Blind Speech Separation Research Articles

Distributed Computation for Sparse Recovery via Continuous-Time Neurodynamic Approach

Sparsity has been extensively employed in multimedia sensing and computing in consumer electronics, signal and image processing, depth video codec, adaptive sparse-type equalizer, blind speech separation, and machine learning. Throughout this paper, we propose a novel distributed projection neurodynamic approach for solving the Basis Pursuit (BP) with flexible partition methods in a distributed manner. The proposed neurodynamic approach requires only that the network is undirected and connected, and no node can access the entire matrix simultaneously. First, we equivalently formulate the BP into a standard distributed optimization problem with a flexible partition-by-blocks method to obtain global information, and discuss the equivalence of their optimality conditions. Then, we propose a distributed continuous-time neurodynamic approach on the basis of primal-dual dynamical systems and projection operators, and also study its global convergence property. Finally, numerical experiments on sparse signals and image recovery further verify the effectiveness and superiority of our proposed neurodynamic approach.

NNMF with Speaker Clustering in a Uniform Filter-Bank for Blind Speech Separation

This study proposes a blind speech separation algorithm that employs a single-channel technique. The algorithm’s input signal is a segment of a mixture of speech for two speakers. At first, filter bank analysis transforms the input from time to time-frequency domain (spectrogram). Number of sub-bands for the filter is 257. Non-Negative Matrix Factorization (NNMF) factorizes each sub-band output into 28 sub-signals. A binary mask separates each sub-signal into two groups; one group belongs to the first speaker and the other to the second speaker. The binary mask separates each sub-signal of the (257×28) 7196 sub-speech signals. That separation cannot identify the speaker. Identification of the sub-signal speaker for each sub-signal is achieved by speaker clustering algorithms. Since speaker clustering cannot process without speaker segmentation, the standard windowed-overlap frames have been used to partition the speech. The speaker clustering process fetches the extracted phase angle from the spectrogram (of the mixture speech) and merges it into the spectrogram (of the recovered speech). Filter bank synthesizes these signals to produce a full-band speech signal for each speaker. Subjective tests denote that the algorithm results are accepted. Objectively, the researchers experimented with 66 mixture chats (6 females and 6 males) to test the algorithm. The average of the SIR test is 11.1 dB, SDR is 1.7 dB, and SAR is 2.8 dB.

Efficient Overdetermined Independent Vector Analysis Based on Iterative Projection with Adjustment

In this paper, a computationally efficient optimization algorithm for independent vector analysis (IVA) is proposed to accelerate iterative convergence speed and enhance the overdetermined convolutive blind speech separation performance. An iterative projection with adjustment (IPA) is investigated to estimate the unmixing matrix for OverIVA. The IPA algorithm jointly executes the iterative projection (IP) algorithm and the iterative source steering (ISS) algorithm to jointly update one row and one column of the mixing matrix, which can perform computationally-efficient blind source separation. It is achieved by updating one demixing filter and jointly adjusting all the other sources along its current direction. Motivated by its technology superiorities, this paper proposes a modified algorithm for the OverIVA, fully exploiting the computational efficiency of IPA optimization scheme. Experimental results corroborate the proposed OverIVA-IPA algorithm converges faster and performs better than the existing state-of-the-arts algorithms.

Blind separation of speech from aortic regurgitation signals using Dhoulath’s method

ABSTRACT Conducting auscultation of traumatically distressed patients has always been demanding for medical professionals. The challenge calls for an innovative solution enabling doctors to conduct precise diagnoses despite other sound interference. This suggested study presents an entirely non-invasive and convenient method designed to aid doctors in routine diagnostic procedures. This study is centred on the segregation of aortic regurgitation heart sounds from speech. The mixture utilised for the study is a combination of speech and aortic regurgitation signals. The method applied for the study is a revised procedure of Blind Source component separation utilising a solo sensor method. With this technique, doctors are not compelled to prevent patients from articulating their pain or discomfort while diagnosing heart sounds. Doctors can offer a consoling word to patients while the auscultation is in progress without worrying about how the speech sounds affect the diagnosis. For babies, timely detection of heart-related issues can be life-saving. With Dhoulath’s method, the distressing sounds of a baby’s cries can be effectively separated, thereby offering doctors clear audio of heartbeats. The study was conducted to ascertain if heartbeats can be segregated from the signals of speech or cries. This segregation procedure has succeeded in arriving at an enhanced level of clarity.

Blind Speech Separation and Dereverberation using neural beamforming

In this paper, we present the Blind Speech Separation and Dereverberation (BSSD) network, which performs simultaneous speaker separation, dereverberation and speaker identification in a single neural network. Speaker separation is guided by a set of predefined spatial cues. Dereverberation is performed by using neural beamforming, and speaker identification is aided by embedding vectors and triplet mining. We introduce a frequency-domain model which uses complex-valued neural networks, and a time-domain variant which performs beamforming in latent space. Further, we propose a block-online mode to process longer audio recordings, as they occur in meeting scenarios. We evaluate our system in terms of Scale Independent Signal to Distortion Ratio (SI-SDR), Word Error Rate (WER) and Equal Error Rate (EER).

Online blind source separation and dereverberation of speech based on a joint diagonalizability constraint

Online blind source separation and dereverberation of speech based on a joint diagonalizability constraint

Independent Vector Analysis via Log-Quadratically Penalized Quadratic Minimization

We propose a new algorithm for blind source separation (BSS) using independent vector analysis (IVA). This is an improvement over the popular auxiliary function based IVA (AuxIVA) with iterative projection (IP) or iterative source steering (ISS). We introduce iterative projection with adjustment (IPA), where we update one demixing filter and jointly adjust all the other sources along its current direction. Each update involves solving a non-convex minimization problem that we term log-quadratically penalized quadratic minimization (LQPQM), that we think is of interest beyond this work. In the general case, we show that its global minimum corresponds to the largest root of a univariate function, reminiscent of modified eigenvalue problems. We propose a simple procedure based on Newton-Raphson to efficiently compute it. Numerical experiments demonstrate the effectiveness of the proposed method. First, we show that it efficiently decreases the value of the surrogate function. In further experiments on synthetic mixtures, we study the probability of finding the true demixing matrix and convergence speed. We show that the proposed method combines high success rate and fast convergence. Finally, we validate the performance on a reverberant blind speech separation task. We find that all the AuxIVA-based methods perform similarly in terms of acoustic BSS metrics. However, AuxIVA-IPA converges faster. We measure up to 8.5 times speed-up in terms of runtime compared to the next best AuxIVA-based method, depending on the number of channels and the signal-to-noise ratio (SNR).

Time and frequency based sparse bounded component analysis algorithms for convolutive mixtures

In this paper, we introduce time-domain and frequency-domain versions of a new Blind Source Separation (BSS) approach to extract bounded magnitude sparse sources from convolutive mixtures. We derive algorithms by maximization of the proposed objective functions that are defined in a completely deterministic framework, and prove that global maximums of the objective functions yield perfect separation under suitable conditions. The derived algorithms can be applied to temporal or spatially dependent sources as well as independent sources. We provide experimental results to demonstrate some benefits of the approach, also including an application on blind speech separation.

Kurtosis maximisation for blind speech separation in Hindi speech processing system using swarm intelligence and ICA

Kurtosis maximisation for blind speech separation in Hindi speech processing system using swarm intelligence and ICA

A Source Counting Method Using Acoustic Vector Sensor Based on Sparse Modeling of DOA Histogram

The number of sources present in a mixture is crucial information often assumed to be known or detected by source counting. The existing methods for source counting in underdetermined blind speech separation suffer from the overlapping between sources with low W-disjoint orthogonality. To address this issue, we propose to fit the direction-of-arrival (DOA) histogram with multiple von-Mises density (VM) functions directly and form a sparse recovery problem, where all the source clusters and the sidelobes in the DOA histogram are fitted with VM functions of different spatial parameters. We also developed a formula to perform the source counting taking advantage of the values of the sparse source vector to reduce the influence of sidelobes. Experiments are carried out to evaluate the proposed source counting method, and the results show that the proposed method outperforms two well-known baseline methods.

Separating Multi Speeches in Intelligent Humanoid Robot using FastICA

The main objective of our research is to develop an intelligent humanoid robot for teaching children by listening and answering the questions. In our previous research, we have designed a humanoid robot that can detect human face and receive commands by using speech recognition. Our robot is based on Bioloid GP robot and Raspberry Pi2 as control system. In this study, we would like to expand the capability of the robot system in order to isolate the speech of one speaker from all the other sounds. The problem for separating multi speeches from stereo audio record is called as Blind Speech Separation (BSS). We propose FastICA algorithm to solve the BSS problem. FastICA is an efficient algorithm to separate several signals based on Independent Component Analysis (ICA) algorithm. Some assumption must be met to use FastICA, that is the number of mixtures are equal to the number of sources and the sources are linearly independent from each other. To evaluate the algorithm, we use several simulations based on two speech sources and its mixing matrix. Our simulation shows FastICA algorithm can solve BSS problem by separating two sound signals, but its linearly independent assumption makes it difficult to implement in our humanoid robot

An Unsupervised Two-Talker Speech Separation System Based on CASA

On the basis of the theory about blind separation of monaural speech based on computational auditory scene analysis (CASA), a two-talker speech separation system combining CASA and speaker recognition was proposed to separate speech from other speech interferences in this paper. First, a tandem algorithm is used to organize voiced speech, then based on the clustering of gammatone frequency cepstral coefficients (GFCCs), an object function is established to recognize the speaker, and the best group is achieved through exhaustive search or beam search, so that voiced speech is organized sequentially. Second, unvoiced segments are generated by estimating onset/offset, and then unvoiced–voiced (U–V) segments and unvoiced–unvoiced (U–U) segments are separated respectively. The U–V segments are managed via the binary mask of the separated voiced speech, while the U–V segments are separated evenly. So far the unvoiced segments are separated. The simulation and performance evaluation verify the feasibility and effectiveness of the proposed algorithm.

Implementation of Blind Speech Separation for Intelligent Humanoid Robot using DUET Method

Nowadays, there are many efforts in building intelligent humanoid robot and adding advanced ability such as Blind Speech Separation (BSS). BSS is a problem of separation of several speech signals in a real world from mono or stereo audio record. In this research, we implement BSP system using DUET algorithm which allow to separate any number of sources by using only stereo (two) mixtures. The DUET (Degenerate Unmixing Estimation Technique) algorithm replaces our previous FastICA (Fast Independent Component Analysis) method only success in simulation but failed in the implementation. The main problem of FastICA is that it assumes instantaneous mixing without time delay in the recording process. To deals with audio record in the presence of inevitable time delays, it has to be replaced with DUET algorithm to separate well in real time. Finally, the DUET algorithm is implemented to humanoid robot which is developed using Raspberry Pi and equipped with RaspPi Cam to detect human face. Furthermore, the Cirrus Logic Audio Card is stacked to Raspberry Pi in order to record stereo audio. In our experiments, there are three controlled variables to evaluate algorithm performance, that is: distance, number of sources, and subject’s name. Robot will record stereo audio for four seconds after face is detected by system. The recording is then separated by DUET algorithm and produce two source estimations with average computation time 1.8 seconds. With Google API, the recognition accuracy of separated speech is varying between 40%-70%.

Kurtosis Maximization for Blind Speech Separation in Hindi Speech Processing System using Swarm Intelligence and ICA

Blind source separation (BSS) method divides mixed signals blindly without any data on the mixing scheme. This is a main issue in an actual period world whether have to identify a specific person in the crowd or it is a zone of speech signal is removed. Besides, these BSS approaches are collective with shape and also statistical features to authenticate the performance of each one in outline classification. For resolving this issue proposes an active BSS algorithm on the basis of the group search optimisation (GSO). The kurtosis of the signals is used as the objective performance and the GSO is utilised to resolve it in the suggested algorithm. Primarily, source signals are taken into account as the independent component analysis (ICA) to generate the mixing signals to BSS yield the maximum kurtosis. The source signal constituent that is divided out is then smeared off from mixtures with the help of the deflation technique. Each and every source signals establish that important development of the computation amount and the quality of signal separation is attained using the projected BSS-GSO algorithm if associated with the preceding algorithms.

Source separation employing beamforming and SRP-PHAT localization in three-speaker room environments

This paper presents a new blind speech separation algorithm using beamforming technique that is capable of extracting each individual speech signal from a mixture of three speech sources in a room. The speech separation algorithm utilizes the steered response power phase transform for obtaining a localization estimate for each individual speech source in the frequency domain. Based on those estimates each desired speech signal is extracted from the speech mixture using an optimal beamforming technique. To solve the permutation problem, a permutation alignment algorithm based on the mutual output correlation is employed to group the output signals into the correct sources from each frequency bin. Evaluations using real speech recordings in a room environment show that the proposed blind speech separation algorithm offers high interference suppression level whilst maintaining low distortion level for each desired signal.

Robustness Analysis of Structured Matrix Factorization via Self-Dictionary Mixed-Norm Optimization

We are interested in a low-rank matrix factorization problem where one of the matrix factors has a special structure; specifically, its columns live in the unit simplex. This problem finds applications in diverse areas such as hyperspectral unmixing, video summarization, spectrum sensing, and blind speech separation. Prior works showed that such a factorization problem can be formulated as a self-dictionary sparse optimization problem under some assumptions that are considered realistic in many applications, and convex mixed norms were employed as optimization surrogates to realize the factorization in practice. Numerical results have shown that the mixed-norm approach demonstrates promising performance. In this letter, we conduct performance analysis of the mixed-norm approach under noise perturbations. Our result shows that using a convex mixed norm can indeed yield provably good solutions. More importantly, we also show that using nonconvex mixed (quasi) norms is more advantageous in terms of robustness against noise.

Joint Tensor Factorization and Outlying Slab Suppression With Applications

We consider factoring low-rank tensors in the presence of outlying slabs. This problem is important in practice, because data collected in many real-world applications, such as speech, fluorescence, and some social network data, fit this paradigm. Prior work tackles this problem by iteratively selecting a fixed number of slabs and fitting, a procedure which may not converge. We formulate this problem from a group-sparsity promoting point of view, and propose an alternating optimization framework to handle the corresponding $\ell_p$ ($0<p\leq 1$) minimization-based low-rank tensor factorization problem. The proposed algorithm features a similar per-iteration complexity as the plain trilinear alternating least squares (TALS) algorithm. Convergence of the proposed algorithm is also easy to analyze under the framework of alternating optimization and its variants. In addition, regularization and constraints can be easily incorporated to make use of \emph{a priori} information on the latent loading factors. Simulations and real data experiments on blind speech separation, fluorescence data analysis, and social network mining are used to showcase the effectiveness of the proposed algorithm.

A Post-filtering Algorithm for Crosstalk Subtraction in Blind Speech Separation Outputs

In this paper, exclusive activity periods (EAPs) are used to model crosstalks in which only one blind source separation (BSS) output signal is assumed to be active and the others are inactive. Inactive intervals of each EAP are used to estimate the crosstalk leaked from each active signal using a least squares method. Then, we use the Wiener filter to cancel the estimated crosstalks from BSS outputs. The benefit of using EAPs is to simplify the estimation of crosstalks from all other signals to the estimation of them from just one active signal in each interval. Thus, it leads to estimate and suppress crosstalks more precisely. A comparison of our method with other popular post-processing algorithms is drawn. The results show an improved performance of the proposed method over earlier approaches.

Modified subspace method based on convex model for underdetermined blind speech separation

Recently, many researches have been done to solve the speech separation problems in the underdetermined cases, and the Two-step method is widely used, which estimates the mixing matrix first and then separates the sources. Subspace projection is an effective method for signal separation, whereas it has some limitations because it fixes the active number at the Time-Frequency (TF) point as a constant. A commonly used solution is to add an extra step to estimate the active source number at every TF point, which will increase the computation cost greatly. This paper provides a new convex-model-based subspace projection method with enhanced functionality, which can be used for Underdetermined Blind Source Separation (UBSS). The model takes into account both projection and size of the signal's subspace, without estimating the real source numbers at the TF points. Simulation results show that the proposed method overcomes the shortage of conventional subspace method and achieves higher separation performance. Furthermore, the proposed algorithm can be employed as a preprocessing technology in the fields of acoustic signals enhancement, recognition, and biomedical images, etc.

Multi-band multi-centroid clustering based permutation alignment for frequency-domain blind speech separation

This paper investigates the permutation ambiguity problem in frequency-domain blind source separation and proposes a robust permutation alignment algorithm based on inter-frequency dependency, which is measured by the correlation coefficient between the time activity sequences of separated signals. To calculate a global reference for permutation alignment, a multi-band multi-centroid clustering algorithm is proposed where at first the permutation inside each subband is aligned with multi-centroid clustering and then the permutation among subbands is aligned sequentially. The multi-band scheme can reduce the dynamic range of the activity sequence and improve the efficiency of clustering, while the multi-centroid clustering scheme can improve the precision of the reference and reduce the risk of wrong permutation among subbands. The combination of two techniques enables to capture the variation of the time–frequency activity of a speech signal precisely, promising robust permutation alignment performance. Extensive experiments are carried out in different testing scenarios (up to reverberation time of 700 ms and 4×4 mixtures) to investigate the influence of two parameters, the number of subbands and the number of clustering-centroids, on the performance of the proposed algorithm. Comparison with existing permutation alignment algorithms proves that the proposed algorithm can improve the robustness in challenging scenarios and can reduce block permutation errors effectively.