Jul 22, 2019 | Updated: 09:15 AM EDT

Artificial Intelligence To Help Translate Infant Cries

Jun 25, 2019 07:39 AM EDT

Close
Baby Crying
(Photo : joffi)

When babies start to cry, their primary caregivers are signalled that they are either hungry, sleepy or need a diaper change. However, no matter how simple this may seem, to first time parents, a baby's cry always means something they have yet to comprehend. Luckily, a group of researchers from the Northern Illinois University is working on an AI that will help parents make a distinction between what are normal cries and cries that are a result of an illness.

The results of this new study may be aimed at parents, but they could also help doctors and nurses in the hospital who want to provide the infant the kind of help it needs as identified by the distinct crying that they made. While they say that a baby's cry is unique, there are some things about it that seem to share some features.

An algorithm was developed by the team based on the automatic speech recognition system along with the technique experts call the compressed sensing. With both techniques working together, the team aimed at being able to reconstruct a signal that will be based on every single data collected to help detect and recognize the various features of infant cries. They made sure that the technology was able to collect the data despite the high levels of background noise present.

The algorithm analyzes the timbre, the loudness as well as the pitch in an infant's cry. Then, it compares it to the database of previously recorded cries identified by the primary care givers of infants as well as the hospital staff exposed to them.

"Much like a special language, a baby's cry related to so many health-related information. It is important that the primary caregiver understands that in order to give the care that the baby needs at a specific time," said Prof. Lichuan Liu. "The differences in the sound signals are actually the carriers of the information."

In order to recognize what the sound means, researchers had to extract the features of the cries and obtain the information from there. The team is hopeful that the method could be widened out to reach out to more areas of medicine where such technology could prove helpful

"The ultimate goal is to have healthier babies and put less pressure on their primary caregivers," said Liu. "We are looking into forming collaborations with hospitals and their medical research teams to obtain more data and create more possible scenarios."

©2017 ScienceTimes.com All rights reserved. Do not reproduce without permission. The window to the world of science times.
Real Time Analytics
<