The Race to Translate Animal Sounds Into Human Language

In 2025 we will see AI and machine learning make real advances in understanding animal communication, answering a question that has puzzled humans for as long as we have existed: “What are animals saying to each other?” recent Caller-Dolittle PrizeCash prizes of up to half-a-million dollars for scientists who “crack the code” are a strong indication of the confidence that recent technological developments in machine learning and large language models (LLMs) are putting this goal within our grasp.

Many research groups have been working for years on algorithms to understand animal sounds. Project Ceti, for example, has been decoding Click on the train of the sperm whale and the song of the humpback. These modern machine learning tools require extremely large amounts of data, and until now such high-quality and well-annotated data have been lacking.

Consider LLMs like ChatGPT that have training data available that includes all the texts available on the internet. Such information on animal communication was not accessible in the past. It’s not just that human data corpora are orders of magnitude larger than the kind of data we have access to for wild animals: more than 500 gigabytes of words were used to train GPT-3, compared to just over 8,000 “codas” for the sperm whale communication project Ceti. For recent analysis of (or voices).

Furthermore, when dealing with human language, we already i know which is being said. We even know what constitutes “sound,” which is a huge advantage in interpreting animal communication, where scientists rarely know whether a particular wolf’s howl, for example, is something different from another wolf’s howl, or even whether wolves perceive a howl as a howl. . Somehow analogous to a “word” in human language.

Still, 2025 will bring new advances, both in the amount of animal communication data available to scientists and in the type and power of AI algorithms that can be applied to that data. With low-cost recording devices such as the Audiomoth exploding in popularity, automatic recording of animal sounds has been placed within easy reach of every scientific research group.

Huge datasets are now coming online, as recorders can be left in the field, listening to gibbon calls in the forest or birds in the forest, 24/7, for long periods of time. There were times when it was impossible to handle such huge datasets manually. Now, new automatic detection algorithms based on convolutional neural networks can run through thousands of hours of recordings, pick out animal sounds and cluster them into different types according to their natural acoustic properties.

Once those large animal datasets become available, new analytical algorithms become a possibility, such as using deep neural networks to find hidden structures within animal vocal sequences, which may resemble meaningful structures in human language.

However, the fundamental question that remains unclear is, what exactly are we expected to do with these animal sounds? Some organizations, such as Interspecies.io, set their goals quite clearly as, “converting signals from one species into coherent signals for other species.” In other words, from Translation Animal communication in human language. Yet most scientists agree that non-human animals do not have a real language of their own—at least not in the way that we humans do.

The Kolar Dolittle prize is a bit more sophisticated, looking for a way “to communicate or decipher the communication of an organism.” Deciphering is a somewhat less ambitious goal than translation, considering the possibility that animals do not, in fact, have a language that can be translated. Today we do not know how much information, or how little, animals express among themselves. In 2025, humanity will likely leapfrog our understanding of not just how much animals are saying but what exactly they are saying to each other.