A new regularized forward blind source separation algorithm for automatic speech quality enhancement

This paper addresses the problem of speech enhancement and acoustic noise reduction by adaptive filtering algorithms in a moving car through blind source separation (BSS) structures. In this paper we propose a new regularized forward blind source separation (RFBSS) algorithm that does not need voice activity detection (VAD) systems, and allows getting efficient speech enhancement performances with low complexity.     

Real-time blind source separation system with applications to distant speech recognition

A real-time BSS system based on DUET was developed and implemented in order to assess its potential as the front-end for a DSR engine. The system uses only two closely-spaced standard omni-directional microphones and a computer soundcard and was developed for low reverberation environments with several human speakers and different noise sources.     

子带t分布的快速独立向量分析在语声盲源 分离中的应用*

为了提高独立向量分析算法在盲语声分离任务中的分离性能,降低算法计算复杂度,并改善目前尚未完全解决的顺序模糊性的问题,该文提出一种基于子带t分布的快速独立向量分析算法。在声源模型方面,该算法首先利用语声信号重尾分布的特性,假设声源概率密度函数服从t分布,同时采用子带建模的方法来增强同一声源相邻频点的高阶依赖性,进而减轻频点间的顺序不一致问题。在空间模型方面,该算法采用秩1更新的方式估计声源信号,避免矩阵求逆操作和分离矩阵的估计,从而降低计算复杂度。实验结果表明,与现有的基于独立向量分析的盲源分离算法相比,该算法能够在相同的迭代次数下取得更优的语声分离性能。     

Using blind source separation techniques to improve speech recognition in bilateral cochlear implant patients

Bilateral cochlear implants seek to restore the advantages of binaural hearing by improving access to binaural cues. Bilateral implant users are currently fitted with two processors, one in each ear, operating independent of one another. In this work, a different approach to bilateral processing is explored based on blind source separation (BSS) by utilizing two implants driven by a single processor. Sentences corrupted by interfering speech or speech-shaped noise are presented to bilateral cochlear implant users at 0 dB signal-to-noise ratio in order to evaluate the performance of the proposed BSS method. Subjects are tested in both anechoic and reverberant settings, wherein the target and masker signals are spatially separated. Results indicate substantial improvements in performance in both anechoic and reverberant settings over the subjects' daily strategies for both masker conditions and at various locations of the masker. It is speculated that such improvements are due to the fact that the proposed BSS algorithm capitalizes on the variations of interaural level differences and interaural time delays present in the mixtures of the signals received by the two microphones, and exploits that information to spatially separate the target from the masker signals.     

瞬时混合盲源分离和最优波束形成的关系研究

最优波束形成和盲源分离是近几十年迅速发展起来的两种重要阵列信号处理方法,在瞬时混合情况下它们具有相似的信号模型,所以本文研究了二者之间的关系。指出,当不存在噪声时,多数盲源分离算法构造的是最优波束形成;当存在噪声时,可以利用盲源分离算法估计的信号导向向量设计最优波束形成。计算机仿真实验验证了这一结论。     

Hamiltonian fluid reductions of electromagnetic drift-kinetic equations for an arbitrary number of moments

We present an infinite family of Hamiltonian electromagnetic fluid models for plasmas, derived from drift-kinetic equations. An infinite hierarchy of fluid equations is obtained from a Hamiltonian drift-kinetic system by taking moments of a generalized distribution function and using Hermite polynomials as weight functions of the velocity coordinate along the magnetic guide field. Each fluid model is then obtained by truncating the hierarchy to a finite number N+1$N+1$ of equations by means of a closure relation. We show that, for any positive N$N$, a linear closure relation between the moment of order N+1$N+1$ and the moment of order N$N$ guarantees that the resulting fluid model possesses a Hamiltonian structure, thus respecting the Hamiltonian character of the parent drift-kinetic model. An orthogonal transformation is identified which maps the fluid moments to a new set of dynamical variables in terms of which the Poisson brackets of the fluid models become a direct sum and which unveils remarkable dynamical properties of the models in the two-dimensional (2D) limit. Indeed, when imposing translational symmetry with respect to the direction of the magnetic guide field, all models belonging to the infinite family can be reformulated as systems of advection equations for Lagrangian invariants transported by incompressible generalized velocities. These are reminiscent of the advection properties of the parent drift-kinetic model in the 2D limit and are related to the Casimirs of the Poisson brackets of the fluid models. The Hamiltonian structure of the generic fluid model belonging to the infinite family is illustrated treating a specific example of a fluid model retaining five moments in the electron dynamics and two in the ion dynamics. We also clarify the connection existing between the fluid models of this infinite family and some fluid models already present in the literature.     

Thermodynamics of solid solutions with an arbitrary number of sites in sublattices

Specifics of thermodynamic calculations for solid solutions is discussed, when the number of sites in sublattices, which may be occupied by the atoms, is not fixed.Translated from Izvestiya Vysshikh Uchebnykh Zavedenii, Fizika, No. 5, pp. 29–31, May, 1989.     

一种混沌信号的盲分离方法

多个混合混沌信号的分离，在混沌及其应用中是一个重要的问题.提出了一种线性混合的混沌信号的瞬时盲分离方法，它利用了各个混沌信号源之间的互不相关性，在未知混合矩阵和混沌方程的情况下，通过求解特征向量的方法从观测量中直接估计出解混合矩阵，以重构出源混沌信号.仿真结果表明，即使在低信噪比情况下，该方法仍可以有效地从噪声背景中分离出多个混合的混沌信号. 关键词： 混沌 盲分离 信噪比 特征向量     

A transformation theory for camouflaging arbitrary heat sources

Camouflage devices have attracted intensive research interest for their significant applications. However, most camouflage devices are specifically designed according to target heat sources. Here, by applying the transformation thermotics approach, we develop a coordinate transformation, and design an unspecific camouflage device which can camouflage arbitrary heat sources into a circular one with an anisotropic shell. We verify the ability of our unspecific camouflage device with both steady and transient simulations. We also find the “apparent negative thermal conductivity” under certain conditions without violating the second law of thermodynamics. To ensure the feasibility, we further put forward the effective medium approximation for sample fabrication, and only two natural materials are required. Our results have relevance to the different applications of infrared misleading, uniform heating, and so on; they may also provide guidance to the research on other diffusive fields, such as magnetostatic and electrostatic fields.     

Test of nonlocality for a continuous-variable state based on an arbitrary number of measurement outcomes

We propose a scheme to test Bell's inequalities for an arbitrary number of measurement outcomes on entangled continuous-variable (CV) states. The Bell correlation functions are expressible in terms of phase-space quasiprobability functions with complex ordering parameters, which can experimentally be determined via a local CV-qubit interaction. We demonstrate that CV systems can give higher violations of these Bell's inequalities than of the ones developed for two-outcome observables.     

Coalescence of an arbitrary number of photons into one at an electron in the plane-wave-approximation model

The case when an arbitrary number of photons moving in parallel coalesce to a single photon at an electron is considered for the first time. The calculation is based on the plane-wave-approximation model, describing the interaction between the quantum field of a plane electromagnetic wave and an electron. All the matrix elements determining the differential probability of the process are accurately calculated. As an illustration, results are given for the coalescence of three photons to one at an electron.Translated from Izvestiya Vysshikh Uchebnykh Zavedenii, Fizika, No. 12, pp. 117–124, December, 1978.     

非线性混合模式的语音盲分离算法

本文针对FIR非线性混合模型，基于最大熵算法，提出了一种以高斯混合模式概率密度函数估计替代传统对数化概率密度估计的盲分离算法，以偶函数为非线性激活函数，采用最大期望（EM）迭代算法推导了分离算法的权向量迭代公式，通过模拟仿真实验结果与传统的最大熵和高阶累积量方法比较，新算法提高了收敛速度，并有效地完成了非线性语音分离任务，抑制了干扰语音信号的影响，提高了输出信噪比。     

一种适用于微弱信号提取的盲源分离算法

本文首次提出了适用于微弱信号提取的盲源分离算法,这种方法是在常用的自然梯度串行更新算法基础上采用了有监督的机制。通过观察两个信号互相关的程度自动判别激活函数的种类,实现了超高斯亚高斯强弱信号混合下的盲源分离。信号仿真表明,有监督的盲源分离技术收敛速度较快,精度较高。     

一种基于聚类的门控卷积网络语声分离方法*

基于深度聚类的语音分离方法已被证明能有效地解决混合语音中说话人输出标签排列的问题,然而,现有关于聚类进行说话人分离方法,大多数是优化嵌入使每个源的重建误差最小化。本文以时域卷积网络（ConvTasNet）为基础网络,设计了一种改进基于聚类的门控卷积（Gate-conv Cluster）语音分离方法,在时域上通过堆叠的门控卷积网络,实现端到端深度聚类的源分离。该框架将非线性门控激活用于时域卷积网络中,提取语音信号的深层次特征;同时在高维特征空间中聚类对语音信号的特征进行表示和划分,为恢复不同信号源提供了一个长期的说话者表示信息。该框架解决了说话人输出标签排列问题并对语音信号的长期依赖性进行建模。通过华尔街日报数据集进行实验得出,该方法在SDRi（信源失真比）和Si-SNR（尺度不变信源噪声比）指标上分别达到了16.72 dB和16.33 dB的效果。     

Soliton structures of a wave field with an arbitrary number of oscillations in nonresonance media

A new class of solitary solutions for a wave field is found. This class describes soliton-like structures of a circularly polarized radiation that propagate in a nonresonance medium and which involve an arbitrary number of field oscillations. A feature peculiar to these solutions is that they undergo a smooth transformation from solitons of the Schrödinger type, which correspond to long pulses involving many oscillations, to extremely short visible pulses, which, in fact, do not extend beyond one period. Realizability of such soliton structures is considered for a field of linear polarization, and their structural stability is shown numerically.     

Fast blind extraction of arbitrary unknown phase shifts by an iterative tangent approach in generalized phase-shifting interferometry

A novel fast convergent algorithm to extract arbitrary unknown phase shifts in generalized phase-shifting interferometry (PSI) is proposed and verified by a series of computer simulations. In this algorithm an error function is introduced and then the unknown phase shifts are found by an iterative tangent approach. In combination with the statistical method, this algorithm can give the most exact results in the fewest iteration steps. It can be used for generalized PSI of arbitrary frames for both smooth and diffusing objects and can usually reach the exact phase shifts with only four or five iterations for three- or four-frame PSI.     

Ensemble average of an arbitrary number of pairs of different eigenvalues using Grassmann integration

An identity satisfied by the eigenvalues of a real-symmetric matrix and an integral representation of a determinant using Grassmann variables are used to show that the ensemble average ofS different pairs of eigenvalues of a GOE is given by (–1) ^S 2^{–S –1/2}(S+1/2).