Sound synthesis system for structured timbre

2016 Fiscal Year Final Research Report

• Project/Area Number: 26370111
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Aesthetics and studies on art
• Research Institution: Tokyo Denki University
• Principal Investigator: OSAKA NAOTOSHI 東京電機大学, 未来科学部, 教授 (20366389)
• Project Period (FY): 2014-04-01 – 2017-03-31
• Keywords: 音色 / 音色合成 / 構造的音色 / 音の音 / サウンドモーフィング / サウンドハイブリッド

Outline of Final Research Achievements

In computer music, timbre synthesis is a very important technical area since it directly devotes to the new expression. So far sound morphing and sound hybridization have been studied isolatedly. We add "sound by sounds" and defined a new concept of structured timbre for all of three. 'Sound by sounds' used many other timbres to synthesize one particular timbre. The difference from sound hybridization is that it is defined to have a hierarchical structure and the timbre of the lower layer is identified with high intelligibility.
Based on this idea, two synthesis examples are studied: one is a speech is expressed in terms of string sound. The other is as an application of NMF (Non Negative Matrix Factorization), speech is expressed in terms of violin and string sound.
The former is example was introduced in a real computer music concert. In the second example, if the intelligibility of speech is higher, the recognition of lower layer instrumental sound becomes weaker.

Free Research Field

楽音合成