Presentation 2004/4/16
Audio Source Separation by Deriving Independent Basis Vectors from Single Mixture Spectrogram
Md. Khademul Islam Molla, Keikichi Hiroset, Nobuaki Minematsu,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This report presents a method to separate the audio sources from their single mixture. The system is based on deriving some independent basis vectors from the mixture spectrogram and clustering them to produce the individual source subspaces. Principal component analysis (PCA) is used to reduce the dimension of mixture spectrogram producing some basis vectors. A data-adaptive decisive factor is proposed for the proper selection of the basis vectors. Independent component analysis (ICA) is employed to make the selected basis vectors independent from each other in their own domain. Kullback-Leibler divergence based k-means clustering algorithm is introduced to grouping the independent basis vectors into given number of sources. The satisfactory result of two-source mixture separation motivates to use this system for real world single mixture audio source separation.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Spectrogram / Basis Vectors / PCA / ICA / Kullback-Leibler Divergence
Paper # SP2004-5
Date of Issue

Conference Information
Committee SP
Conference Date 2004/4/16(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 Speech (SP)
Language ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Audio Source Separation by Deriving Independent Basis Vectors from Single Mixture Spectrogram
Sub Title (in English)
Keyword(1) Spectrogram
Keyword(2) Basis Vectors
Keyword(3) PCA
Keyword(4) ICA
Keyword(5) Kullback-Leibler Divergence
1st Author's Name Md. Khademul Islam Molla
1st Author's Affiliation Graduate School of Frontier Sciences, The University of Tokyo()
2nd Author's Name Keikichi Hiroset
2nd Author's Affiliation Graduate School of Frontier Sciences, The University of Tokyo
3rd Author's Name Nobuaki Minematsu
3rd Author's Affiliation Graduate School of Information Science and Technology The University of Tokyo
Date 2004/4/16
Paper # SP2004-5
Volume (vol) vol.104
Number (no) 30
Page pp.pp.-
#Pages 6
Date of Issue