Scientists create a pig translator that decodes the emotions of grunts

An international team of researchers has reported the development of an algorithm capable of translating the emotional state of pigs from the sound of their growls. The researchers say the system could be used to monitor the welfare of pigs on a farm in real time.

Domestic pigs display very sophisticated varieties of vocal expression. Previous studies have found correlations between high-frequency calls, such as yelling and shouting, associated with negative emotions, and low-frequency growls associated with positive or neutral emotions. But between these two extremes lies an assortment of less well-understood sounds.

The new research first sought to understand the wide range of pig vocalizations. To do this, the researchers listed 7,414 different pig sounds, from 411 animals.

Each pig sound was accompanied by specific behavioral observations and, where possible, heart rate monitoring to establish positive or negative emotional associations. Positive scenarios were studied, such as piglets nursing or playing with toys, and emotionally negative scenarios were also tracked, including fights, separation from family, and slaughter.

In general, the results validated previous observations linking high-frequency calls to negative emotional states and low-frequency sounds to positive emotional states. However, the researchers found a significant volume of calls that did not fit this simple distinction.

Two particular acoustic characteristics have proven to be as important as frequency in understanding emotional valence: duration and rate of amplitude modulation. For example, a high-frequency scream was determined to represent positive emotion when it was short and contained few amplitude modulations.

“There are clear differences in pig cries when looking at positive and negative situations,” explained Elodie Briefer, author of the study from the University of Copenhagen. “In positive situations, screams are much shorter, with slight fluctuations in amplitude. Growls, specifically, start high and gradually decrease in frequency.

The researchers then used a neural network to develop an algorithm capable of translating the emotional characteristic of the pig’s sounds. In this proof-of-concept study, the researchers claim that the initial iteration of the algorithm can correctly translate pigs’ emotions from their calls with 92% accuracy.

The long-term goal, according to the researchers, would be to develop some sort of app that can monitor the emotional well-being of commercial pigs in real time. Briefer also hypothesizes that their analytical method is transferable to other types of mammals, suggesting the possibility of some sort of universal translator that could track animals’ emotions through the sounds they make.

“We trained the algorithm to decode pig grunts,” Briefer said. “Now we need someone who wants to develop the algorithm into an app that farmers can use to improve the welfare of their animals.”

The new study has been published in the journal Scientific reports.

Source: University of Copenhagen