Correlation analysis of audio and video contents: A metadata based approach

The study of the associations between audio and video contents has numerous important applications in the fields of multimedia mining and information retrieval. Many works are implemented to correlate audio and video contents based on signal processing. This work represents audio visual correlation using metadata of audio and video as a leading role. The proposed method focuses on the web videos which exhibit a broad range of structural and semantic relationships between audio and video contents. To identify such relationships, it is necessary to undergo correlation analysis of audio and video contents. In this work, attempts are made to correlate audio and video contents of Entertainments, Sports and News categories web videos effectively and efficiently, and also correlation coefficient are calculated. The experimental procedures and results on the correlation of audio-video sequences by considering audio spectrogram frequency value to corresponding video bitrate value is described using different methods.