With 70% accuracy, scientists use AI to interpret a dog's bark

IT House

2024-06-07 19:41Posted on the official account of Shandong IT Home

IT House reported on June 7 that researchers are trying to use AI to interpret whether a dog's barking is playing or angry. At the same time, the researchers also tried to use AI to identify the age, sex, and breed of dogs.

Researchers at the University of Michigan, in collaboration with Mexico's National Institute of Astrophysics, Optics and Electronics (INAOE) in Puebla, conducted the study and found that the AI model originally used to train human speech could serve as a starting point for training animal communication models.

Image via Pixabay

Rada Mihalcea, head of the University of Michigan's AI Lab, said AI has made significant advances in understanding the subtleties of speech, distinguishing the nuances of pitch, pitch and accent, and can rely on these research foundations to understand dog barking.

One of the main hurdles to the development of such AI models that analyze animal vocalizations is the lack of publicly available data. While there are many resources and opportunities to record human speech, it is more difficult to collect data from animals.

The team attempted to collect information about the dogs' barking using human voice data, collecting barking, growling, and whining sounds from 74 dogs of different breeds, ages, and genders in a variety of situations.

The team used this vocal information to analyze human speech in a machine model that understands the communication between dogs very well, and the model is 70% accurate in various tests.

According to Rada Mihalcea, "sounds and patterns from human speech can serve as a basis for analyzing and understanding the acoustic patterns of other sounds, such as animal vocalizations". Other researchers on the team also say that a better understanding of the nuances of the various sounds that animals make could improve human interpretation and response to their emotional and physical needs.

It is reported that the results of the experiment were presented at the 2024 International Joint Conference on Computational Linguistics, Language Resources and Assessment. IT Home with paper link:

View original image 53K

•