A Review on Audio Features Based Extraction of Songs From Movies

Abstract

A movie is a collection of different portions like comedy scenes, action, dialogs and songs. Extraction of a selected portion such as songs is a complex task. Basic three approaches namely visual based, text based and audio based are used by authors over the time for video content classification. This paper indicates that audio based approach is most suitable for extraction of songs. Audio based approach utilizes various audio features which can differentiate songs and non-songs as mentioned here. Different combinations of these features have been used by researchers from which a generic model is derived that can become the base for songs extraction application.