Innovative IoT and Machine Learning Approach for Smart Baby Monitoring

The identification of a baby cry is crucial to understand infant needs to understand properly the methods by which infants require care. In one way, systems often focus on sound recognition based on feature extraction and classification, but they suffer from problems regarding noise interference as well as inaccuracy in classification. This challenge arises basically due to the hard presence of external and internal noises making cry patterns hard to differentiate. In the proposed approach, MFCC coefficients for feature extraction and CNN based on AlexNet architecture will classify it. The proposed solution features pre-processing techniques including DWT to filter out noise from the data so that clean data is available, which will ensure proper classification. This way, baby cry classification will become improved and allow parents to monitor their infant over the internet, particularly when the conditions are critical. The proposed system provides improvements concerning improved accuracy for cry recognition, noise reduction, and real-time classification. Thus, it improves the efficiency and reliability of monitoring. This will make the system a reliable tool for baby care.