Scientists use AI to interpret bird calls

Researchers at the University of Michigan are using artificial intelligence (AI) to better understand whether a dog's barks communicate playful or angry emotions.

They are also investigating whether AI can correctly identify a dog's age, sex and breed from the way it barks.

Scientists have been able to make progress towards deciphering dog communication by reusing existing computer models trained on human speech.

“Advances in AI have the potential to revolutionize our understanding of animal communication,” said Radha Mihalcea, director of the University of Michigan's AI Institute.

“Our research leverages what we've built in speech processing to open up new avenues for understanding the nuances of dog barks.”

AI has made great strides in our ability to understand the subtle nuances of conversation.

AI-powered systems are used to distinguish nuances in tone, pitch and accent, enabling technologies such as voice recognition software.

It reaches that level of sophistication by training on a huge number of real human voices.

However, no comparable database exists for dogs.

“It is physically very difficult to pick up and record animal sounds,” said Artem Abzaliyev, the study's lead author.

His team set out to explore whether scientists could get around the data gap by piggybacking on research done in humans.

So his team collected the barks, growls and whimpers of 74 dogs of different breeds, ages and sexes in a variety of situations.

They fed that into a machine learning model (a type of algorithm that identifies patterns in large data sets) designed to analyze human speech.

The robot also proved effective at picking up on what the dog was trying to communicate.

The researchers found that on average, the model was 70% accurate across a range of tests.

“This is the first time that technology optimized for human speech has been used to help decipher animal communication,” Michalcea said.

“Our findings show that the sounds and patterns derived from human vocalizations can serve as a basis for analysing and understanding the acoustic patterns of other sounds, such as animal vocalizations.”

The researchers said the findings could have “important implications” for animal welfare.

The researchers suggest that a better understanding of the nuances of the different sounds animals make could improve the way humans interpret and respond to their emotional and physical needs.

The results were presented at the Joint International Conference on Computational Linguistics, Language Resources and Evaluation.

Mexico's National Institute of Astrophysics, Optics and Electronics also worked on the project in collaboration with the University of Michigan.