Study Uncovers New Way For Humans To Translate Animal Talk: With AI

A lion’s roar, the birds chirping in the tree, the neigh of a horse, and even down to the bark of your furry best friend could soon sound like conversations you’ll understand as clear as day.

The dream of being able to speak to every animal in nature has always been something that seems so unfathomable, but one team of scientists is ready to take on that dream and turn it into a reality. Researchers are looking at the effects of artificial intelligence and machine learning to understand and translate what wildlife is saying to each other and us.

A nonprofit called Earth Species Project (ESP) has delivered a roadmap for this distant future of cross-species communication. “We hypothesize that ML models can provide new insights into non-human communication, and believe that discoveries fueled by these new techniques have the potential to transform our relationship with the rest of nature,” said the organization in a blog post.

One of the first challenges in training the ML was that obtaining data from animals was hard to collect. As a result, the team mainly relies on a small group of scientists who examine certain species to help decode said collated data.
 
Furthermore, the model predictions can only be based on what is already understood about how wildlife communicates.
 
The ML ultimately will also help answer questions such as under which conditions wildlife makes these signals and how receivers of these signals behave. It also hopes to solve how communication strategies differ between species and populations. 

ESP is creating methods motivated by recent advancements in computer vision that highlight the signal variations crucial for ML-based categorization. The team is also experimenting with using a species’ vocalization to predict its motion. This could lead to a distinction between dialects between two different sets of test subjects.  

 
Aza Raskin, co-founder and president of ESP, noted in the first published study that the findings would not be limited to house pets but to all species, or at least as many as possible.

Regardless of the team’s monumental feat, some other roadblocks could get in their pursuit ahead as well. For one, there is the fact that wildlife, like humans, use non-verbal cues as well when communicating with one another. And just as it plays a huge part in how we converse with one another, so does it for animals.

Despite this, an experimental algorithm can already identify which animal in a noisy group is actually “speaking.” As a partner to this model, a secondary system is being developed where a machine can mimic animal calls.

Lastly, ESP does note that it will go about the study in an ethical way that will not harm or put any wildlife in danger.

 
 
 
[via Lifewire and Futurism, opening image via A © Petrarichli | Dreamstime.com]