Imagine a device that could monitor your pet dog when you aren’t at home and tell you if it’s anxious or upset.
Or one that could monitor the emotional state of dogs in a shelter or help owners decode the barks and other vocalisations of their assistance or therapy dogs. It could even be used by vets to help them diagnose whether a dog is ill or not.
- What are cats thinking? How to understand what your feline friend is trying to tell you
- Best dog quotes: our favourite sayings about man's best friend
A device, in other words, that can literally help us understand what dogs are saying.
These are just some of the possible end goals of research being carried out by scientists at the University of Michigan who have turned to Artificial Intelligence (AI) to help them uncover the secrets of canine conversation.
In the paper, the researchers explain how they used AI developed to analyse human speech to see if it could be repurposed to predict what the sound units contained within a dog’s vocalisations – not just barks, but grunts and squeals, too – actually mean.
They set up an experiment in which 74 dogs in Mexican domestic homes were exposed to a range of stimuli – for example, the researchers made the dogs bark by knocking on the door or by pretending to attack the owner and made audio recordings of how they reacted. Other simulations included playing with the dog, speaking affectionately to it or even pretending to take the dog for a walk and then walking out of sight.
The idea was to see if the AI model could accurately interpret what the dog vocalisations meant. And it could, with an accuracy rate of up to 70 per cent.
Of course, experts in dog behaviour are extremely good at understanding what dogs mean by barking and other noises they make. But you don’t have an expert at home, at a shelter or helping vets diagnose pets in their care.
“Advances in AI can be used to revolutionise our understanding of animal communication,” said Rada Mihalcea, professor or computer science and director of the university’s AI laboratory.
Part of the point of the research was that the scientists were using a model originally created to analyse human speech as the basis for their work. “Our results show that the sounds and patterns derived from human speech can serve as a foundation for analysing and understanding the acoustic patterns of other sounds, such as animal vocalisations,” said Mihalcea.
As well as further developing and improving the model for dogs, the scientists also want to see if they can use AI to interpret the vocalisations of other animals. “In the future, we are planning to extend our work to birds and marine mammals, since those species have a large amount of data available,” they say in the paper.
Love dogs? Check out our pets section for advice on dog care from how to get rid of fleas to why dogs eat grass and how to house train your puppy You can also find dog walking inspiration and expert advice to help your dog behave beautifully on walks in our dog guides
We have also rounded up a selection of dog whistles for recall, dog water bottles and even dog jackets for your four-legged friend.
