Emotion Detection in Baby Cries Using Machine Learning with Lullaby Recommendations

Abstract: The interpretation of baby cries is a vital yet challenging task in caregiving. Babies express their needs—hunger, sleepiness, discomfort, or happiness—through crying, but decoding these emotional cues is subjective and prone to error. This paper introduces a machine learning-based system to analyze baby cries, classify their emotional states, and recommend emotion-specific lullabies to caregivers. Utilizing audio feature extraction techniques like spectral centroid, Mel-Frequency Cepstral Coefficients (MFCCs), and RMS intensity, coupled with a Random Forest Classifier, the system achieves remarkable accuracy and robustness. Deployed via a Django-based web application, the system ensures accessibility and real-time responsiveness, making it suitable for diverse caregiving contexts. Extensive evaluation in both controlled and noisy environments confirms the system's effectiveness. By bridging traditional caregiving practices with cutting-edge technology, this research offers a novel tool for enhancing infant care while preserving the cultural richness of lullabies.