Researchers at Northern Illinois University in the United States have devised an algorithm that can listen to and decipher the cries of babies.
It’s not always easy to calm baby when he or she starts crying. Is he or she hungry? Is he or she tired? Or it time for diaper change? Artificial intelligence could soon help distraught parents decode the cries of their toddlers.
Researchers at Northern Illinois University in the United States have developed an algorithm that can listen to and decipher the cries of a baby. The artificial intelligence behind the algorithm analyzes the differences between the sound levels of the cries emitted by the child to find the cause.
The algorithm could also identify if the crying is abnormal, which can sometimes announce health problems such as infection, pneumonia or central nervous system disorders. “As in a special language, there is a lot of health information in different sounds,” says Lichuan Liu, author of the study published in Journal of Automatica Sinica and relayed by Newsweek magazine. “The differences between the sound signals actually carry the information. (…) To recognize and exploit this information, we must extract the characteristics [of the cry signals] and then retrieve the information contained therein.”
The goal is “to have healthier babies and relieve pressure on parents and caregivers,” added Lichuan Liu.
To develop their argorithm, the researchers recorded 48 tears from 26 babies hospitalized in neonatology. These cries have been interpreted by relatives of infants or experienced nurses, reports The Canadian Huffington Post.
The algorithm needs to be further improved and researchers are using hospitals and medical research centers with whom they could collaborate to make the tool even more accurate. But the scientists agree: the AI will remain a assistant tool that will never supplant parents.