Infinite-state spectrum model for music signal analysis

Author

Nakano, Masahiro ; Le Roux, Jonathan ; Kameoka, Hirokazu ; Ono, Nobutaka ; Sagayama, Shigeki

Author_Institution

Grad. Sch. of Inf. Sci. & Technol., Univ. of Tokyo, Tokyo, Japan

fYear

2011

fDate

22-27 May 2011

Firstpage

1972

Lastpage

1975

Abstract

This paper presents a nonparametric Bayesian extension of non-negative matrix factorization (NMF) for music signal analysis. Instrument sounds often exhibit non-stationary spectral characteristics. We introduce infinite-state spectral bases into NMF to represent time-varying spectra in polyphonic music signals. We describe our extension of NMF with infinite-state spectral bases generated by the Dirichlet process in a statistical framework, derive an efficient optimization algorithm based on collapsed variational inference, and validate the framework on audio data.

Keywords

Bayes methods; audio signal processing; inference mechanisms; matrix decomposition; music; nonparametric statistics; optimisation; spectral analysis; Dirichlet process; NMF; collapsed variational inference; infinite-state spectrum model; nonnegative matrix factorization; nonparametric extension; nonstationary spectral characteristics; optimization algorithm; polyphonic music signal analysis; time-varying spectra; Bayesian methods; Hidden Markov models; Inference algorithms; Instruments; Multiple signal classification; Signal analysis; Spectrogram; Collapsed variational Bayes; Dirichlet process; Nonnegative matrix factorization (NMF); Nonparametric Bayes;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech and Signal Processing (ICASSP), 2011 IEEE International Conference on

Conference_Location

Prague

ISSN

1520-6149

Print_ISBN

978-1-4577-0538-0

Electronic_ISBN

1520-6149

Type

conf

DOI

10.1109/ICASSP.2011.5946896

Filename

5946896