DEEP LEARNING ALGORITHM-BASED SNORE MONITORING METHOD AND SYSTEM, AND CORRESPONDING ELECTRIC BED CONTROL METHOD AND SYSTEM

Disclosed is a deep learning algorithm-based snore monitoring method, which is based on deep learning and speech recognition technology, comprising: acquiring an audio signal and slicing same according to a preset sample duration; using a silence detection algorithm to determine whether a slice contains sound; extracting acoustic spectral features for an audio slice containing sound; inputting the generated spectral features into a deep neural network to extract deep learning features; using a fully connected layer to classify the extracted deep learning features; and according to a preset duration, determining a snore event and performing intervention. Also disclosed is a related system. Compared with the traditional method, a deep learning algorithm-based snore recognition method and system provided by the present invention greatly improve snore determination accuracy, thereby bringing better user experience. L'invention concerne un procédé de surveillance de ronflement reposant sur un algorithme d'apprenti