I Wonder what my Cats are really Saying to me!

A new study from the University of Michigan leveraged AI models trained on human speech to decode the meaning behind dog noises — identifying details like breed, age, gender, and emotional state with 70% accuracy.

The details:

Researchers gathered vocalizations from 74 dogs of varying breeds, ages, and situational contexts.
The noises were inputted into an AI model originally designed to analyze human voices, trained on 960 hours of speech, and fine-tuned for dogs.
The AI was able to predict individual dogs from barks, distinguish dog breed and gender, and match barks to emotional context like play and aggression with 70% accuracy.

Why it matters: AI is not only bridging the language gap for humans across the globe — but also potentially across species as well. Communicating with other intelligent animals (or at least better understanding them) seems like a skill issue that’s going to be solved sooner rather than later.

5 comments