ARLINGTON, Texas — Have you ever wondered what your dog's barks mean? A new team of researchers at UT Arlington thinks they just might be able to tell you.
"People have always tried to understand what animals are talking about," Kenny Zhu, Professor of computer science at UTA told WFAA.
Zhu received a three-year, $483,804 National Science Foundation Research Experiences for Undergraduates grant to fund his research. He and his team of 14 students are on a mission: to use machine learning and artificial intelligence to decode 'dog words' (in other words: what dog barks might mean.)
"We are trying to do this completely automatically, and systematically," he explained.
Biologists and animal scientists have manually collected dog bark data before to analyze for potential patterns. But what Zhu is doing is new.
"So far we have collected more than 30,000 videos of dogs," he explained.
Zhu's team is using the machine learning and AI technology to translate dog sounds into phonetic representations, and hopefully, one day, words.
"We’ve been able to discover certain word-like patterns, in the dog's so-called language, but we are still trying to verify if these are really words."
The technology essentially catalogs the barks, analyzes the sounds, and segments them into numbers like syllables.
"The sequence of numbers can make up a word, that’s what we think."
Zhu knows there are skeptics but the science, he said, is real and promising.
And if Artificial Intelligence can crack the code, could we really communicate with dogs, there’s no telling what that might mean.
Zhu said he believes the technology could help detect diseases in dogs quicker, assist dogs with mining or even disaster response.
"The knowledge we can get from animals can be hugely beneficial to society," Zhu said.
The project will also study possible language in other animals, not just dogs.