Similarity Graph
Published:
Paper: http://sibgrapi.sid.inpe.br/archive.cgi/sid.inpe.br/sibgrapi/2015/08.20.14.29
Abstract: Music collections are widely available on the Internet and, with the increasing storage and bandwidth capability, users can currently access thousands of songs, what brings challenges to music organization and exploration. Therefore, there is a growing demand towards automated Music Information Retrieval (MIR) tools for organizing, retrieving and processing music data sets. Recent visualization systems have focused in showing music similarities using audio data and multidimensional projection. Nonetheless, they do not take into account similarities between parts of songs, that is, they compare pieces as a whole, neglecting the aspect that some of them have similarities only in small parts. This work proposes a novel methodology for users to visually explore music collections considering that the similarity can take place only in small parts of the song. Our approach uses MIR to find similar segments between pairs of audio files and a graph metaphor to display the detected similarities.
