期刊名称：International Journal of Computer Science and Network Security
印刷版ISSN：1738-7906
出版年度：2019
卷号：19
期号：4
页码：140-145
出版社：International Journal of Computer Science and Network Security
摘要：Humans have a remarkable ability to classify sound signals into classes: music, speech, applause, laughter, etc. Faced with an excessive abundance of multimedia documents, we propose in this paper to develop a new configuration of multimedia documents based on entropy and entropy energy for automatic segmentation. A sound classification plays an important role in rich and varied applications, ranging from indexing audio documents to protecting copyright and archiving the diversity of radio and television channels. Given the diversity of requirements of these potential applications in the sound of classification, our object is to choose a generalist approach to the classification of sound documents that can easily adapt to classes defined according to its particular application. The proposed approach is based on entropy energy modulation. The classical problems of sound classification are summarized by three classes: the classification into music/speech, man/woman and action/non-action. Our application concerns the segmentation of a sound track in speech/non-speech.
