An iterative inversion approach to blind source separation

We present an iterative inversion (II) approach to blind source separation (BSS). It consists of a quasi-Newton method for the resolution of an estimating equation obtained from the implicit inversion of a robust estimate of the mixing system. The resulting learning rule includes several existing algorithms for BSS as particular cases giving them a novel and unified interpretation. It also provides a justification of the Cardoso and Laheld (1996) step size normalization. The II method is first presented for instantaneous mixtures and then extended to the problem of blind separation of convolutive mixtures. Finally, we derive the necessary and sufficient asymptotic stability conditions for both the instantaneous and convolutive methods to converge.     

Blind Source Separation of Postnonlinear Convolutive Mixture

In this paper, a novel solution is developed to solve blind source separation of postnonlinear convolutive mixtures. The proposed model extends the conventional linear instantaneous mixture model to include both convolutive mixing and postnonlinear distortion. The maximum-likelihood (ML) approach solution based on the expectation-maximization (EM) algorithm is developed to estimate the source signals and the parameters in the proposed nonlinear model. In the proposed solution, the sufficient statistics associated with the source signals are estimated in the E-step, while the model parameters are optimized through these statistics in the M-step. However, the complication resulted from the postnonlinear function associated with the mixture renders these statistics difficult to be formulated in a closed form and hence causes intractability in the parameter optimization. A computationally efficient algorithm is proposed which uses the extended Kalman smoother (EKS) to facilitate the E-step tractable and a set of self-updated polynomials is used as the nonlinearity estimator to facilitate closed form estimations of the parameters in the M-step. The theoretical foundation of the proposed solution has been rigorously developed and discussed in details. Both simulations and recorded speech signals have been carried out to verify the success and efficacy of the proposed algorithm. Remarkable improvement has been obtained when compared with the existing algorithms.     

Model selection for convolutive ICA with an application to spatiotemporal analysis of EEG

We present a new algorithm for maximum likelihood convolutive independent component analysis (ICA) in which components are unmixed using stable autoregressive filters determined implicitly by estimating a convolutive model of the mixing process. By introducing a convolutive mixing model for the components, we show how the order of the filters in the model can be correctly detected using Bayesian model selection. We demonstrate a framework for deconvolving a subspace of independent components in electroencephalography (EEG). Initial results suggest that in some cases, convolutive mixing may be a more realistic model for EEG signals than the instantaneous ICA model.     

具有非奇异约束的线性卷积混合信号盲分离联合对角化方法

联合对角化能够成功解决盲分离问题, 但在求解时会得到非期望的奇异解, 从而无法完全分离出源信号. 鉴于此, 提出一种用于线性卷积混合盲分离的联合对角化方法, 将卷积混合模型变换为瞬时模型, 并对变换后的模型应用联合对角化求取分离矩阵. 在求解过程中, 引入约束条件对解的范围进行限定, 避免了奇异解的出现. 仿真结果表明, 所提出的方法能够成功实现卷积混合信号盲分离.

     

Spatio–Temporal FastICA Algorithms for the Blind Separation of Convolutive Mixtures

This paper derives two spatio-temporal extensions of the well-known FastICA algorithm of Hyvarinen and Oja that are applicable to the convolutive blind source separation task. Our time-domain algorithms combine multichannel spatio-temporal prewhitening via multistage least-squares linear prediction with novel adaptive procedures that impose paraunitary constraints on the multichannel separation filter. The techniques converge quickly to a separation solution without any step size selection or divergence difficulties, and unlike other methods, ours do not require special coefficient initialization procedures to obtain good separation performance. They also allow for the efficient reconstruction of individual signals as observed in the sensor measurements directly from the system parameters for single-input multiple-output blind source separation tasks. An analysis of one of the adaptive constraint procedures shows its fast convergence to a paraunitary filter bank solution. Numerical evaluations of the proposed algorithms and comparisons with several existing convolutive blind source separation techniques indicate the excellent relative performance of the proposed methods.     

Mixing Audiovisual Speech Processing and Blind Source Separation for the Extraction of Speech Signals From Convolutive Mixtures

Looking at the speaker's face can be useful to better hear a speech signal in noisy environment and extract it from competing sources before identification. This suggests that the visual signals of speech (movements of visible articulators) could be used in speech enhancement or extraction systems. In this paper, we present a novel algorithm plugging audiovisual coherence of speech signals, estimated by statistical tools, on audio blind source separation (BSS) techniques. This algorithm is applied to the difficult and realistic case of convolutive mixtures. The algorithm mainly works in the frequency (transform) domain, where the convolutive mixture becomes an additive mixture for each frequency channel. Frequency by frequency separation is made by an audio BSS algorithm. The audio and visual informations are modeled by a newly proposed statistical model. This model is then used to solve the standard source permutation and scale factor ambiguities encountered for each frequency after the audio blind separation stage. The proposed method is shown to be efficient in the case of 2 times 2 convolutive mixtures and offers promising perspectives for extracting a particular speech source of interest from complex mixtures     

Multichannel blind deconvolution for source separation in convolutive mixtures of speech

This paper addresses the blind separation of convolutive and temporally correlated mixtures of speech, through the use of a multichannel blind deconvolution (MBD) method. In the proposed framework (LP-NGA), spatio-temporal separation is carried out by entropy maximization using the well-known natural gradient algorithm (NGA), while a temporal pre-whitening stage, based on linear prediction (LP), manages to fully preserve the original spectral characteristics of each source contribution. Confronted with synthetic convolutive mixtures, we show that the LP-NGA-an unconstrained natural extension to the multichannel BSS problem-benefits not only from fewer model constraints, but also from other factors, such as an overall increase in separation performance, spectral preservation efficiency and speed of convergence.     

Two-microphone separation of speech mixtures.

Separation of speech mixtures, often referred to as the cocktail party problem, has been studied for decades. In many source separation tasks, the separation method is limited by the assumption of at least as many sensors as sources. Further, many methods require that the number of signals within the recorded mixtures be known in advance. In many real-world applications, these limitations are too restrictive. We propose a novel method for underdetermined blind source separation using an instantaneous mixing model which assumes closely spaced microphones. Two source separation techniques have been combined, independent component analysis (ICA) and binary time - frequency (T-F) masking. By estimating binary masks from the outputs of an ICA algorithm, it is possible in an iterative way to extract basis speech signals from a convolutive mixture. The basis signals are afterwards improved by grouping similar signals. Using two microphones, we can separate, in principle, an arbitrary number of mixed speech signals. We show separation results for mixtures with as many as seven speech signals under instantaneous conditions. We also show that the proposed method is applicable to segregate speech signals under reverberant conditions, and we compare our proposed method to another state-of-the-art algorithm. The number of source signals is not assumed to be known in advance and it is possible to maintain the extracted signals as stereo signals.     

基于多频段排序的语音卷积混合盲源分离

针对语音卷积盲源分离频域法排列顺序不确定性问题,提出一种多频段能量排序算法。首先,通过对混合信号的短时傅立叶变换(STFT),在频域上各个频点建立一个瞬时混合模型进行独立分量分析,之后结合能量相关排序法和波达方向(DOA)排序法解决排序不确定性问题,再利用分裂语谱方法解决幅度不确定性问题,进而得到每个频点正确的分离子信号,最后利用逆短时傅立叶(ISTFT)变换得到分离的源信号。仿真结果表明,与Murata的排序算法对比,改进的算法在信号偏差比、信道干扰比、系统误差比上都所提高。     

基于频域卷积信号盲源分离的乐曲数据库构建*

将通过频域卷积信号盲源分离算法从MP3歌曲音频信号中分离出人声主唱信号,再从人声主唱信号中提取出能够表征歌曲的旋律特征构建哼唱检索系统的歌曲数据库。盲源分离要求观测信号数目不小于源信号数目,因此先用小波多分辨率分析构造一路观测信号,再用频域独立成分分析(FDICA)实现MP3歌曲音频信号的盲源分离(BSS)。实验证明,采用FDICA-based BSS从歌曲MP3中分离出的人声主唱信号的旋律特征与待检索的人声哼唱信号的旋律特征有较高的相似度,可以用歌曲MP3构建哼唱检索系统的歌曲数据库。     

基于ICA的相位编码信号欠定盲分离算法

针对二相编码信号时域或频城上不充分稀疏的情况,提出了欠定盲源分离中估计混合矩阵和恢复源信号的新方法.首先,利用二相编码信号成型模型的特异性,将欠定盲分离问题转化成卷积盲分离问题,然后通过抽头延时将其转化为线性瞬时混叠问题,通过独立分量分析(ICA)方法对延时后的观测信号进一步处理.为了准确地分离出源信号,利用峭度准则对...     

基于稳健联合分块对角化的卷积盲分离

针对卷积盲分离问题,提出一种新的矩阵联合分块对角化(Joint block diagonalization, JBD)算法. 现有的迭代非正交联合分块对角化算法都存在不收敛的情况,本文利用分离矩阵的特殊结构确保其可逆性,使得算法的迭代过程稳定. 在已知矩阵分块结构的条件下,首先,将卷积盲分离模型写成瞬时形式,并说明其满足联合分块对角化结构; 然后,提出联合分块对角化的代价函数,依据代价函数的最小化等价于矩阵中每个分块的范数最小化, 将整个分离矩阵的迭代更新转化成每个分块的迭代更新;最后,利用最小化条件得到迭代算法. 实数和复数两种情况下的算法都进行了推导.基本实验验证了新算法在不同条件下的性能; 仿真实验中对在时域和频域都重叠的信号的卷积混合进行盲分离,实验结果验证了新算法具有更好的分离性能和更稳定的分离能力.     

一种新的联合块对角化卷积盲分离时域算法

提出一种基于高阶累积量联合块对角化的时域算法求解卷积混合盲信号分离问题。引入白化处理,将混叠矩阵转变成酉矩阵,混合信号转变为互不相关的,进而计算出其对应的一系列高阶累积量矩阵,通过最小化代价函数来实现高阶累积量矩阵联合块对角化的目的,在时域中解决超定卷积盲分离问题。实验表明,相比于经典的自然梯度算法,所提方法的分离精度更高,且运算速度也更快。     

Constrained non-negative matrix factorization for score-informed piano music restoration

In this work, we propose a constrained non-negative matrix factorization method for the audio restoration of piano music using information from the score. In the first stage (instrument training), spectral patterns for the target source (piano) are learned from a dataset of isolated piano notes. The model for the piano is constrained to be harmonic because, in this way, each pattern can define a single pitch. In the second stage (noise training), spectral patterns for the undesired source (noise) are learned from the most common types of vinyl noises. To obtain a representative model for the vinyl noise, a cross-correlation-based constraint that minimizes the cross-talk between different noise components is used. In the final stage (separation), we use the trained instrument and noise models in an NMF framework to extract the clean audio signal from undesired non-stationary noise. To improve the separation results, we propose a novel score-based constraint to avoid activations of notes or combinations that are not present in the original score. The proposed approach has been evaluated and compared with commercial audio restoration softwares, obtaining competitive results.     

Nonnegative matrix factorization with the Itakura-Saito divergence: with application to music analysis

This letter presents theoretical, algorithmic, and experimental results about nonnegative matrix factorization (NMF) with the Itakura-Saito (IS) divergence. We describe how IS-NMF is underlaid by a well-defined statistical model of superimposed gaussian components and is equivalent to maximum likelihood estimation of variance parameters. This setting can accommodate regularization constraints on the factors through Bayesian priors. In particular, inverse-gamma and gamma Markov chain priors are considered in this work. Estimation can be carried out using a space-alternating generalized expectation-maximization (SAGE) algorithm; this leads to a novel type of NMF algorithm, whose convergence to a stationary point of the IS cost function is guaranteed. We also discuss the links between the IS divergence and other cost functions used in NMF, in particular, the Euclidean distance and the generalized Kullback-Leibler (KL) divergence. As such, we describe how IS-NMF can also be performed using a gradient multiplicative algorithm (a standard algorithm structure in NMF) whose convergence is observed in practice, though not proven. Finally, we report a furnished experimental comparative study of Euclidean-NMF, KL-NMF, and IS-NMF algorithms applied to the power spectrogram of a short piano sequence recorded in real conditions, with various initializations and model orders. Then we show how IS-NMF can successfully be employed for denoising and upmix (mono to stereo conversion) of an original piece of early jazz music. These experiments indicate that IS-NMF correctly captures the semantics of audio and is better suited to the representation of music signals than NMF with the usual Euclidean and KL costs.     

Regularized nonnegative matrix factorization using Gaussian mixture priors for supervised single channel source separation

We introduce a new regularized nonnegative matrix factorization (NMF) method for supervised single-channel source separation (SCSS). We propose a new multi-objective cost function which includes the conventional divergence term for the NMF together with a prior likelihood term. The first term measures the divergence between the observed data and the multiplication of basis and gains matrices. The novel second term encourages the log-normalized gain vectors of the NMF solution to increase their likelihood under a prior Gaussian mixture model (GMM) which is used to encourage the gains to follow certain patterns. In this model, the parameters to be estimated are the basis vectors, the gain vectors and the parameters of the GMM prior. We introduce two different ways to train the model parameters, sequential training and joint training. In sequential training, after finding the basis and gains matrices, the gains matrix is then used to train the prior GMM in a separate step. In joint training, within each NMF iteration the basis matrix, the gains matrix and the prior GMM parameters are updated jointly using the proposed regularized NMF. The normalization of the gains makes the prior models energy independent, which is an advantage as compared to earlier proposals. In addition, GMM is a much richer prior than the previously considered alternatives such as conjugate priors which may not represent the distribution of the gains in the best possible way. In the separation stage after observing the mixed signal, we use the proposed regularized cost function with a combined basis and the GMM priors for all sources that were learned from training data for each source. Only the gain vectors are estimated from the mixed data by minimizing the joint cost function. We introduce novel update rules that solve the optimization problem efficiently for the new regularized NMF problem. This optimization is challenging due to using energy normalization and GMM for prior modeling, which makes the problem highly nonlinear and non-convex. The experimental results show that the introduced methods improve the performance of single channel source separation for speech separation and speech–music separation with different NMF divergence functions. The experimental results also show that, using the GMM prior gives better separation results than using the conjugate prior.     

Indeterminacy Free Frequency-Domain Blind Separation of Reverberant Audio Sources

Blind separation of convolutive mixtures is a very complicated task that has applications in many fields of speech and audio processing, such as hearing aids and man-machine interfaces. One of the proposed solutions is the frequency-domain independent component analysis. The main disadvantage of this method is the presence of permutation ambiguities among consecutive frequency bins. Moreover, this problem is worst when reverberation time increases. Presented in this paper is a new frequency-domain method, that uses a simplified mixing model, where the impulse responses from one source to each microphone are expressed as scaled and delayed versions of one of these impulse responses. This assumption, based on the similitude among waveforms of the impulse responses, is valid for a small spacing of the microphones. Under this model, separation is performed without any permutation or amplitude ambiguity among consecutive frequency bins. This new method is aimed mainly to obtain separation, with a small reduction of reverberation. Nevertheless, as the reverberation is included in the model, the new method is capable of performing separation for a wide range of reverberant conditions, with very high speed. The separation quality is evaluated using a perceptually designed objective measure. Also, an automatic speech recognition system is used to test the advantages of the algorithm in a real application. Very good results are obtained for both, artificial and real mixtures. The results are significantly better than those by other standard blind source separation algorithms.     

Blind separation of non-stationary sources using continuous density hidden Markov models

Blind source separation (BSS) has attained much attention in signal processing society due to its ‘blind’ property and wide applications. However, there are still some open problems, such as underdetermined BSS, noise BSS. In this paper, we propose a Bayesian approach to improve the separation performance of instantaneous mixtures with non-stationary sources by taking into account the internal organization of the non-stationary sources. Gaussian mixture model (GMM) is used to model the distribution of source signals and the continuous density hidden Markov model (CDHMM) is derived to track the non-stationarity inside the source signals. Source signals can switch between several states such that the separation performance can be significantly improved. An expectation-maximization (EM) algorithm is derived to estimate the mixing coefficients, the CDHMM parameters and the noise covariance. The source signals are recovered via maximum a posteriori (MAP) approach. To ensure the convergence of the proposed algorithm, the proper prior densities, conjugate prior densities, are assigned to estimation coefficients for incorporating the prior information. The initialization scheme for the estimates is also discussed. Systematic simulations are used to illustrate the performance of the proposed algorithm. Simulation results show that the proposed algorithm has more robust separation performance in terms of similarity score in noise environments in comparison with the classical BSS algorithms in determined mixture case. Additionally, since the mixing matrix and the sources are estimated jointly, the proposed EM algorithm also works well in underdetermined case. Furthermore, the proposed algorithm converges quickly with proper initialization.     

基于卷积非负矩阵部分联合分解的强噪声单声道语音分离

非负矩阵部分联合分解(Nonnegative matrix partial co-factorization, NMPCF)将指定源频谱作为边信息参与混合信号频谱的联合分解, 以帮助确定指定源的基向量进而提高信号分离性能.卷积非负矩阵分解(Convolutive nonnegative matrix factorization, CNMF)采用卷积基分解的方法进行矩阵分解, 在单声道语音分离方面取得较好的效果.为了实现强噪声条件下的语音分离, 本文结合以上两种算法的优势, 提出一种基于卷积非负矩阵部分联合分解(Convolutive nonnegative partial matrix co-factorization, CNMPCF)的单声道语音分离算法.本算法首先通过基音检测算法得到混合信号的语音起始点, 再据此确定混合信号中的纯噪声段, 最后将混合信号频谱和噪声频谱进行卷积非负矩阵部分联合分解, 得到语音基矩阵, 进而得到分离的语音频谱和时域信号.实验中, 混合语音信噪比(Signal noise ratio, SNR)选择以-3 dB为间隔从0 dB至-12 dB共5种SNR.实验结果表明, 在不同噪声类型和噪声强度条件下, 本文提出的CNMPCF方法相比于以上两种方法均有不同程度的提高.