A Comprehensive Review of Infant Cry Analysis and Classification: Techniques, Challenges, and Future Directions

Crying is a newborn's natural and spontaneous form of communication, expressing their needs and discomforts. Variations in the acoustic properties of cries can serve as early indicators of medical conditions such as asphyxia, hypothyroidism, or hearing impairments. Recognizing these variations is essential for timely and appropriate medical intervention. This comprehensive review traces the evolution of infant cry analysis and classification, from manual observation to automatic recognition systems. It examines key phases of the analysis process, including data collection, signal preprocessing, feature extraction, feature selection and dimensionality reduction, classification techniques, and performance evaluation. Artificial intelligence‐based approaches are examined, with a focus on machine learning (ML) algorithms, neural network architectures, and deep learning (DL) frameworks. In addition, this work identifies persistent challenges such as dataset scarcity, lack of standardized benchmarks, and limited cross‐dataset generalization, and outlines future research priorities, including explainable AI, robust modeling in noisy environments, and ethical deployment in clinical care. Complementing the review, the study empirically benchmarks the Baby Chillanto dataset, using established CNNs as baselines for future research. Overall, it provides a foundational reference for developing accurate, interpretable, and practical cry recognition systems to support infant care and clinical applications. This article is categorized under: Application Areas > Health Care Technologies > Artificial Intelligence Technologies > Machine Learning