Presentation 2012-11-07
Infinite Composite Autoregressive Models based on Gamma Processes for Music Signal Analysis
Kazuyoshi YOSHII, Masataka GOTO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This paper presents novel statistical models that can be used for multiple fundamental frequency (F0) estimation and timbre-based source separation of polyphonic music signals. These models are based on nonnegative matrix factorization (NMF) extended by the source-filter paradigm, in which an amplitude or power spectrogram is decomposed as the product of two kinds of spectral atoms (sources and filters) and time-varying gains of source-filter pairs. However, a critical problem is that the numbers of sources and filters should be given in advance. To solve this problem, we propose nonparametric Bayesian models based on gamma processes and efficient variational and multiplicative learning algorithms. These infinite composite autoregressive models (iCARMs) can discover effective numbers of sources and filters in a data-driven manner.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Music signal analysis / F0 estimation / source separation / gamma process / nonparametric Bayes
Paper # IBISML2012-51
Date of Issue

Conference Information
Committee IBISML
Conference Date 2012/10/31(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 Information-Based Induction Sciences and Machine Learning (IBISML)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Infinite Composite Autoregressive Models based on Gamma Processes for Music Signal Analysis
Sub Title (in English)
Keyword(1) Music signal analysis
Keyword(2) F0 estimation
Keyword(3) source separation
Keyword(4) gamma process
Keyword(5) nonparametric Bayes
1st Author's Name Kazuyoshi YOSHII
1st Author's Affiliation National Institute of Advanced Industrial Science and Technology (AIST)()
2nd Author's Name Masataka GOTO
2nd Author's Affiliation National Institute of Advanced Industrial Science and Technology (AIST)
Date 2012-11-07
Paper # IBISML2012-51
Volume (vol) vol.112
Number (no) 279
Page pp.pp.-
#Pages 8
Date of Issue