This paper addresses the problem of speech enhancement and acoustic noise reduction by blind structures. Recently, the backward blind source separation (BBSS) structure has shown efficiency in cancelling the acoustic ...
详细信息
This paper addresses the problem of speech enhancement and acoustic noise reduction by blind structures. Recently, the backward blind source separation (BBSS) structure has shown efficiency in cancelling the acoustic noise and improving corrupted speech signals form very noisy observations without any a priori information of source signals. In this paper, we propose a new algorithm based on the combination between the BBSS structure and the simplified fast transversal filter (SFTF) algorithm. The proposed two-channel simplified fast transversal filter (TCSFTF) algorithm succeeded an important blind improvement of steady state and convergence speed performances in diverse noisy situations when only the noisy signals are known. The performances of the new TCSFTF algorithm are compared with four state-of-the-art algorithms in different noisy conditions. This comparison is evaluated in terms of cepstral distance (CD), system mismatch (SM), segmental signal to noise ratio (SegSNR), and segmental mean square error (SegMSE) criteria. (C) 2018 Elsevier Ltd. All rights reserved.
In this exposition, a simple practical adaptive algorithm is developed for efficient and accurate reconstruction of Neumann boundary data in the inverse Stefan problem, which is a highly nontrivial task. Primarily, th...
详细信息
In this exposition, a simple practical adaptive algorithm is developed for efficient and accurate reconstruction of Neumann boundary data in the inverse Stefan problem, which is a highly nontrivial task. Primarily, this algorithm detects the satisfactory location of the source points from the boundary in reconstructing the boundary data in the inverse Stefan problem efficiently. To deal with the ill-conditioning of the matrix generated by the MFS, we use Tikhonov regularization and the algorithm is designed in such a way that the optimal regularization parameter is detected automatically without any use of traditional methods like the discrepancy principle, the L-curve criterion or the generalized cross-validation (GCV) technique. Furthermore, this algorithm can be thought of as an alternative to the concept of Beck's future temperatures for obtaining stable and accurate fluxes, but without it being necessary to specify data on any future time interval. A MATLAB code for the algorithm is discussed in more-than-usual detail. We have studied the effects of accuracy and measurement error (random noise) on both optimal location and number of source points. The effectiveness of the proposed algorithm is shown through several test problems, and numerical experiments indicate promising results. (C) 2018 Elsevier B.V. All rights reserved.
This paper proposes an efficient algorithm for impulsive active noise control (IANC) systems. The impulsive sources cannot be modeled by Gaussian distribution, and hence the standard adaptive algorithm based on second...
详细信息
ISBN:
(纸本)9781538673928
This paper proposes an efficient algorithm for impulsive active noise control (IANC) systems. The impulsive sources cannot be modeled by Gaussian distribution, and hence the standard adaptive algorithm based on second order statistics would give poor performance or even fail to converge. One solution is to derive adaptive algorithm by minimizing a fractional low order moment, resulting in the famous filtered-x least mean p-power (FxLMP) algorithm. The proposed algorithm discussed in this paper is based on a previously proposed generalized FxLMP algorithm. The key idea here is to introduce a variable step-size using a convex-combination approach. A large value is used at the start-up of IANC system to achieve a fast convergence speed. As the AINC system converges, the step-size automatically reduces to a small value to improve the steady-state noise reduction performance. Simulations demonstrate the effectiveness of the proposed algorithm.
A fast adaptive method to eliminate a single sinusoidal interference of known frequency is proposed. The coupled-oscillator is used to generate the reference signal instead of look-up table. Unlike the least mean squa...
详细信息
ISBN:
(纸本)9781538623176
A fast adaptive method to eliminate a single sinusoidal interference of known frequency is proposed. The coupled-oscillator is used to generate the reference signal instead of look-up table. Unlike the least mean square (LMS) based adaptive algorithm, the weighted least squared error criterion is employed instead to evaluate the system parameters. It has been shown that the proposed method can quickly eliminate unwanted interference as compared to previous methods. Moreover, the extensive simulations have been done and revealed that proposed algorithm yields a constant output signal to interference ratio (SIR) of about 20 dB regardless of input SIR.
In this paper, we address the problem of speech enhancement by adaptive filtering algorithm. A particular attention has been made for the backward blind source separation (BBSS) algorithm and its use in speech enhance...
详细信息
ISBN:
(纸本)9781450363921
In this paper, we address the problem of speech enhancement by adaptive filtering algorithm. A particular attention has been made for the backward blind source separation (BBSS) algorithm and its use in speech enhancement application. In this paper, we propose to implement the BBSS algorithm in the Wavelet-domain. The proposed backward wavelet BBSS (WBBSS) algorithm is then used in speech enhancement test. The new WBBSS shows better performances in terms of convergence speed and steady state in comparison with BBSS one. The obtained results have been evaluated in terms of segmental SNR and cepstral distance criteria and confirm the best performance of the proposed WBBSS algorithm.
Data compression is a key part of database management systems for storage saving and performance enhancement. In column-oriented databases, records belong to the same attribute are stored nearby, and the similarity be...
详细信息
ISBN:
(纸本)9781538662434
Data compression is a key part of database management systems for storage saving and performance enhancement. In column-oriented databases, records belong to the same attribute are stored nearby, and the similarity between these records increases the compressibility of data and expands the range of compression algorithms to choose. Since different data compression algorithms process data in different manners, the achieved compression ratio varies significantly. This makes it worth studying the choice of compression algorithms depending on features of data to be compressed. As Recurrent Neural Networks is good at processing and making predictions based on series of data, we propose a Long-Short Term Memory network based model to select compression algorithm for input data blocks adaptively. Given a typical database benchmark, we implemented our model to formulate compression strategies for each data block and managed to reduce at most 15% storage size than using a single compression algorithm scheme.
A system for wireless data transmission is presented based on the use of an adaptive algorithm for processing spatial-temporal signals using antenna arrays. To solve the problem of increasing the maximum data transmis...
详细信息
ISBN:
(纸本)9781538627822
A system for wireless data transmission is presented based on the use of an adaptive algorithm for processing spatial-temporal signals using antenna arrays. To solve the problem of increasing the maximum data transmission speed and providing a low probability of error in data transmission under multipath conditions, technologies such as MIMO (Multiple input-Multiple output) and OFDM (Orthogonal Frequency Division Multiplexing) were used. The adaptation process is based on the formation of an equivalent directional pattern of the amplitude array antenna in the direction of the arrival of the signal with the highest power along one of the paths in the channel. The results of the research showed that application of the adaptive algorithm on the receiving side allows to significantly reduce the probability of bit error, thereby increasing the throughput in the wireless data channel. The adaptation process is based on the formation of an equivalent directional pattern of the amplitude array antenna in the direction of the arrival of the signal with the highest power along one of the paths in the channel. The results of the research showed that application of the adaptive algorithm on the receiving side allows to significantly reduce the probability of bit error, thereby increasing the throughput in the wireless data channel.
The paper discusses an adaptive algorithm for processing hydroacoustic signals of antenna arrays in real time. We perform mathematical analysis of the adaptation algorithm. The algorithm is based on an iterative proce...
详细信息
ISBN:
(纸本)9781509048656
The paper discusses an adaptive algorithm for processing hydroacoustic signals of antenna arrays in real time. We perform mathematical analysis of the adaptation algorithm. The algorithm is based on an iterative procedure for finding weight coefficients. We provide recommendations for algorithm application under different conditions.
Generative adversarial networks (GANs) have shown significant progress in generating high-quality visual samples, however they are still well known both for being unstable to train and for the problem of mode collapse...
详细信息
Generative adversarial networks (GANs) have shown significant progress in generating high-quality visual samples, however they are still well known both for being unstable to train and for the problem of mode collapse, particularly when trained on data collections containing a diverse set of visual objects. In this paper, we propose an adaptive-step Generative Adversarial Network (-GAN), which is designed to mitigate the impact of instability and saturation in the original by dynamically adjusting the ratio of the training steps of both the generator and discriminator. To accomplish this, we track and analyze stable training curves of relatively narrow datasets and use them as the target fitting lines when training more diverse data collections. Furthermore, we conduct experiments on the proposed procedure using several optimization techniques (e.g., supervised guiding from previous stable learning curves with and without momentum) and compare their performance with that of state-of-the-art models on the task of image synthesis from datasets consisting of diverse images. Empirical results demonstrate that Ak-GAN works well in practice and exhibits more stable behavior than regular GANs during training. A quantitative evaluation has been conducted on the Inception Score (IS) and the relative inverse Inception Score (RIS);compared with regular GANs, the former has been improved by 61% and 83%, and the latter by 21% and 60%, on the CelebA and the Anime datasets, respectively.
In model free adaptive control (MFAC), a virtual equivalent dynamic linearized model is built. The linearization length constants (LLCs) of the virtual equivalent dynamic linearized model are selected by the practitio...
详细信息
In model free adaptive control (MFAC), a virtual equivalent dynamic linearized model is built. The linearization length constants (LLCs) of the virtual equivalent dynamic linearized model are selected by the practitioner based on experience. In this paper, the optimal LLCs are investigated, and compact model free adaptive control (CMFAC) is introduced for a class of unknown discrete-time nonlinear systems. Compared with MFAC, the proposed CMFAC does not need to consider the values of LLCs, and the optimal LLCs are decided by the desired tracking error of systems. Simulation experiments are taken, and the simulation results indicate that the proposed control algorithm is effective and can achieve asymptotic tracking.
暂无评论