Enhancing Caregiving with Multi-Modal Baby Cr...

Enhancing Caregiving with Multi-Modal Baby Cry Analysis

Author Name : Prajwal Helunde, Abhishek Dhole, Bhushan Banswal , Kaushik Morayya , Altamash Pathan , Pratik Shrote, Ravindra Kale

DOI: https://doi.org/10.56025/IJARESM.2023.11112316

ABSTRACT A baby's universal language is their cry, a moving expression of their emotions, from discomfort to joy. However, for many new parents, deciphering the meaning of their baby's cries can be a confusing challenge. The demands of modern life often leave parents with limited time for their babies, leading to frustration and confusion when faced with prolonged crying. In response to this common dilemma, in this paper we present an innovative solution: a technology-driven approach to predicting the reasons for infant crying using state-of-the-art machine learning techniques. Our system is designed to capture baby cries using sound analysis. Utilizing advanced audio preprocessing methods and feature extraction techniques, including Mel-Frequency Ceptral Coefficients (MFCC) and Librosa, it achieves an impressive accuracy rate of 78%. To further refine the analysis, we use the Random Forest classifier, which categorizes a baby's cry into one of seven different types: hunger, sleepiness, fear, discomfort, pain, and the need to burp. To make this technology easily accessible to caregivers, we developed a user-friendly mobile app with a real-time cry recording feature. This app allows parents to better understand and respond to their child's needs, ultimately strengthening the parent-child bond.