Musical morphing by parallel processing of musical components

摘要

Many morphing methods for image, speech and musical tone have been proposed. But no morphing method for musical song have been proposed. In this paper, we propose a morphing method for musical song (one measure, 3 tracks) by parallel processing of musical components - timbre contour, rhythm and harmony. We analyzed questionnaire data about the similarity of components by multidimensional scaling and Neural Networks, and got psychological similarity map of each musical components. We got musical morphing by re-synthesizing morphing results on the similarity map of each musical components. By experiments, we succeeded to get satisfactory morphing results.