Blind source separation for convolutive mixtures based on the joint diagonalization of power spectral density matrices

Mei, Tiemin; Mertins, Alfred; Yin, Fuliang; Xi, Jiangtao; Chicharo, J.F.

doi:10.1016/j.sigpro.2008.02.003

Cited by 20 publications

(16 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The first is to make the separation filters smooth in the frequency domain [4], [10], [11]. This may be achieved by limiting their lengths.…”

Section: A Region-growing Permutation Alignment Approach In Frequencymentioning

confidence: 99%

“…This may be achieved by limiting their lengths. It has been proved in [11] that the permutation ambiguity can be avoided if the separation filters are short enough relative to the FFT block size. Besides, the spectral continuity of separation filters can also be exploited to align the permutation [12].…”

Section: A Region-growing Permutation Alignment Approach In Frequencymentioning

confidence: 99%

“…For current bin and previous bin (supposing the correct permutation is known for bin ), select a permutation which maximizes the average value of correlation coefficients. This is expressed as (11) Now the average correlation coefficient for permutation is (12) Step 3) Divide the full frequency band into low and high bands, and process them separately in step 4-5. Step 4) Partition the high (low) frequency band into regions, with the highly related frequency bins assigned to the same region.…”

Section: B Proposed Permutation Alignment Schemementioning

confidence: 99%

See 2 more Smart Citations

A Region-Growing Permutation Alignment Approach in Frequency-Domain Blind Source Separation of Speech Mixtures

Wang

Ding

Yin

2011

IEEE Trans. Audio Speech Lang. Process.

Self Cite

View full text Add to dashboard Cite

Abstract-The convolutive blind source separation (BSS) problem can be solved efficiently in the frequency domain, where instantaneous BSS is performed separately in each frequency bin. However, the permutation ambiguity in each frequency bin should be resolved so that the separated frequency components from the same source are grouped together. To solve the permutation problem, this paper presents a new alignment method based on an inter-frequency dependence measure: the powers of separated signals. Bin-wise permutation alignment is applied first across all frequency bins, using the correlation of separated signal powers; then the full frequency band is partitioned into small regions based on the bin-wise permutation alignment result. Finally, region-wise permutation alignment is performed in a region-growing manner. The region-wise permutation correction scheme minimizes the spreading of the misalignment at isolated frequency bins to others, hence to improve permutation alignment. Experiment results in simulated and real environments verify the effectiveness of the proposed method. Analysis demonstrates that the proposed frequency-domain BSS method is computationally efficient.

show abstract

“…The first is to make the separation filters smooth in the frequency domain [4], [10], [11]. This may be achieved by limiting their lengths.…”

Section: A Region-growing Permutation Alignment Approach In Frequencymentioning

confidence: 99%

Section: A Region-growing Permutation Alignment Approach In Frequencymentioning

confidence: 99%

Section: B Proposed Permutation Alignment Schemementioning

confidence: 99%

See 1 more Smart Citation

A Region-Growing Permutation Alignment Approach in Frequency-Domain Blind Source Separation of Speech Mixtures

Wang

Ding

Yin

2011

IEEE Trans. Audio Speech Lang. Process.

Self Cite

View full text Add to dashboard Cite

show abstract

“…The answer to the question of why this FIOF-based BSS approach can avoid the permutation problem is that the time-domain impulse responses have been limited by length, which is equivalent to the frequency-domain smoothness constraints on the unmixing filters [6] [7].…”

Section: Model Of Convolutive Bssmentioning

confidence: 99%

On the generalization of blind source separation algorithms from instantaneous to convolutive mixtures

Mei

Mertins

Yin

2008

2008 5th IEEE Sensor Array and Multichannel Signal Processing Workshop

Self Cite

View full text Add to dashboard Cite

Many convolutive blind source separation (BSS) approaches are generalized from instantaneous BSS methods in either time or frequency domain. In this paper, we establish in a general way the inner relationship between the time-domain instantaneous BSS and the frequency-domain convolutive BSS. From this point of view, the time-domain approaches for instantaneous mixture separation are generalized to those for convolutive mixture separation in the frequency domain. Two examples are given to illustrate the feasibility of the proposed approach.

show abstract

“…In this paper, we choose the hybrid works of [3] and [8] in order to investigate the performance of information theoretic and SOS-based BSS methods for the cell-phone application. In these algorithms, the demixing filter coefficients are optimized in the time domain based on an integrated cost function.…”

Section: Introductionmentioning

confidence: 99%

Examination of convolutive blind source separation algorithms based on information theoretic criterion and second-order statistics for cell-phone application

Kheradmand

Sheikhzadeh

Raahemifar

et al. 2011

2011 24th Canadian Conference on Electrical and Computer Engineering(CCECE)

View full text Add to dashboard Cite

In this paper, we have investigated the performance of a blind source separation (BSS) method based on second order statistics (SOS) and an information theoretic BSS approach based on Kullback-Leibler (KL) divergence criterion for the data of cell-phone application. For this purpose, a hybrid time-frequency approach and two different SOS and KL-based algorithms within this framework are exploited. The performance of these algorithms is examined using the real-world data recorded for evaluating BSS algorithms for cell-phone application. Simulation results reveal that both algorithms have nearly the same performance after convergence. However, due to the simplicity and slightly better convergence properties, the SOS-based algorithm is preferred for practical purposes.

show abstract

Blind source separation for convolutive mixtures based on the joint diagonalization of power spectral density matrices

Cited by 20 publications

References 34 publications

A Region-Growing Permutation Alignment Approach in Frequency-Domain Blind Source Separation of Speech Mixtures

A Region-Growing Permutation Alignment Approach in Frequency-Domain Blind Source Separation of Speech Mixtures

On the generalization of blind source separation algorithms from instantaneous to convolutive mixtures

Examination of convolutive blind source separation algorithms based on information theoretic criterion and second-order statistics for cell-phone application

Contact Info

Product

Resources

About