A cover song is a new performance or recording of a previously recorded music by an artist other than the original one. The automatic identification of cover songs is useful for a wide range of tasks, from fans looking for new versions of their favorite songs to organizations involved in licensing copyrighted songs. This is a difficult task given that a cover may differ from the original song in key, timbre, tempo, structure, arrangement and even language of the vocals. Cover song identification has attracted some attention recently. However, most of the state-of-the-art approaches are based on similarity search, which involves a large number of similarity computations to retrieve potential cover versions for a query recording. In this paper, we adapt the idea of time series shapelets for contentbased music retrieval. Our proposal adds a training phase that finds small excerpts of feature vectors that best describe each song. We demonstrate that we can use such small segments to identify cover songs with higher identification rates and more than one order of magnitude faster than methods that use features to describe the whole music.
Download Full PDF Version (Non-Commercial Use)