This is a great idea!
Started offering "Papatto Ikkoiku@Baby Notebook" to analyze emotions from baby's crying.
It appears to have a feature that allows a smartphone application to analyze the pattern of a baby's cries and why the baby is crying. The following is a quote from the press release published by PRTimes.
■About the cry diagnosis function
First Ascent Inc. has built a crying voice diagnostic algorithm by collecting and analyzing crying voices from more than 20,000 monitor users. Users can activate the cry diagnosis function from the application and hold the microphone over the crying baby to find out why it is crying. The diagnostic algorithm displays the classification and probability analyzed as highly likely in five categories: "hungry," "sleepy," "uncomfortable," "angry," and "wants to play. According to the results of preliminary monitor user feedback, the correct response rate is over 80%.
The aim seems to be to increase the number of users and collect more sampling cries to improve the accuracy of cries diagnostics through machine learning.
One question I have is what is the percentage of correct responses determined by? Well, would it be something like: stopped crying = understood the baby's appeal? The difference between uncomfortable and angry also seems to be quite a fine line. For example, if the body is hot and uncomfortable, if it is hot and angry, I think it is hard to say whether it is uncomfortable or angry. Still, it is better than having no information at all, and if accurate information can be provided, it will greatly help third parties to watch over the baby. I look forward to the future.
