Presentation 2013-05-23
The proposal of multiplex sound analysis scheme based on spectrum basis using Gauss process regression model
Akihiro KAGIMOTO, Hiroyuki KASAI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In multiplex sound analysis, it is generally difficult to recognize the pitch of each sound and the instrument category which belongs. In the conventional system, the sound quantity belonging to the same musical instrument category assumes that it has the same spectrum power ratio (harmonic structure), and analyzes using a fixed harmonic structure. However, harmonic structure is, in general, not constant. In this paper, harmonic structure is variably made by learning instrument sound in advance and making a model by a gauss process regression. And sound quantity to recognize and the category which belongs are specified by finding the optimal combination from this spectrum basis. We show the validity of the proposed system by measuring its recognition accuracy and the system which assumes the case of harmonic structure regularity.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Gauss process regression / Specmurt / Non-negative Matrix Factorization
Paper # MoNA2013-2
Date of Issue

Conference Information
Committee MoNA
Conference Date 2013/5/16(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Mobile Network and Applications(MoNA)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) The proposal of multiplex sound analysis scheme based on spectrum basis using Gauss process regression model
Sub Title (in English)
Keyword(1) Gauss process regression
Keyword(2) Specmurt
Keyword(3) Non-negative Matrix Factorization
1st Author's Name Akihiro KAGIMOTO
1st Author's Affiliation University of Electro-Communications()
2nd Author's Name Hiroyuki KASAI
2nd Author's Affiliation University of Electro-Communications
Date 2013-05-23
Paper # MoNA2013-2
Volume (vol) vol.113
Number (no) 56
Page pp.pp.-
#Pages 6
Date of Issue