A Home Sounds Classification For Health Smart Home (HSH)

This work is part of project C EPRU , it describes the design phase of a system to classify everyday life sounds, based on separation and classification of audio sources in a habitat. Based on previous work, where we collected database and analyzed some sounds, an application of telemonitoring of the elderly or disabled is proposed. First the key concepts of the research area are presented, we studied the different sounds used in an application of telemonitoring from which we are inspired to create our database. The features are extracted from segments and models were developed then a classification using dynamic Time Warping is applied. Tests were done on some sounds and results are reported on table. Finally we implemented a Graphic used Interface based on simple features for classification of seven type of home sounds. The obtained results are very encouraging. Keywordshealth smart home; Features and classification, audio channel; DTW classifier.