Audio Content Retrieval Based On Ontology

Audio is an important part of multimedia information and people to h andle the audio information is also growing, to find the target from the vas t amount of audio data audio content, also poses a challenge. Audio(voice audio) information retrieval systems, speech recognition, speaker recognit ion, audio similarity matching of data technology and audio annotation pro cessing technology is the core of the data link.Based on audio content search in the entire process of technique is re quired for analysis and research, analyzing and summarizing the current st atus of domestic research on audio retrieval, the audio feature is analyzed,and some feature extraction using MATLAB software. Studied in research on speech recognition and speaker identification, respectively, based on t he combination of these two different patterns, both identifying model is c onstructed. In addition, the ontology-based retrieval of audio content is al so presented. In audio retrieval experiment system, speech recognition, sp eech synthesis module embedded in them, the recognition results to be ana lyzed and summarized. From the experimental results, audio text recogniti on rate is 79.24%, the recognition rate between men and women is 86.11%, recognition rates may also need to upgrade, but for speakers of differe nt dialects accents, results have reached the expected objectives. This onto logy technology in audio data management, callout model is proposed to s olve audio level for further study of the semantic gap.