An audio classification and speech recognition system for video content analysis

Audio can provide useful information for video content analysis. Audio classification and speech recognition for video content analysis is proposed in this paper. Firstly, audio data from video stream is extracted. Secondly, the audio frames are classified into silence, speech and music based on rules and Support Vector Machine(SVM) algorithm. Finally, an automatic speech recognition(ASR) system is applied for speech-to-text conversion. The experimental result on CCTV_NEWS of TRECVID shows that our approach is effective.