Joyce Poole and Michael Pardo recently published a groundbreaking study on elephant communication: Using a machine learning model, they were able to show strong evidence that African savannah elephants have unique names for one another. The statistical model they used — known as a random forest model — is nothing new or snazzy. It’s been around for 20 years. But it’s one example of how animal communication researchers are using machine learning to decode animal calls they can’t through observation alone.

This video covers some other ways machine learning is solving the limits of human observation in the study of animal communication. And it explains a wild plan for what might be next: deep learning models that could facilitate interspecies communication.