Automatic Respiratory Sound Classification Using Deep Learning

This project focuses on the automatic classification of respiratory sounds using deep learning techniques. Lung sound recordings are preprocessed and converted into mel-spectrogram representations, which are then used to train a Convolutional Neural Network (CNN). The system classifies respiratory sounds into normal and abnormal categories, including crackles and wheezes, using the ICBHI 2017 dataset. By learning features directly from audio signals, the approach reduces reliance on manual feature extraction and improves robustness to noise and variability in real-world recordings. The results demonstrate the effectiveness of deep learning-based analysis for supporting objective and consistent respiratory sound diagnosis.