Researchers from the University of Michigan, Mexico's National Institute of Astrophysics and the Institute of Optics and Electronics collaborated on the study using an artificial intelligence model called Wav2Vec2, which was trained for more than 1,000 hours to analyze human speech. But this time, they used it to analyze dog barking instead. For a basic understanding of animal language?
“There is still a lot we don't know about the animals that share this world with us.”
“Advancements in artificial intelligence can be used to revolutionize our understanding of animal communication,” Rada Mihalcea, director of the University of Michigan’s Artificial Intelligence Laboratory, said in a press release. “Our findings suggest that we may not need to start from scratch.”
The study used the advanced AI-based speech model Wav2Vec2 to determine the mood, gender and breed of dogs in their barking. The researchers used two different data sets to train and compare the results. One group was trained from scratch using only dog barking sounds. One group was pre-trained on human speech. Then adjust the barking sound. The model was pre-trained on nearly 1,000 hours of human speech recordings, and the researchers then fine-tuned the model on a dataset that included barking sounds from 74 dogs, including 42 Chihuahuas, 21 Doodles, and 11 Schnauzers.
