NMF-Based Speech Enhancement Using Bases Update

Kwon, Kisoo; Shin, Jong Won; Kim, Nam Soo

doi:10.1109/lsp.2014.2362556

Cited by 60 publications

(17 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the time-frequency analysis of audio signals, however, the obtained basis may not be adequate to capture the temporal dependency of repeating patterns within the signal, and the success of these methods strongly relies on the prior knowledge of noise or speech or both, which limits implementations of the models. Recently, the online dictionary learning methods have been proposed in two aspects of implementing scheme [46][47][48][49][50] and circumventing the mismatch problem between the training and testing stages [24,52].…”

Section: Online Dictionary Learningmentioning

confidence: 99%

Dictionary Learning-Based Speech Enhancement

Duong¹,

Bui²,

Wang³

2019

Active Learning - Beyond the Future

View full text Add to dashboard Cite

This chapter presents an overview of dictionary learning-based speech enhancement methods. Specifically, we review the existing algorithms that employ sparse representation (SR), nonnegative matrix factorization (NMF), and their variations applying for speech enhancement. We emphasize that there are two stages in a speech enhancement system, namely learning dictionary and enhancement. The two scenarios of learning dictionary process, offline and online, are discussed carefully as well. We finally present some evaluation methods and suggest the future lines of work.

show abstract

Section: Online Dictionary Learningmentioning

confidence: 99%

Dictionary Learning-Based Speech Enhancement

Duong¹,

Bui²,

Wang³

2019

Active Learning - Beyond the Future

View full text Add to dashboard Cite

show abstract

“…Instead of directly using the estimated magnitude spectra in (3), a spectral gain function similar to the Wiener filter is adopted in [12] and [9]. In this scheme, the gain function is given by…”

Section: Nmf-based Audio Source Separationmentioning

confidence: 99%

“…Over the last few years, audio source separation has been one of the interesting topics in audio signal processing such as speech enhancement, speech recognition, music signal processing, and so on [1]- [9]. Data-representation methods and template-based approaches have been widely applied to audio source separation.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Target Source Separation Based on Discriminative Nonnegative Matrix Factorization Incorporating Cross-Reconstruction Error

Kwon

Shin

Kim

2015

IEICE Trans. Inf. & Syst.

Self Cite

View full text Add to dashboard Cite

SUMMARYNonnegative matrix factorization (NMF) is an unsupervised technique to represent nonnegative data as linear combinations of nonnegative bases, which has shown impressive performance for source separation. However, its source separation performance degrades when one signal can also be described well with the bases for the interfering source signals. In this paper, we propose a discriminative NMF (DNMF) algorithm which exploits the reconstruction error for the interfering signals as well as the target signal based on target bases. The objective function for training the bases is constructed so as to yield high reconstruction error for the interfering source signals while guaranteeing low reconstruction error for the target source signals. Experiments show that the proposed method outperformed the standard NMF and another DNMF method in terms of both the perceptual evaluation of speech quality score and signal-to-distortion ratio in various noisy environments. key words: nonnegative matrix factorization, discriminative basis, crossreconstruction error

show abstract

“…NMF takes the advantages of simple calculation, fast factorization and obvious physical properties of the results, which has attracted great attention in academia. Various signal separation algorithms based on NMF have been studied and proposed [11][12][13]. The basic NMF algorithm includes NMF algorithm based on Euclidean distance and NMF algorithm based on Kullback-Leibler (KL).…”

Section: Introductionmentioning

confidence: 99%

Signal Separation Based on Extended Least Squares

Gao¹,

Chen²,

Tian³

et al. 2018

Proceedings of the 2018 3rd Joint International Information Technology，Mechanical and Electronic Engineering Conference (JIMEC

View full text Add to dashboard Cite

In this paper, a new signal separation method is proposed in this paper to solve the problem of poor separation effect of mixed signals in strong noise environment. Based on least squares (LS), the improved optimization model is extended by Kullback-Leibler dispersion to remove the random noise. Theoretical analysis and simulation experiments show that the algorithm proposed in this paper is superior to the existing algorithm in estimating the source signal, especially when the mixed signal is completely immersed in noise, the recovery effect of the source signal is more obvious than the existing algorithm.

show abstract

NMF-Based Speech Enhancement Using Bases Update

Cited by 60 publications

References 20 publications

Dictionary Learning-Based Speech Enhancement

Dictionary Learning-Based Speech Enhancement

Target Source Separation Based on Discriminative Nonnegative Matrix Factorization Incorporating Cross-Reconstruction Error

Signal Separation Based on Extended Least Squares

Contact Info

Product

Resources

About