Blind signal separation and speech recognition in the frequency domain

Koutras, Athanasios; Dermatas, Evangelos; Kokkinakis, G.

doi:10.1109/icecs.1999.812314

Cited by 7 publications

(5 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This recognition, as well as the Signal to Interference Ratio (SIR) results [8], although promising for the field of speech recognition, showed a certain incompetence in performing efficiently due to the stochastic nature of the gradient descent optimization technique that was used. Furthermore, the convergence behavior of the above optimization methods depends on the step size choice and the initial separation filter coefficient values.…”

Section: Introductionmentioning

confidence: 94%

The Hybrid Genetic Algorithm for Blind Signal Separation

Shyr

2006

Neural Information Processing

View full text Add to dashboard Cite

Abstract. In this paper, a hybrid genetic algorithm for blind signal separation that extracts the individual unknown independent source signals out of given linear signal mixture is presented. The proposed method combines a genetic algorithm with local search and is called the hybrid genetic algorithm. The implemented separation method is based on evolutionary minimization of the separated signal cross-correlation. The convergence behaviour of the network is demonstrated by presenting experimental separating signal results. A computer simulation example is given to demonstrate the effectiveness of the proposed method. The hybrid genetic algorithm blind signal separation performance is better than the genetic algorithm at directly minimizing the Kullback-Leibler divergence. Eventually, it is hopeful that this optimization approach can be helpful for blind signal separation engineers as a simple, useful and reasonable alternative.

show abstract

Section: Introductionmentioning

confidence: 94%

The Hybrid Genetic Algorithm for Blind Signal Separation

Shyr

2006

Neural Information Processing

View full text Add to dashboard Cite

show abstract

“…Therefore, this last matrix can be written by means of the Z transform as Several ICA-based algorithms can be applied in this case to carry the separation process out. In the context of audio, the convolutive problem is classically analysed by means of second order statistics (Ehlers & Schuster, 1997;Ikram & Morgan, 2001;Kawamoto et al, 1999;Sahlin & Broman, 1998;Weinstein et al, 1993), higher order statistics (Charkani & Deville, 1999;Jutten et al, 1991b;Nguyen et al, 1992;Nguyen & Jutten, 1995;Van Gerven et al, 1994) and probability density function Koutras et al, 1999;Lee et al, 1997a;.…”

Section: Blind Implicit Source Separation -A New Concept In Bss Theorymentioning

confidence: 99%

Independent Component Analysis for Audio and Biosignal Applications

Naik¹

2012

View full text Add to dashboard Cite

show abstract

“…Numerous methods have been used in the published literature to indicate separation performance. The list includes mean-square-error (MSE) [20], [21] bit per symbol error rate [22], [23], Frobenius distance [24], multichannel row and multichannel column intersymbol interference (ISI) [25], plot of global mixing filter responses [26]- [28], signal-to-noise ratio (SNR) [29]- [31], interference-to-signal ratio (ISR) [19], signal-to-interference ratio (SIR) [32]- [38], one-at-a-time SIR [8], ISI [39]- [41], bias and standard deviation of filter coefficients [42], [43], plot of estimated sources in the time or frequency domain [44]- [46], hand-segmented SIR [47], automatic speech recognition rate [48], [33], [34], and the mean opinion score [18]. Several of these are not ideal for comparisons because they are either subjective, such as the plots and the mean opinion score, or require knowledge of the mixing filters, which makes them inapplicable for real mixtures.…”

Section: Figures Of Merit For Separationmentioning

confidence: 99%

Experimental upper bound for the performance of convolutive source separation methods

Hild

Erdoğmuş

Prı́ncipe

2006

IEEE Trans. Signal Process.

View full text Add to dashboard Cite

Abstract-An important problem in the field of blind source separation (BSS) of real convolutive mixtures is the determination of the role of the demixing filter structure and the criterion/optimization method in limiting separation performance. This issue requires the knowledge of the optimal performance for a given structure, which is unknown for real mixtures. Herein, the authors introduce an experimental upper bound on the separation performance for a class of convolutive blind source separation structures, which can be used to approximate the optimal performance. As opposed to a theoretical upper bound, the experimental upper bound produces an estimate of the optimal separating parameters for each dataset in addition to specifying an upper bound on separation performance. Estimation of the upper bound involves the application of a supervised learning method to the set of observations found by recording the sources one at a time. Using the upper bound, it is demonstrated that structures other than the finite-impulse-response (FIR) structure should be considered for real (convolutive) mixtures, there is still much room for improvement in current convolutive BSS algorithms, and the separation performance of these algorithms is not necessarily limited by local minima.Index Terms-Blind source separation (BSS), convolutive source separation, independent components analysis (ICA), speech enhancement, upper bound.

show abstract

Blind signal separation and speech recognition in the frequency domain

Cited by 7 publications

References 4 publications

The Hybrid Genetic Algorithm for Blind Signal Separation

The Hybrid Genetic Algorithm for Blind Signal Separation

Independent Component Analysis for Audio and Biosignal Applications

Experimental upper bound for the performance of convolutive source separation methods

Contact Info

Product

Resources

About