Imagine a future where you could have real heart-to-hearts with your pets. This future is near, … [+] thanks to the unprecedented boom in machine learning technology.
gettyDog owners will attest that our four-legged besties are highly expressive creatures. For many of us, coming back home from a long day at work involves wagging tails, happy jumps or a series of playful barks and howls from their dogs. While the sentiment of “I’m happy you are back” is conveyed, the nuances of the conversation are, naturally, lost in translation.
The same can be said for human-animal interactions in the wild. Consider a hypothetical scenario in which you encounter a chimpanzee in its natural habitat. Suddenly, it begins making loud, threatening noises and displaying aggressive gestures. While we understand these signals as hostile based on the chimp’s body language and vocalizations, the finer points of what it’s expressing—whether it’s fear, a territorial warning, a question or a direct threat—remain unclear.
What if there was a way to “talk” yourself out of the situation, much like you might attempt to de-escalate a confrontation with another human?
Advancements in machine learning are beginning to make such scenarios conceivable. There are now tools that are designed to sift through vast datasets of animal sounds and actions to create a proto-dictionary of animal language. Here are two situations where we’ve successfully managed to decipher what animals are saying.
1. Sperm Whale Language Is Being Codified
Sperm whales, which are highly social animals, use a sophisticated system of clicking sounds called “codas” to communicate with each other, much like humans use Morse code. These sounds are produced when these whales shoot air out of their nasal passages.
Although we were able to accurately tie distinct codas to the actual sperm whale making them, most aspects of their communication system were still not understood—that is, until a recent study published in Nature Communications used artificial intelligence to comb through close to 9,000 codas and make sense of the clicks.
Specifically, the study discovered new characteristics of codas, which they’ve termed “rubato” and “ornamentation,” that change based on the conversation’s context and are consistently used and mimicked among whales. The researchers found that these codas use a system that mixes rubato and ornamentation with steady features—rhythm and tempo—to create a wide range of distinct codas.
By analyzing these features using AI brainpower, the researchers have proposed a “Sperm Whale Phonetic Alphabet,” akin to our very own International Phonetic Alphabet.
“Like the International Phonetic Alphabet for human languages, this ‘Sperm Whale Phonetic Alphabet’ shows how a small set of axes of variation (place of articulation, manner of articulation and voicedness in humans; rhythm, tempo, ornamentation, and rubato in sperm whales) give rise to the diverse set of observed phonemes (in humans) or codas (in sperm whales),” the researchers explain.
2. ‘DeepSqueak’ Is Something Like Google Translate For Rodent Noises
Rodents, including mice and rats, communicate using a language of high-pitched ultrasonic vocalizations (USVs). These vocalizations are often at frequencies above the range of human hearing.
Rats communicate through various ultrasonic calls: 50 kHz sounds, often likened to laughter, emerge … [+] during joyful moments such as play or courtship, and even in the exhilaration from a drug effect. Conversely, their 22 kHz vocalizations signal distress, reflecting feelings of pain or illness.
gettyHumans can typically hear sounds up to about 20 kHz, while these rodent vocalizations can range from 20 kHz to over 100 kHz, making them inaudible to us without specialized equipment. Studying these sounds, although noninvasive, were considered prohibitively expensive and laborious until DeepSqueak, a machine learning tool, was developed in 2018 to unravel these high-pitched squeaks.
Speaking to DW, the lead researcher behind DeepSqueak, Kevin Coffey says, “AI and deep-learning tools are not magic. They are not going to suddenly translate all animal sounds into English. The hard work is being done by biologists who need to observe animals in a multitude of situations and connect the calls to behaviors, emotions, etcetera.”
What the software does is streamline the process of research by isolating rodent calls from raw audio, matching them with similar vocalizations and delivering insights into the behaviors of these animals. It is essentially a sophisticated interpreter, decoding the ultrasonic language of rodents into data that researchers can analyze.
Examples like these, where we are starting to understand animal communication, are increasingly common, illustrating the rapid evolution of our technological capabilities. Though still in its early stages, the feasibility of human-animal communication—both theoretically and practically—is becoming clear.
Today, we might not yet chat freely with our pets, but consider this: just a few years ago, advanced AI tools like ChatGPT seemed like futuristic dreams. Now, they’re part of our everyday lives. This swift progress hints that meaningful dialogues with animals might soon be within our reach.
{Categories} _Category: Applications,*ALL*{/Categories}
{URL}https://www.forbes.com/sites/scotttravers/2024/05/15/pet-translator-apps-were-closer-to-decoding-animal-language-than-you-think/{/URL}
{Author}Scott Travers, Contributor{/Author}
{Image}https://imageio.forbes.com/specials-images/imageserve/66437440926cc67c8e989ede/0x0.jpg?format=jpg&crop=3268,1533,×0,y409,safe&height=600&width=1200&fit=bounds{/Image}
{Keywords}Science,/science,Innovation,/innovation,Science,/science,standard{/Keywords}
{Source}POV{/Source}
{Thumb}{/Thumb}