Snoring and breathing detection based on empirical mode decomposition

Snoring sound may indicate the presence of obstructive sleep apnea (OSA). In this study, we propose a novel method to detect snoring events in sleep audio recordings. Firstly, the mean and the standard deviation of instantaneous frequency are exacted as the features of the OSA acoustic features in each intrinsic mode functions(IMF) of Empirical Mode Decomposition (EMD). A support vector machine is then applied to perform the frame-based classification procedure in each IMF. This method was demonstrated experimentally to be effective for snoring and breathing detection. The database for detection included fullnight audio recordings from 5 individuals who have snoring habits. The performance of the proposed method was evaluated by classifying different events (expiration, inspiration, snoring, and breath) from the snoing sound recordings. In the experiments, proposed algorithm achieved an accuracy of 99.65% for detecting snoring and 99.30% for breathing events.