20 in the journal Ecology and Evolution. Using field recordings from Africa and machine learning techniques to analyze the acoustics, scientists found that African lions (Panthera leo) produce two distinct types of roars: the familiar, guttural one that anchors a roaring bout -- and carries vocal signatures unique to each animal -- plus an overlooked "intermediary" roar that is shorter and lower-pitched than the classic full-throated version. The findings challenge decades-old assumptions. Biologists have long known that a lion's roar helps advertise territory, attract mates and locate pride members -- and that a complete roaring bout begins with moans and ends with grunts. But everything in the middle was treated as a single, undifferentiated roar. Now, by decoding that roar into its component parts, and with artificial intelligence trained to tell one lion's voice from the next, conservation groups may be able to count and track lions by sound alone. "If you can identify a lion by its roar, this could potentially be a tool to count the number of individuals within a landscape," says Jonathan Growcott, a conservation technologist and large carnivore biologist at the University of Exeter in England. Such insights could prove especially valuable at a time of shrinking habitat and poaching pressures, when lions have vanished from more than 90 percent of their historic range, he adds. Still, what the newly identified intermediary roar actually communicates remains unclear. "We don't know yet," Growcott says. "Unfortunately, we don't speak lion. There is no option of 'lion' on Duolingo." The discovery emerged from tens of thousands of hours of audio captured by remote recorders in Tanzania's Nyerere National Park and by acoustic collars fitted to lions in Zimbabwe. When Growcott's team ran more than 3,000 calls through pattern-recognition algorithms, subtle differences jumped out. Full-throated roars traced a clear arc, rising in pitch before ending in a trailing fall, while intermediary roars were flatter and far less elaborate. Focusing on how long an utterance lasted and how high in pitch it climbed, the researchers could then build an algorithm capable of classifying each type of roar, moan and grunt with high precision. In at least one lion population, accuracy topped 91 percent. By parsing roar types and pulling out the more informative full-throated call, the tool even identified which individual lion was roaring, outperforming human experts. According to Tanya Berger-Wolf, a computational ecologist at Ohio State University in Columbus, this is one of the first clear demonstrations that machine learning can reliably interpret the vocalizations of a mammal. "It is a good example of bioacoustic monitoring beyond birds, amphibians and insects," she says. But because the recordings lacked behavioral context, scientists still can't say why lions choose one roar type over the other -- an open question that intrigues lion experts like Craig Packer of the University of Minnesota in St. Paul. "It would be interesting to have enough recordings in known contexts to know if lions roar more loudly in certain situations," he says. As for the MGM lion, his iconic roar contains no hidden intermediary for one simple reason: It doesn't belong to a lion at all. In a bit of Hollywood movie magic, sound designers opted for something even more ferocious, Growcott says. "The MGM lion is actually a tiger."

Lions have two types of roar - new research

University of Exeter provides funding as a member of The Conversation UK. The roar of an African lion is one of the most iconic sounds of the animal kingdom. However, my new research suggests it should actually be separated into two distinct vocalisations: the full-throated roar, and an "intermediary roar" with a flatter, less varied sound. Making this distinction could have important implications for lions' conservation. The total population of wild lions in Africa is estimated to be between 22,000 and 25,000, but this number is half what it was 25 years ago. The main drivers of this decline are habitat loss and fragmentation, reduction in prey, and conflict with local people. According to the International Union for Conservation of Nature red list, lions are now vulnerable to extinction. My colleagues and I investigated roaring in lions to get better at distinguishing between their different vocalisations. But our findings may make it easier to monitor lions' numbers, which in turn would make it easier to protect them. You might think you know a lion's roar from the clip used by MGM at the start of all its films - but that isn't quite right. It's actually a tiger's roar dubbed on top of this famous piece of cinema. Compared with a lion, a tiger's roar is often raspier and higher-pitched. In fact, male and female lions produce what scientists call a "roaring bout". Each begins with a series of soft moans, followed by a subsection of intermediary and full-throated roars, which finally subside into a repetition of grunts. There is no set length of time a roaring bout will last (though most are between 30 and 45 seconds) and the number of vocalisations within each subsection does not keep to a strict formula. The roaring bout is important behaviour. Not only does it signal to other lions in their pride where they are, but to unfriendly lions, bouts can be used to advertise territorial boundaries. The loudest, most complex component of a lion's roaring bout is the full-throated roar, which is an individually identifiable sound. Each lion's full-throated roar is as specific to the individual as the pattern of spots are to a leopard (and as my 2024 paper found, as their roar is too). Population density estimates are a key metric for identifying priority areas for conservation. If individual lions can be identified by their full-throated roars, then researchers could use this to count them. However, picking out the full-throated roars from other vocalisations within a roaring bout is tricky. Even for those with expert ears, it is a subjective process which is prone to human bias. The reason becomes clearer when you look at a spectrogram of a lion's roaring bout - a visual representation of its sounds using an x-axis of time (seconds) and y-axis of frequency (hertz). The full-throated roar at the start of the mid-section of the bout rarely looks or sounds the same as the roar that occurs right before the grunts kick in. Which made me wonder: should these different roars be classified the same? My colleagues and I leaned on AI to help us analyse our roar recordings. Perhaps this could help solve the issue of subjectivity, we thought, and classify lion vocalisations automatically, creating a tool so that other researchers always know which roar is right for counting lions. We used supervised machine learning to classify the vocalisations which occur in a lion's roaring bout into three call types: full-throated roars, grunts, and our newly identified intermediary roar. From the spectrogram, we could see that the full-throated roar is loud, complex and arcs in pitch. The intermediary roar was a flatter sound with less variation - and it always followed the full-throated roars. Grunts were shorter and even more compact. Using simple acoustic parameters - the duration of each vocalisation and its maximum frequency - we could then identify each call type with an accuracy of 95.4%. As the full-throated roars are unique to each individual lion, we wanted to test whether our AI analysis of full-throated roars was better at distinguishing between different lions than human hearing. We found we could identify individual lions at an accuracy of 94.3% - an improvement of 2.2% over when human-selected full-throated roars were used. Using this technique for identifying full-throated roars could hopefully lead to more accurate population density estimates of lions. Read more: Lions are still being farmed in South Africa for hunters and tourism - they shouldn't be It is exciting to discover the language of lions is more complex than previously thought. However, it is unclear what the communicative differences of the two roar types may be. Scientists have long believed that lion roars may convey information relating to pride size, age and identity - but without Dr Doolittle to translate the meaning of moans, grunts and roars, this is still guesswork. Therefore, it may take some time before "lion" appears as an option on Duolingo. For now, we should just celebrate the fact that AI can help us to discover more about wild phenomena as iconic as a lion's roar.

AI detects a secret lion roar no one knew existed

A recent investigation has revealed that African lions use two separate kinds of roars, not just one. This finding is expected to play an important role in improving how conservation groups track and study these big cats. Researchers at the University of Exeter uncovered a previously unrecognized "intermediary roar" that appears alongside the well-known full-throated version. The study, published in Ecology and Evolution, is the first to apply artificial intelligence to automatically sort lion roars into different types. The system reached a 95.4 per cent accuracy rate and greatly reduced the influence of human interpretation, allowing for more consistent identification of individual lions. Lead author Jonathan Growcott of the University of Exeter said: "Lion roars are not just iconic -- they are unique signatures that can be used to estimate population sizes and monitor individual animals. Until now, identifying these roars relied heavily on expert judgment, introducing potential human bias. Our new approach using AI promises more accurate and less subjective monitoring, which is crucial for conservationists working to protect dwindling lion populations." Lion Numbers Continue to Fall The International Union for Conservation of Nature red list categorizes lions as vulnerable to extinction. Current estimates suggest Africa holds only 20,000 to 25,000 wild lions, and this population has dropped by about half over the last quarter century. The new study concludes that a lion's roaring sequence includes both the established full-throated roar and the intermediary version, overturning the long-standing assumption that only one roar type existed. Similar developments have been reported in research on other large carnivores, including spotted hyaenas, and reinforce the expanding value of bioacoustics in ecological science. AI Improves Monitoring Accuracy By applying machine learning to classify full-throated roars, the research team advanced the ability to distinguish individual lions. The automated, data-focused method also streamlines passive acoustic monitoring, offering a more dependable and accessible option than common techniques such as spoor surveys or camera trapping. Jonathan Growcott added: "We believe there needs to be a paradigm shift in wildlife monitoring and a large-scale change to using passive acoustic techniques. As bioacoustics improve, they'll be vital for the effective conservation of lions and other threatened species." Broad Collaboration Supports New Findings The project was carried out by the University of Exeter in partnership with the Wildlife Conservation Unit at the University of Oxford, Lion Landscapes, Frankfurt Zoological Society, TAWIRI (Tanzania Wildlife Institute for Research) and TANAPA (Tanzania National Parks Authority). Computer scientists from Exeter and Oxford also contributed to the work. Funding came from the Lion Recovery Fund, WWF Germany, the Darwin Initiative, and the UKRI AI Centre for Doctoral Training in Environmental Intelligence.

AI Reveals Lions Don't Just Roar, They Have a Second Call - Neuroscience News

Summary: A new study reveals African lions produce two types of roars, overturning long-held assumptions and opening the door to more precise wildlife monitoring. Using machine learning, researchers automatically distinguished between full-throated and newly identified intermediary roars with over 95% accuracy, eliminating much of the human bias in vocal identification. This breakthrough greatly enhances conservation efforts by enabling reliable, noninvasive population tracking. As lion numbers continue to drop across Africa, AI-driven bioacoustics may become a critical tool for protecting vulnerable big-cat populations. A new study has found African lions produce not one, but two distinct types of roars - a discovery set to transform wildlife monitoring and conservation efforts. Researchers at the University of Exeter have identified a previously unclassified "intermediary roar" alongside the famous full-throated roar. The study, published in Ecology and Evolution, used artificial intelligence to automatically differentiate between lion roars for the first time. This new approach had a 95.4 per cent accuracy and significantly reduced human bias to improve the identification of individual lions. Lead author Jonathan Growcott from the University of Exeter said: "Lion roars are not just iconic - they are unique signatures that can be used to estimate population sizes and monitor individual animals. Until now, identifying these roars relied heavily on expert judgment, introducing potential human bias. "Our new approach using AI promises more accurate and less subjective monitoring, which is crucial for conservationists working to protect dwindling lion populations." According to the International Union for Conservation of Nature red list, lions are listed as vulnerable to extinction. The total population of wild lions in Africa is estimated to be between 20,000 and 25,000, but this number has decreased by half in the last 25 years. The study establishes that a lion's roaring bout contains both a full-throated roar and a newly named intermediary roar, challenging the long-held belief that only one roar type existed. These findings echo similar advances in the study of other large carnivores, such as spotted hyaenas, and highlight the growing potential of bioacoustics in ecological research. Researchers used advanced machine learning techniques and by implementing this automated, data-driven approach to classify full-throated roars, the team improved the ability to distinguish individual lions. The new process simplifies passive acoustic monitoring, making it more accessible and reliable compared to traditional methods like camera traps or spoor surveys. Jonathan Growcott continued: "We believe there needs to be a paradigm shift in wildlife monitoring and a large-scale change to using passive acoustic techniques. As bioacoustics improve, they'll be vital for the effective conservation of lions and other threatened species." The research was a collaborative effort between the University of Exeter, the Wildlife Conservation Unit at the University of Oxford, Lion Landscapes, Frankfurt Zoological Society, TAWIRI (Tanzania Wildlife Institute for Research) and TANAPA (Tanzania National Parks Authority), as well as computer scientists from Exeter and Oxford. Funding: The work was supported by the Lion Recovery Fund, WWF Germany, the Darwin Initiative, and the UKRI AI Centre for Doctoral Training in Environmental Intelligence. Roar Data: Redefining a lion's roar using machine learning For territorial advertisement and intra-pride communication African lions emit a roaring bout, of which one component, is their iconic roar. The full-throated roar of a lion has recently been shown to be a unique and individually identifiable signature. At the same time, the frequency of large-scale passive acoustic monitoring surveys has increased. As such, a lion's roar may soon become a useful tool to count individuals and estimate population density, to supplement traditional survey techniques. Currently, selecting full-throated roars is heavily dependent on expert inference and is therefore subject to human-induced bias. We propose a data-driven approach to automatically classify lions' full-throated roars from the other vocalisations that constitute a roaring bout. By using two-state Gaussian Hidden-Markov Models, we also demonstrate that two types of roars exist within a lion's roaring bout -- a full-throated roar and a newly named intermediary roar -- and these can be classified at an accuracy of 84.7%. We further demonstrate that using simple metrics to describe lion vocalisations -- maximum frequency (Hz) and vocalisation length (s) -- and K-means clustering is sufficient to classify lion call types, at a high accuracy (95.4%), and that using data-driven predicted full-throated roars results in an improved ability to identify individuals (F1-score 0.87 vs. manual full-throated roar classification 0.80). Here, we establish an easy-to-understand and implement process that will reduce the knowledge gap and make passive acoustic monitoring more accessible in a field currently dominated by other monitoring techniques (e.g., camera surveys), paving the way for novel research.

Lions have a secret roar that was just uncovered by AI

Across the savanna, a lion's roar breaks the night silence. The sound is unmistakable, yet the deeper structure behind it has gone unnoticed for decades. A new study now reveals clear layers within each roaring sequence. The research arrives at an important moment, as conservation teams search for better ways to track a species under pressure. The researchers identified four sound types within a roaring bout. A lion begins with soft moans, moves into full throated roars, shifts to a shorter intermediary roar, and ends with grunts. Earlier thinking focused on one roar only. The new investigation shows a second roar type with a shorter rise, a quicker fall, and a lower maximum frequency. Its repeated position within each sequence signals a defined role rather than a weakened roar. "Lion roars are not just iconic - they are unique signatures that can be used to estimate population sizes and monitor individual animals," said Jonathan Growcott from the University of Exeter. "Until now, identifying these roars relied heavily on expert judgment, introducing potential human bias. Our new approach using AI promises more accurate and less subjective monitoring, which is crucial for conservationists working to protect dwindling lion populations." Roaring behavior changes with age, social position, and daily rhythms. Younger individuals call less. Males without a pride avoid calling because loud announcements may draw rivals. Calling peaks before dawn when sound travels far. Water sources attract more vocal activity. Females do call, though new mothers often stay quiet to avoid drawing attention to cubs. Every choice reflects risk, territory, and pride connections. Spectrograms in the study show clear contrasts between each sound type. Full throated roars stretch longer and reach higher frequencies. Intermediary roars rise and fall quickly. Moans have softer, uneven contours at the start of the bout. Grunts finish the sequence with brief, low notes. Consistent placement of the intermediary roar confirms its identity as a distinct call. The pattern repeats across many samples. Hidden Markov Models helped classify sound types by following the movement of the fundamental frequency over time. This approach handled the structure well. A simpler method also worked: K means clustering based on duration and maximum frequency. Once moans were removed, classification accuracy rose sharply. The simpler workflow avoids heavy computational needs and allows wider use by field teams. Recordings from collared lions in Zimbabwe helped test the system further. AI based classification improved identification of individuals. Automatic selection captured more usable full throated roars and matched those calls to specific lions with higher accuracy. Manual spotting still handles moans because moans always appear at the start and have an unmistakable shape. Research on spotted hyaenas revealed multiple whoop types inside a single calling sequence. The lion study follows the same pattern of expanded understanding. Many large carnivores rely on layered sequences rather than isolated calls. Hidden acoustic structure appears more common than earlier work suggested. Passive acoustic monitoring offers strong advantages for wide landscapes. Camera traps miss animals in dense or rugged areas. Sound sensors detect calls over long distances. With reliable classification, acoustic surveys can produce steadier population estimates and reduce human bias in data processing. "We believe there needs to be a paradigm shift in wildlife monitoring and a large-scale change to using passive acoustic techniques. As bioacoustics improve, they'll be vital for the effective conservation of lions and other threatened species," said Growcott. Frequency and duration varied among lions from Tanzania, Zimbabwe, and Botswana. One male from Botswana produced unusually short roars. Geographic origins may influence vocal traits. Nomadic males often travel huge distances, so understanding regional variation will support accurate classification across borders. The researchers designed a straightforward system. It avoids deep learning models that require massive datasets. Lions do not vocalize often enough to support such systems. The chosen workflow uses accessible techniques, making adoption easier for conservation workers. The roar no longer stands as a single towering sound. It exists as a sequence with distinct stages and valuable information. With stronger tools, researchers can track individuals, estimate populations, and understand behavior more clearly. Improved listening may help protect lions as pressures on wild populations continue to rise. Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates.