One microphone singing voice separation using source-adapted models

Author

Ozerov, Alexey ; Philippe, Pierrick ; Gribonval, Rémi ; Bimbot, Frédéric

Author_Institution

France Telecom R&D, France

fYear

2005

fDate

16-19 Oct. 2005

Firstpage

90

Lastpage

93

Abstract

In this paper, the problem of one microphone source separation applied to singing voice extraction is studied. A probabilistic approach based on Gaussian mixture models (GMM) of the short time spectra of two sources is used. The question of source model adaptation is investigated in order to improve separation quality. A new adaptation method consisting in a filter adaptation technique via the maximum likelihood linear regression (MLLR) is presented with an associated filter-adapted training phase.

Keywords

acoustic signal processing; filtering theory; maximum likelihood estimation; microphones; regression analysis; source separation; Gaussian mixture models; filter adaptation technique; maximum likelihood linear regression; microphone singing voice separation; probabilistic approach; separation quality; singing voice extraction; source-adapted models; Adaptation model; Audio recording; Disk recording; MONOS devices; Maximum likelihood linear regression; Microphones; Nonlinear filters; Research and development; Source separation; Wiener filter;

fLanguage

English

Publisher

ieee

Conference_Titel

Applications of Signal Processing to Audio and Acoustics, 2005. IEEE Workshop on

Print_ISBN

0-7803-9154-3

Type

conf

DOI

10.1109/ASPAA.2005.1540176

Filename

1540176