A novel framework to create customary soundtracks using MPEG-7 descriptors

In this paper, we propose a framework to create customary soundtracks for video files by adding a method layer upon MPEG-7. This method layer is used to identify potential frames in an annotated or non-annotated video file. Similarly an audio file for the soundtrack is selected based on its high level features, and descriptor for this file is generated if required. The two descriptors of the files are then combined and adjusted together to form a final joint MPEG-7 descriptor. The actual sound file is then manipulated to adjust with the video file according to the new descriptor, thia creating a customary soundtrack for the video file. Experimental analysis reports that the proposed framework is efficient enough to deliver results with good audio-visual coherence.