Home // International Journal On Advances in Software, volume 7, numbers 1 and 2, 2014 // View article
Authors:
Shuichi Kurabayashi
Yoshiyuki Kato
Keywords: Cross-Media Retrieval, Emotion-Aware Search, User Interface
Abstract:
VizMIR is a music retrieval system that provides an intuitive user interface to search for music based on the sentiment that it evokes. The unique feature of this system is a cross-media retrieval mechanism that accepts a sequence of images to describe user requirements for mood transitions within a musical composition. VizMIR has a hybrid metric space for converting color change in images to continuous tonal changes in music, and vice versa. When a user enters a sequence of images as a query specifying the desired changes of mood in music, VizMIR measures the color distance in the image sequence, converts the calculated distance into the distance of the movement of musical tonality, and finds music that has the same or similar tonality movement. To support this bidirectional conversion of distance, we design two metric spaces as topologically equivalent structures, and provide a bridge function that maps a distance measured in the color metric space into one in the tonality metric space. This system enables users to search music by subtly manipulating queries through trial and error, and this is easy to use because images are suited to interactive manipulation. This method is useful in searching for music unknown to the user that evinces a mood satisfying the user's preferences.
Pages: 161 to 170
Copyright: Copyright (c) to authors, 2014. Used with permission.
Publication date: June 30, 2014
Published in: journal
ISSN: 1942-2628