The race to translate animal sounds into human language

In 2025, we will see artificial intelligence and machine learning being leveraged to make real progress in understanding animal communication, answering a question that has perplexed humans for as long as we have existed: “What do animals say to each other?” The last Kohler-Doolittle AwardOffering cash prizes of up to half a million dollars to scientists who “crack the code” is an indication of growing confidence that recent technological advances in machine learning and large language models (LLMs) put this goal within our reach.

Many research groups have been working on algorithms to understand animal sounds for years. For example, the Ceti project decrypted Click sperm whale trains and humpback songs. These modern machine learning tools require very large amounts of data, and until now, such amounts of high-quality, well-annotated data do not exist.

Consider LLMs like ChatGPT that have training data available to them that includes all the text available online. Such information about animal communication was not accessible in the past. It’s not just that the human dataset is much larger than the kind of data we have access to on animals in the wild: more than 500GB of words were used to train GPT-3, compared to just over 8,000 code words. “(or pronunciation) of Project Ceti’s recent analysis of sperm whale communications.

Additionally, when working with human language, we actually… He knows What is said. We even know what constitutes a “word,” which is a huge advantage over explaining communication between animals, where scientists rarely know whether a particular wolf’s howl, for example, means something different from another wolf’s howl, or even whether wolves are considered Howling is like a howl. It’s a bit like a “word” in human language.

However, 2025 will bring new developments, both in the amount of animal communication data available to scientists, and in the types and power of AI algorithms that can be applied to that data. Automated recording of animal sounds has become within the reach of every scientific research group, with the growing popularity of low-cost recording devices such as the AudioMoth.

Huge data sets are now available online, where recorders can be left in the field, to listen to the calls of gibbons in the woods or birds in the woods, 24 hours a day, 7 days a week, over long periods of time. There have been occasions when it was impossible to manage large data sets manually. Now, new automatic detection algorithms based on convolutional neural networks can race through thousands of hours of recordings, picking up animal sounds and grouping them into different types, according to their natural vocal characteristics.

Once these large animal datasets become available, new analytical algorithms become possible, such as using deep neural networks to find hidden structure in sequences of animal sounds, which may be analogous to the meaningful structure in human language.

However, the fundamental question that remains unclear is: what exactly do we hope to do with these animal sounds? Some organizations, such as Interspecies.io, have quite clearly defined their goal as “transforming signals from one species into coherent signals for another.” In other words, l Translate Animal communication with human language. However, most scientists agree that non-human animals do not have an actual language of their own, at least not in the way that we humans have one.

The Coller Dolittle Prize is a bit more complex, looking for a way to “communicate with a living being or decipher its communications.” Decoding is a slightly less ambitious goal than translation, considering the possibility that animals may not actually have a language that can be translated. Today, we don’t know how much, or how little, information animals transmit to each other. In 2025, humanity will have the potential to transcend our understanding of not only what animals say but also exactly what they say to each other.