DMCA
Latent variable decomposition of spectrograms for single channel speaker separation (2005)
| Venue: | in IEEE WASPAA |
| Citations: | 21 - 10 self |
BibTeX
@INPROCEEDINGS{Raj05latentvariable,
author = {Bhiksha Raj and Paris Smaragdis},
title = {Latent variable decomposition of spectrograms for single channel speaker separation},
booktitle = {in IEEE WASPAA},
year = {2005},
pages = {17--20}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
In this paper we present an algorithm for the separation of multiple speakers from mixed singlechannel recordings by latent variable decompoistion of the speech spectrogram. We model each magnitude spectral vector in the short-time Fourier transform of a speech signal as the outcome of a discrete random process that generates frequency bin indices. The districution of the process is modelled a mixture of multinomial distributions, such that the mixture weights of the component multinomials vary from analysis window to analysis window. The component multinomials are assumed to be speaker specific and are learnt from taining signals for each speaker. The distributions representing magnitude spectral vectors for the mixed signal are decomposed into mixtures of the multinomials for all component speakers. The frequency distribution, i.e. the spectrum for each speaker is reconstructed from this decomposition. Experimental results show that the proposed method is very effective at separating mixed signals.
Keyphrases
single channel speaker separation latent variable decomposition analysis window component multinomial magnitude spectral vector mixed signal speech spectrogram speech signal mixed singlechannel recording experimental result frequency distribution short-time fourier transform frequency bin index mixture weight multinomial distribution latent variable decompoistion component speaker discrete random process multiple speaker
