Danish scientists from Copenhagen University have trained artificial intelligence (AI) to recognize the emotions of seven animals caused by analyzing their pronunciation. Machine learning model was developed by experts who were able to successfully distinguish between a positive and negative feeling. The research was published in the scientific journal ICience.
The team analyzed thousands of sounds created by cows, pigs and other animals in different emotional states. The main indicators of emotions have become changes in the time of signals, energy distribution, frequency and main amplitude modulation.
These sound patterns turn out the same in different types, which shows an emotional conservative manifestation system.
According to the researchers, the accuracy of AI model in classifying animal emotions reaches 89.49%. This finding can contribute to the understanding of the history of emotional development, as well as promptly determine stress or discomfort between the representatives of the animal world.