Scaling and compressing melodies using geometric similarity measures

• We propose efficient algorithms for two geometric problems that arise in music information retrieval; linear scaling and data compression. Linear is used tempo variation melody matching, while an important application of compression clustering classification. use matching techniques to measure the similarity between melodies. Our take advantage properties optimal solution are based on line sweeping dynamic programming techniques. Melodic measurement key importance retrieval. In this paper, we represent as sets points or horizontal segments Euclidean plane optimization inspired operations melodies; audio problem, incoming query scaled forward until a reference minimized. The problem asks subset notes given such cost selected