Regularized nonnegative matrix factorization using Gaussian mixture priors for supervised single channel source separation

Grais, Emad M.; Erdoğan, Hakan

doi:10.1016/j.csl.2012.09.002

Cited by 23 publications

(33 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Whereas speech enhancement focuses on stationary or nearly stationary backgrounds, speech separation refers to the case where the background is highly non-stationary and can contain difficult sources such as music or other speech signals. This problem has traditionally been addressed using model-based approaches, for example based on hidden Markov models (HMMs) [1], or non-negative matrix factorization (NMF) [2] and its extensions [3][4][5].…”

Section: Introductionmentioning

confidence: 99%

Phase-sensitive and recognition-boosted speech separation using deep recurrent neural networks

Erdoğan

Hershey

Watanabe

et al. 2015

2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

586

465

View full text Add to dashboard Cite

Separation of speech embedded in non-stationary interference is a challenging problem that has recently seen dramatic improvements using deep network-based methods. Previous work has shown that estimating a masking function to be applied to the noisy spectrum is a viable approach that can be improved by using a signalapproximation based objective function. Better modeling of dynamics through deep recurrent networks has also been shown to improve performance. Here we pursue both of these directions. We develop a phase-sensitive objective function based on the signal-to-noise ratio (SNR) of the reconstructed signal, and show that in experiments it yields uniformly better results in terms of signal-to-distortion ratio (SDR). We also investigate improvements to the modeling of dynamics, using bidirectional recurrent networks, as well as by incorporating speech recognition outputs in the form of alignment vectors concatenated with the spectral input features. Both methods yield further improvements, pointing to tighter integration of recognition with separation as a promising future direction.

show abstract

Section: Introductionmentioning

confidence: 99%

Phase-sensitive and recognition-boosted speech separation using deep recurrent neural networks

Erdoğan

Hershey

Watanabe

et al. 2015

2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

586

465

View full text Add to dashboard Cite

show abstract

“…In order to enforce this structure to the NMF model, Lefèvre et al [2] introduced a penalty term on the activations that favors sparsity at the group level. Grais and Erdogan [3] proposed a method that regularizes the columns of U by using Gaussian mixture models. Ş imşekli and Cemgil [4] presented a coupled factorization model where they hierarchically decomposed the activation matrix U into basis and weight matrices where the basis matrix aims to capture the hierarchical structure of the spectral dictionary.…”

Section: Introductionmentioning

confidence: 99%

Hierarchical and coupled non-negative dynamical systems with application to audio modeling

Şimşekli¹,

Roux

Hershey

2013

2013 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics

View full text Add to dashboard Cite

Many kinds of non-negative data, such as power spectra and count data, have been modeled using non-negative matrix factorization. Even though this modeling paradigm has yielded successful applications, it falls short when the data have certain hierarchical and temporal structure. In this study, we propose a novel dynamical system model that can handle these kinds of complex structures that often arise in non-negative data. We show that our model can be extended to handle heterogeneous data for data-driven regularization. We present convergence-guaranteed update rules for each latent factor. In order to assess the performance, we evaluate our model on the transcription of classical piano pieces, and show that it outperforms related models. We also illustrate that the performance can be further improved by making use of symbolic data.

show abstract

“…Research on NMF ranges from developing more accurate and efficient algorithms for factorization to its usage in various applications including acoustic/audio signal processing [2], source separation [3], hyperspectral analysis [4] and cognitive radio [5].…”

Section: Introductionmentioning

confidence: 99%

Sparsity promoted non-negative matrix factorization for source separation and detection

Wang

et al. 2014

2014 19th International Conference on Digital Signal Processing

View full text Add to dashboard Cite

The effectiveness of non-negative matrix factorization (NMF) depends on a suitable choice of the number of bases, which is often difficult to decide in practice. This paper imposes sparseness on the factorization coefficients in order to determine the number of bases automatically during the decomposition process.The benefit of sparse promotion for NMF is demonstrated through application to sound source separation as well as acoustic-based human fall detection under strong interference.

show abstract

Regularized nonnegative matrix factorization using Gaussian mixture priors for supervised single channel source separation

Cited by 23 publications

References 9 publications

Phase-sensitive and recognition-boosted speech separation using deep recurrent neural networks

Phase-sensitive and recognition-boosted speech separation using deep recurrent neural networks

Hierarchical and coupled non-negative dynamical systems with application to audio modeling

Sparsity promoted non-negative matrix factorization for source separation and detection

Contact Info

Product

Resources

About