Researchers from various institutes developed a way to decode dog barks using human speech-trained AI models.
The study was initiated by the University of Michigan and Mexico's National Institute of Astrophysics, and Optics and Electronics Institute.
Researchers Rely on AI Model to Translate Animal Communication
In a press release, the University of Michigan revealed that the team used AI models that were originally trained on human speech as a starting point to develop a unique system that can translate animal communication.
There is a huge research gap in developing tools that could analyze animal vocalizations. Moreover, collecting data from animals is harder than obtaining speech samples from a human.
"There is so much we don't yet know about the animals that share this world with us. Advances in AI can be used to revolutionize our understanding of animal communication, and our findings suggest that we may not have to start from scratch," said Janice M. Jenkins, collegiate professor of Computer Science and Engineering, and director of the University's AI Laboratory.
Existing AI Models Decode Dog Barks
While there is a lack of research and data on analyzing animal speech, the researchers repurposed existing models that were trained with human speech. The trained AI models allowed the team to form the foundation of the new system.
Human speech-trained AI models are capable of identifying tone, pitch, and accent. These nuances are recorded and analyzed by a computer to distinguish a person's speaking intent.
The study recorded vocalizations from 74 dogs. They varied in breed, age, sex, and other contexts. The data were submitted to the lab and were used to modify the learning system of the existing AI models.
The researchers managed to generate representations and interpretations from the collected recordings. The study could be vital for further understanding animal welfare in the future.
