The Infant Cries Translator was developed at the National Taiwan University Hospital Yunlin and can differentiate between four separate crying sounds by recording the sounds of babies and comparing them against a vast database.
Over a two year period researchers collected around 200,000 crying sounds from approximately 100 new-born babies, and uploaded them to an online database. Analysis of the frequency of individual screams among these helped researchers, led by Chang Chuan-yu and Dr Chen Si-da, distinguish subtle differences in acoustics.
The resulting app shows analysis of a baby’s cries on the user’s phone within 15 seconds. Researchers say the app has an accuracy of 92 percent for infants under two weeks old, helping inform parents when their child is hungry, sleepy, in pain, or has a wet diaper. The analysis becomes less accurate the older the baby is.
“The Infant Cries Translator can differentiate four different statuses of sounds of baby crying, including hunger, the diaper getting wet, sleepy and pain,” said Chuan-yu. “So far, according to the feedback from users, the accuracy of the app we’ve tested can reach 92 percent for babies under two weeks old. As for the babies under one or two months, the accuracy of the app can also reach up to 84 or 85 percent. Even for the four month old baby, the accuracy can reach 77 percent.”
Its creators say there is little point using the app past the age of six months because the baby has become more affected by its environment, but they believe it will be a useful tool for parents, especially those with their first child.
The app constantly updates its database to a Cloud Drive. Its machine learning algorithm means that mothers and fathers can set up a personal setting for their infant, based on the parents’ feedback.
“After downloading this app, we only need to set the birth of date and nationality for the new born baby,” said Chang. “Once the baby cries, we only need to press the recording button for 10 seconds, and the sound will be uploaded to the Cloud Drive. After the differentiating process, the analysis result of the sound would be transferred to Mum’s mobile phone. So Mum can depend on the actual situation to determine whether the analysis result is correct or not, to make a revision for the app.”