Supervised Monaural Speech Enhancement Using Complementary Joint Sparse Representations

Luo, You Xin; Bao, Guangzhao; Xu, Yangfei; Ye, Zhongfu

doi:10.1109/lsp.2015.2509480

Cited by 21 publications

(7 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…OVL is a linear combination of three objective measures including PESQ, log likelihood ratio (LLR) [21] and weighted-slope spectral (WSS) distance [22]. Moreover, we compare all the experimental results of our proposed method with five state-of-the-art speech enhancement methods involving GDL [6], CJSR [9], RPCA [11], LSLD [13] and CLSMD [10].…”

Section: Resultsmentioning

confidence: 99%

“…Different from the GDL method, the complementary joint sparse representation (CJSR) method [9] utilizes the noisy speech magnitude spectra in the learning stage together with the clean speech magnitude spectra and the noise magnitude spectra, respectively, to form two different training datasets to train two distinct mixture dictionaries. The trained mixture dictionaries offer latent mappings from the noisy speech to the clean speech and noise, respectively, which enable the sparse coding algorithm in the enhancement stage to improve the accuracy of both speech and noise estimations in the frequency domain.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Speech Enhancement Based on Dictionary Learning and Low-Rank Matrix Decomposition

Zhu

Champagne

2019

IEEE Access

View full text Add to dashboard Cite

In this paper, a new speech enhancement method is proposed based on dictionary learning and low-rank matrix decomposition with the objective to improve speech quality and intelligibility. In both the learning stage and the enhancement stage of the proposed method, a new noise model is employed to capture the noise characteristics with a combination of a low-rank matrix and an overcomplete noise dictionary. In the learning stage, an unsupervised dictionary learning algorithm is first proposed based on this new noise model to train a noise dictionary with low mutual coherence to the clean speech. Then, a supervised low-rank matrix decomposition algorithm is proposed to extract the low-rank component of noise using the clean speech, noise, and noisy speech training data sets. The estimated low-rank matrix is clustered through the K-means method to generate a codebook which provides useful information for the enhancement stage. Finally, in the enhancement stage, by using the well-trained dictionaries and the reference codebook from the learning stage, a decomposition algorithm for the low-rank matrix and sparse component is developed to effectively estimate the clean speech from the noisy speech. The experimental results show that our proposed method achieves better enhancement performance than some state-of-the-art reference methods in terms of four objective performance evaluation measures especially in low signal-to-noise ratio adverse environments. INDEX TERMS Clustering, noise model, speech enhancement, supervised low-rank matrix decomposition, unsupervised dictionary learning.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Speech Enhancement Based on Dictionary Learning and Low-Rank Matrix Decomposition

Zhu

Champagne

2019

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Then, the clean speech PSD estimate is found based on multiplication of the sparse code with the dictionary. Luo et al [17] proposed a complementary joint sparse representation, where two mixture dictionaries to model "mixture and speech" and "mixture and noise" are added to the Generative Dictionary Learning (GDL) problem formulation, and sparse codes of clean speech are forced to represent the noisy mixture on the mixture and clean speech sub-dictionary, while the sparse codes for the noise are forced to represent the noisy mixture on the mixture and noise sub-dictionary. Though this joint sparse representation alleviates, to some extent, the problem of source confusion, it is characterized by high complexity due to the need of learning four subdictionaries instead of two.…”

Section: Introductionmentioning

confidence: 99%

Incoherent Discriminative Dictionary Learning for Speech Enhancement

Shaheen¹,

Dakkak²,

Wainakh³

2018

JTIT

View full text Add to dashboard Cite

Speech enhancement is one of the many challenging tasks in signal processing, especially in the case of nonstationary speech-like noise. In this paper a new incoherent discriminative dictionary learning algorithm is proposed to model both speech and noise, where the cost function accounts for both “source confusion” and “source distortion” errors, with a regularization term that penalizes the coherence between speech and noise sub-dictionaries. At the enhancement stage, we use sparse coding on the learnt dictionary to ﬁnd an estimate for both clean speech and noise amplitude spectrum. In the ﬁnal phase, the Wiener ﬁlter is used to reﬁne the clean speech estimate. Experiments on the Noizeus dataset, using two objective speech enhancement measures: frequency-weighted segmental SNR and Perceptual Evaluation of Speech Quality (PESQ) demonstrate that the proposed algorithm outperforms other speech enhancement methods tested.

show abstract

“…Then the target type was determined by minimizing the reconstruction error of every type of training sample. Sparse representation has been used extensively in various recognition tasks, such as facial recognition, speech recognition, and hyperspectral image classification [ 10 , 11 , 12 ]. Sparse representation based recognition has gained in-depth development in accordance to different requirements of image recognition.…”

Section: Introductionmentioning

confidence: 99%

Aspect-Aided Dynamic Non-Negative Sparse Representation-Based Microwave Image Classification

Zhang

Yang

Liu

et al. 2016

Sensors

View full text Add to dashboard Cite

Classification of target microwave images is an important application in much areas such as security, surveillance, etc. With respect to the task of microwave image classification, a recognition algorithm based on aspect-aided dynamic non-negative least square (ADNNLS) sparse representation is proposed. Firstly, an aspect sector is determined, the center of which is the estimated aspect angle of the testing sample. The training samples in the aspect sector are divided into active atoms and inactive atoms by smooth self-representative learning. Secondly, for each testing sample, the corresponding active atoms are selected dynamically, thereby establishing dynamic dictionary. Thirdly, the testing sample is represented with ℓ1-regularized non-negative sparse representation under the corresponding dynamic dictionary. Finally, the class label of the testing sample is identified by use of the minimum reconstruction error. Verification of the proposed algorithm was conducted using the Moving and Stationary Target Acquisition and Recognition (MSTAR) database which was acquired by synthetic aperture radar. Experiment results validated that the proposed approach was able to capture the local aspect characteristics of microwave images effectively, thereby improving the classification performance.

show abstract

Supervised Monaural Speech Enhancement Using Complementary Joint Sparse Representations

Cited by 21 publications

References 25 publications

Speech Enhancement Based on Dictionary Learning and Low-Rank Matrix Decomposition

Speech Enhancement Based on Dictionary Learning and Low-Rank Matrix Decomposition

Incoherent Discriminative Dictionary Learning for Speech Enhancement

Aspect-Aided Dynamic Non-Negative Sparse Representation-Based Microwave Image Classification

Contact Info

Product

Resources

About