Skip to main content
Applied and Numerical Harmonic Analysis

Supervised non-negative matrix factorization for audio source separation

Publication ,  Chapter
Sprechmann, P; Bronstein, AM; Sapiro, G
January 1, 2015

Source separation is a widely studied problem in signal processing. Despite the permanent progress reported in the literature it is still considered a significant challenge. This chapter first reviews the use of non-negative matrix factorization (NMF) algorithms for solving source separation problems, and proposes a new way for the supervised training in NMF. Matrix factorization methods have received a lot of attention in recent year in the audio processing community, producing particularly good results in source separation. Traditionally, NMF algorithms consist of two separate stages: a training stage, in which a generative model is learned; and a testing stage in which the pre-learned model is used in a high level task such as enhancement, separation, or classification. As an alternative, we propose a task-supervised NMF method for the adaptation of the basis spectra learned in the first stage to enhance the performance on the specific task used in the second stage. We cast this problem as a bilevel optimization program efficiently solved via stochastic gradient descent. The proposed approach is general enough to handle sparsity priors of the activations, and allow non-Euclidean data terms such as β-divergences. The framework is evaluated on speech enhancement.

Duke Scholars

DOI

Publication Date

January 1, 2015

Start / End Page

407 / 420
 

Citation

APA
Chicago
ICMJE
MLA
NLM
Sprechmann, P., Bronstein, A. M., & Sapiro, G. (2015). Supervised non-negative matrix factorization for audio source separation. In Applied and Numerical Harmonic Analysis (pp. 407–420). https://doi.org/10.1007/978-3-319-20188-7_16
Sprechmann, P., A. M. Bronstein, and G. Sapiro. “Supervised non-negative matrix factorization for audio source separation.” In Applied and Numerical Harmonic Analysis, 407–20, 2015. https://doi.org/10.1007/978-3-319-20188-7_16.
Sprechmann P, Bronstein AM, Sapiro G. Supervised non-negative matrix factorization for audio source separation. In: Applied and Numerical Harmonic Analysis. 2015. p. 407–20.
Sprechmann, P., et al. “Supervised non-negative matrix factorization for audio source separation.” Applied and Numerical Harmonic Analysis, 2015, pp. 407–20. Scopus, doi:10.1007/978-3-319-20188-7_16.
Sprechmann P, Bronstein AM, Sapiro G. Supervised non-negative matrix factorization for audio source separation. Applied and Numerical Harmonic Analysis. 2015. p. 407–420.

DOI

Publication Date

January 1, 2015

Start / End Page

407 / 420