By Jacob Gronholt-Pedersen
VIPPEROD, Denmark (Reuters) – European scientists have developed an artificial intelligence (AI) algorithm that can interpret pig sounds, aiming to create a tool that could help farmers improve animal welfare.
The algorithm could potentially alert farmers to negative emotions in pigs, improving their welfare, said Elodie Mandel-Briefer, a behavioral biologist at the University of Copenhagen who is co-leading the research.
The scientists, from universities in Denmark, Germany, Switzerland, France, Norway and the Czech Republic, used thousands of recorded pig sounds in a variety of scenarios, including play, isolation and competition for food, to find that growling, grunting and squeaking reveal positive effects. or negative emotions.
Although many farmers already have a good insight into the well-being of their animals by looking at them in the pigsty, existing instruments mainly measure their physical condition, according to Mandel-Briefer.
“Animals’ emotions are critical to their well-being, but we don’t measure this much on farms,” she said.
The algorithm showed that pigs raised outdoors, free-range or organic farms with the ability to roam and dig in the ground produced fewer stress calls than conventionally raised pigs. The researchers believe that once fully developed, this method could also be used to label farms, allowing consumers to make informed choices.
“Once we get the tool working, farmers will be able to have an app on their phone that can translate what their pigs are saying in terms of emotions,” Mandel-Briefer said.
Short grunts generally indicate positive emotions, while long grunts often signal discomfort, such as when pigs push each other at the trough. High-frequency sounds such as screams or squeaks usually mean that the pigs are stressed, for example if they are in pain, fighting or separated.
The scientists used these findings to create an algorithm that uses AI.
“Artificial intelligence really helps us to process the enormous amount of sounds we ingest, but also to classify them automatically,” says Mandel-Briefer.
(Reporting by Jacob Gronholt-Pedersen, editing by William Maclean)
