Content Based Search Engine for Arabic Audio Files

Abstract

Information systems were subject to considerable changes. These changes were arisen due to the fact that information representation underwent massive evolution. This evolution from heavyweight information representation (i.e. papers, books, music sheets) towards lightweight information (i.e. multimedia and digital forms) is driven by spectacular changes to the constrains of storing, managing and retrieving such types of information. Multimedia Information Retrieval (MIR) is about obtaining semantic information from different multimedia data sources (i.e. audio, video or images). The research in this field faces many challenges due to the variety of data types, methodologies and research problems. One of the challenging sub-field of MIR is audio retrieval. Audio retrieval aims to extract certain piece of information from audio signals contents based on a submitted query. Audio data types may include pure music, environmental sounds, songs and speech utterances. In this work, we have implemented a content based retrieval system for Arabic speech signals to retrieve the speech audios relevant to the user text query. A Large Vocabulary Arabic Continuous