Underdetermined source separation with structured source priors

Abstract : We consider the source extraction problem for stereo instantaneous musical mixtures with more than two sources. We prove that usual separation methods based only on spatial diversity have performance limitations when the sources overlap in the time-frequency plane. We propose a new separation scheme combining spatial diversity and structured source priors. We present possible priors based on nonlinear Independent Subspace Analysis (ISA) and Hidden Markov Models (HMM), whose parameters are learnt on solo musical excerpts. We show with an example that they actually improve the separation performance.
Liste complète des métadonnées

Cited literature [10 references]  Display  Hide  Download

https://hal.inria.fr/inria-00544694
Contributor : Emmanuel Vincent <>
Submitted on : Wednesday, December 8, 2010 - 4:43:40 PM
Last modification on : Thursday, March 21, 2019 - 2:20:03 PM
Document(s) archivé(s) le : Thursday, March 10, 2011 - 12:29:16 PM

File

vincent_ICA04.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : inria-00544694, version 1

Collections

Citation

Emmanuel Vincent, Xavier Rodet. Underdetermined source separation with structured source priors. 5th Int. Conf. on Independent Component Analysis and Blind Signal Separation (ICA), Sep 2004, Granada, Spain. pp.327--332. ⟨inria-00544694⟩

Share

Metrics

Record views

71

Files downloads

126