PROCEDE ET SYSTEME POUR DETECTER DES EVENEMENTS SONORES DANS UN ENVIRONNEMENT DONNE

摘要

The method involves performing unsupervised initialization of a group of classes or subspace of normal ambiance of audio data. Normality of a model (M) is defined. The groups that use the model are optimized to redistribute the data to different groups. An unknown audio signal obtained from a to-be analyzed environment is analyzed. The unknown audio signal is compared to the obtained model. Comparison of the scores is obtained by classifiers using decision rules to determine the presence or absence of anomaly in the analyzed audio signal. An independent claim is also included for a system for determining abnormal events in an environment.