Multichannel Wiener Filter Research Articles

RELAX-Jr: An Automated Pre-Processing Pipeline for Developmental EEG Recordings.

Automated EEG pre-processing pipelines provide several key advantages over traditional manual data cleaning approaches; primarily, they are less time-intensive and remove potential experimenter error/bias. Automated pipelines also require fewer technical expertise as they remove the need for manual artefact identification. We recently developed the fully automated Reduction of Electroencephalographic Artefacts (RELAX) pipeline and demonstrated its performance in cleaning EEG data recorded from adult populations. Here, we introduce the RELAX-Jr pipeline, which was adapted from RELAX and designed specifically for pre-processing of data collected from children. RELAX-Jr implements multi-channel Wiener filtering (MWF) and/or wavelet-enhanced independent component analysis (wICA) combined with the adjusted-ADJUST automated independent component classification algorithm to identify and reduce all artefacts using algorithms adapted to optimally identify artefacts in EEG recordings taken from children. Using a dataset of resting-state EEG recordings (N = 136) from children spanning early-to-middle childhood (4-12 years), we assessed the cleaning performance of RELAX-Jr using a range of metrics including signal-to-error ratio, artefact-to-residue ratio, ability to reduce blink and muscle contamination, and differences in estimates of alpha power between eyes-open and eyes-closed recordings. We also compared the performance of RELAX-Jr against four publicly available automated cleaning pipelines. We demonstrate that RELAX-Jr provides strong cleaning performance across a range of metrics, supporting its use as an effective and fully automated cleaning pipeline for neurodevelopmental EEG data.

Characterization of Cochlear Implant Artifact and Removal Based on Multi-Channel Wiener Filter in Unilateral Child Patients.

Cochlear implants (CI) allow deaf patients to improve language perception and improving their emotional valence assessment. Electroencephalographic (EEG) measures were employed so far to improve CI programming reliability and to evaluate listening effort in auditory tasks, which are particularly useful in conditions when subjective evaluations are scarcely appliable or reliable. Unfortunately, the presence of CI on the scalp introduces an electrical artifact coupled to EEG signals that masks physiological features recorded by electrodes close to the site of implant. Currently, methods for CI artifact removal have been developed for very specific EEG montages or protocols, while others require many scalp electrodes. In this study, we propose a method based on the Multi-channel Wiener filter (MWF) to overcome those shortcomings. Nine children with unilateral CI and nine age-matched normal hearing children (control) participated in the study. EEG data were acquired on a relatively low number of electrodes (n = 16) during resting condition and during an auditory task. The obtained results obtained allowed to characterize CI artifact on the affected electrode and to significantly reduce, if not remove it through MWF filtering. Moreover, the results indicate, by comparing the two sample populations, that the EEG data loss is minimal in CI users after filtering, and that data maintain EEG physiological characteristics.

Multichannel Wiener filter in active sound‐navigation‐and‐ranging systems—A joint beamformer and matched filter approach

AbstractConventional active SONAR systems often use beamformers and matched filters separately to extract bearing and range information from the received signal and offer a straightforward way of creating a two‐dimensional map of the environment. In SONAR systems the minimum‐variance‐distortionless‐response beamformer (MVDR beamformer) is a commonly used type of beamformer, which will reconstruct the receive signal from a certain direction optimally. In terms of detecting the transmit signal, the most used method is the conventional matched filter. Both algorithms are simple to implement and perform well under various noise scenarios. The proposed method combines the beamformer and matched filter by introducing an extended channel model that allows the derivation of a multichannel Wiener filter to solve for the unknown reflection coefficients of the complete two‐dimensional environment. This results in adaptively calculated filter weights that will drastically improve the performance compared to a separate MVDR beamformer and matched filter. In addition, a parameter is introduced with which one can arbitrarily adjust the focus between angular and temporal resolution depending on the application. After the derivation, the performance is demonstrated with simulations and measurements.

Frame-wise speech extraction with recursive expectation maximization for partially deformable microphone arrays

Large-apertured wearable and deformable microphone arrays have shown their better performance than conventional small wearable and rigid microphone arrays for speech extraction tasks, especially when the desired speech is far away from the arrays or there exist a great number of directional noises in practical applications. Most existing beamforming methods that require the information of the desired sound direction often cannot be used directly on deformable arrays because their shape might continuously change over time. To tackle the problem of utilizing partially deformable microphone arrays for target speech extraction, this paper derives a frame-wise speech extraction method, which is referred to as the recursive multichannel Wiener filter with generalized Gamma priors. In this method, the complex spectrum of the desired speech is assumed to follow a zero-mean complex Gaussian distribution with time-varying variance, and the reciprocal of this time-varying variance known as the precision is further assumed to follow a Gamma distribution, while the complex spectrum of the noise is assumed to follow a zero-mean complex Gaussian distribution. Under these two assumptions, we derive the recursive expectation maximization algorithm to extract the desired speech in a frame-by-frame manner. Experimental results show the promising performance of the proposed method in terms of multiple objective metrics.

Frequency-domain distributed multichannel wiener filtering speech enhancement algorithm

A frequency-domain distributed microphone multi-channel Wiener filter speech enhancement algorithm is proposed in this paper. In this paper, the distributed microphone speech model is considered. First, the speech signal in the time domain is converted into the speech signal in the frequency domain by the discrete Fourier transform method. Then, the unconstrained minimization problem of the noise reduction and speech distortion of the complex linear filter in the frequency domain is established. Simulation results show that the proposed algorithm is superior to some existing multi-channel speech enhancement algorithms.

Drone audition: Audio signal enhancement from drone embedded microphones using multichannel Wiener filtering and Gaussian-mixture based post-filtering

In this paper, we consider the problem of recovering desired sound source signals from on-board microphone recordings on a noisy drone. Enhancement of source signal degraded by drone noise is considered to be a difficult task due to the strong noise generated from its motors and propellers causing an extremely low signal-to-drone noise ratio (SD‾NR). We propose a solution (i) by combining the widely known multichannel Wiener filter (MWF) to remove drone noise from microphone recordings, and (ii) further reduction of residual noise using a Gaussian mixture model (GMM) based dual-stage parametric Wiener filter (WF). The method exploits known statistics of motor current-specific drone noise. This combination of techniques to the specific context of signal enhancement for drone audition is applicable to irregular microphone arrays embedded on a drone enabling realistic integration to most drones. We demonstrate the validity of the proposed framework with extensive real data through (i) experimental recordings from two different drone acoustics datasets and (ii) outdoor measurements from a hovering drone for a bioacoustic application. The results confirm improved performance in terms of SD‾NR, speech quality (PESQ), and intelligibility (STOI) at very low SD‾NR (up to −30 dB) and show a strong potential for signal enhancement applications using noisy drones.

A nested generalized sidelobe canceller for source counting, localization, and signal separation in reverberant fields

This paper proposes a nested generalized sidelobe canceller (NGSC) for typical array signal processing tasks, including source counting, localization, and signal separation. Multiple blocking matrices are arranged in a nested structure to successively eliminate dominant sources until the remaining signal is predominantly incoherent noise. The microphone signals are dereverberated using a multichannel weighted prediction error algorithm. In source counting, the number of sound sources is determined by tracking the average power of the blocked signal. In source localization, the direction-of-arrival (DOA) is estimated via cosine similarity in conjunction with the golden section search. In signal separation, the estimated DOA enables speech separation in a linearly constrained minimum variance beamformer with postfiltering (LCMV-PF). Monte Carlo simulations are performed to compare the proposed NGSC approach with four baselines, minimum description length, second order statistic of the eigenvalue, multistage Wiener filtering, and multiple signal classification. The results show that NGSC achieves at least 28.80% higher source counting accuracy with a 1.04° lower root mean square degree error than the baselines. The signal-to-distortion ratio achieved by LCMV-PF is 1.52 dB higher than that achieved by the linearly constrained minimum power beamformer and the multichannel Wiener filter.

Cascade algorithms for combined acoustic feedback cancelation and noise reduction

This paper presents three cascade algorithms for combined acoustic feedback cancelation (AFC) and noise reduction (NR) in speech applications. A prediction error method (PEM)-based adaptive feedback cancelation (PEM-based AFC) algorithm is used for the AFC stage, while a multichannel Wiener filter (MWF) is applied for the NR stage. A scenario with M microphones and 1 loudspeaker is considered, without loss of generality. The first algorithm is the baseline algorithm, namely the cascade M-channel rank-1 MWF and PEM-AFC, where a NR stage is performed first using a rank-1 MWF followed by a single-channel AFC stage using a PEM-based AFC algorithm. The second algorithm is the cascade (M+1)-channel rank-2 MWF and PEM-AFC, where again a NR stage is applied first followed by a single-channel AFC stage. The novelty of this algorithm is to consider an (M+1)-channel data model in the MWF formulation with two different desired signals, i.e., the speech component in the reference microphone signal and in the loudspeaker signal, both defined by the speech source signal but not equal to each other. The two desired signal estimates are later used in a single-channel PEM-based AFC stage. The third algorithm is the cascade M-channel PEM-AFC and rank-1 MWF where an M-channel AFC stage is performed first followed by an M-channel NR stage. Although in cascade algorithms where NR is performed first and then AFC the estimation of the feedback path is usually affected by the NR stage, it is shown here that by performing a rank-2 approximation of the speech correlation matrix this issue can be avoided and the feedback path can be correctly estimated. The performance of the algorithms is assessed by means of closed-loop simulations where it is shown that for the considered input signal-to-noise ratios (iSNRs) the cascade (M+1)-channel rank-2 MWF and PEM-AFC and the cascade M-channel PEM-AFC and rank-1 MWF algorithms outperform the cascade M-channel rank-1 MWF and PEM-AFC algorithm in terms of the added stable gain (ASG) and misadjustment (Mis) as well as in terms of perceptual metrics such as the short-time objective intelligibility (STOI), perceptual evaluation of speech quality (PESQ), and signal distortion (SD).

RELAX part 2: A fully automated EEG data cleaning algorithm that is applicable to Event-Related-Potentials

ObjectiveElectroencephalography (EEG) is often used to examine neural activity time-locked to stimuli presentation, referred to as Event-Related Potentials (ERP). However, EEG is influenced by non-neural artifacts, which can confound ERP comparisons. Artifact cleaning reduces artifacts, but often requires time-consuming manual decisions. Most automated methods filter frequencies <1 Hz out of the data, so are not recommended for ERPs (which contain frequencies <1 Hz). Our aim was to test the RELAX (Reduction of Electroencephalographic Artifacts) pre-processing pipeline for use on ERP data. MethodsThe cleaning performance of multiple versions of RELAX were compared to four commonly used EEG cleaning pipelines across both artifact cleaning metrics and the amount of variance in ERPs explained by different conditions in a Go-Nogo task.ResultsRELAX with Multi-channel Wiener Filtering (MWF) and wavelet-enhanced independent component analysis applied to artifacts identified with ICLabel (wICA_ICLabel) cleaned data most effectively and produced amongst the most dependable ERP estimates. RELAX with wICA_ICLabel only or MWF_only may detect effects better for some ERPs. ConclusionsRELAX shows high artifact cleaning performance even when data is high-pass filtered at 0.25 Hz (applicable to ERP analyses). SignificanceRELAX is easy to implement via EEGLAB in MATLAB and freely available on GitHub. Given its performance and objectivity we recommend RELAX to improve artifact cleaning and consistency across ERP research.

Introducing RELAX: An automated pre-processing pipeline for cleaning EEG data - Part 1: Algorithm and application to oscillations

ObjectiveElectroencephalographic (EEG) data are often contaminated with non-neural artifacts which can confound experimental results. Current artifact cleaning approaches often require costly manual input. Our aim was to provide a fully automated EEG cleaning pipeline that addresses all artifact types and improves measurement of EEG outcomes MethodsWe developed RELAX (the Reduction of Electroencephalographic Artifacts). RELAX cleans continuous data using Multi-channel Wiener filtering [MWF] and/or wavelet enhanced independent component analysis [wICA] applied to artifacts identified by ICLabel [wICA_ICLabel]). Several versions of RELAX were compared using three datasets (N = 213, 60 and 23 respectively) against six commonly used pipelines across a range of artifact cleaning metrics, including measures of remaining blink and muscle activity, and the variance explained by experimental manipulations after cleaning. ResultsRELAX with MWF and wICA_ICLabel showed amongst the best performance at cleaning blink and muscle artifacts while preserving neural signal. RELAX with wICA_ICLabel only may perform better at differentiating alpha oscillations between working memory conditions. ConclusionsRELAX provides automated, objective and high-performing EEG cleaning, is easy to use, and freely available on GitHub. SignificanceWe recommend RELAX for data cleaning across EEG studies to reduce artifact confounds, improve outcome measurement and improve inter-study consistency.

A Unified Algorithmic Framework for Distributed Adaptive Signal and Feature Fusion Problems—Part I: Algorithm Derivation

In this paper, we describe a general algorithmic framework for solving linear signal or feature fusion optimization problems in a distributed setting, for example in a wireless sensor network (WSN). These problems require linearly combining the observed signals (or features thereof) collected at the various sensor nodes to satisfy a pre-defined optimization criterion. The framework covers several classical spatial filtering problems, including minimum variance beamformers, multi-channel Wiener filters, principal component analysis, canonical correlation analysis, (generalized) eigenvalue problems, etc. The proposed distributed adaptive signal fusion (DASF) algorithm is an iterative method that solves these types of problems by allowing each node to share a linearly compressed version of the local sensor signal observations with its neighbors to reduce the energy and bandwidth requirements of the network. We first discuss the case of fully-connected networks and then extend the analysis to more general network topologies. The general DASF algorithm is shown to have several existing distributed algorithms from the literature as a special case, while at the same time allowing to solve new distributed problems as well with guaranteed convergence and optimality. This paper focuses on the algorithm derivation of the DASF framework along with simulations demonstrating its performance. A technical analysis along with convergence conditions and proofs are provided in a companion paper.

EEG-Based Mental Tasks Recognition via a Deep Learning-Driven Anomaly Detector.

This paper introduces an unsupervised deep learning-driven scheme for mental tasks' recognition using EEG signals. To this end, the Multichannel Wiener filter was first applied to EEG signals as an artifact removal algorithm to achieve robust recognition. Then, a quadratic time-frequency distribution (QTFD) was applied to extract effective time-frequency signal representation of the EEG signals and catch the EEG signals' spectral variations over time to improve the recognition of mental tasks. The QTFD time-frequency features are employed as input for the proposed deep belief network (DBN)-driven Isolation Forest (iF) scheme to classify the EEG signals. Indeed, a single DBN-based iF detector is constructed based on each class's training data, with the class's samples as inliers and all other samples as anomalies (i.e., one-vs.-rest). The DBN is considered to learn pertinent information without assumptions on the data distribution, and the iF scheme is used for data discrimination. This approach is assessed using experimental data comprising five mental tasks from a publicly available database from the Graz University of Technology. Compared to the DBN-based Elliptical Envelope, Local Outlier Factor, and state-of-the-art EEG-based classification methods, the proposed DBN-based iF detector offers superior discrimination performance of mental tasks.

Robust Distributed Noise Suppression in Acoustic Sensor Networks

In acoustic sensor networks (ASNs), speech-related applications often perform poorly due to serious noise interference. To address this problem, a robust distributed noise suppression method for ASNs is proposed in this article. Specifically, by using the maximum input signal-to-noise ratio (SNR) subarray as the reference array, the robust distributed parameterized multichannel Wiener filter (RD-PMWF) is proposed for a tradeoff between the desired signal distortion and the noise suppression in the fully connected ASNs, where each subarray fuses its multiple speech signals into a single-channel signal by a compression vector. Then, to seek out the maximum input SNR subarray in any topology, an in-network compare method that takes place on the tree topology is presented based on a data-driven signal flow. Finally, the RD-PMWF without topology constraints is obtained, relying on the in-network comparison of input SNR, and the communication cost and computational complexity are analyzed. The proposed method is robust to the change of network topology and speaker positions even at low input SNR situations and successfully suppresses noise by using only the subarray with the maximum input SNR and its neighbors in any topology. Simulation and real-world experimental results illustrate that the proposed method achieves good noise suppression performance in noisy and reverberant ASNs while exhibiting an even lower computational complexity.

Distributed multichannel Wiener filtering for speech enhancement in acoustic sensor networks

SummaryThe performance of speech processing algorithms in wireless acoustic sensor networks (WASNs) is severely degraded by diffuse noise. To address this problem, a distributed speech enhancement algorithm for WASNs is proposed in this article. The parameterized multichannel Wiener filter is first improved by exploiting all cross‐correlation information among nodes. Next, using the Hadamard element‐wise product of label matrix and cross‐correlation matrix, the distributed parameterized multichannel Wiener filter based on the in‐network summation is presented. Moreover, by making the tradeoff between communication cost and enhancement performance, a simplified distributed algorithm is proposed. Finally, the noise reduction performance of the proposed method is theoretically analyzed. The proposed method achieves good speech enhancement performance and requires only in‐network summation in WASNs. Experimental results reveal the validity of the proposed method.

Automated CCA-MWF Algorithm for Unsupervised Identification and Removal of EOG Artifacts From EEG.

Affective brain computer interface (ABCI) enables machines to perceive, understand, express and respond to people's emotions. Therefore, it is expected to play an important role in emotional care and mental disorder detection. EEG signals are most frequently adopted as the physiology measurement in ABCI applications. Eye blinking and movements introduce lots of artifacts into raw EEG data, which seriously affect the quality of EEG signal and the subsequent emotional EEG feature engineering and recognition. In this paper, we propose a fully automatic and unsupervised ocular artifact identification and removal algorithm named automated canonical correlation analysis (CCA)-multi-channel wiener filter (MWF) (ACCAMWF). Firstly, spatial distribution entropy (SDE) and spectral entropy (SE) are computed to automatically annotate artifact segments. Then, CCA algorithm is used to extract neural signal from artifact contaminated data to further supplement the clean EEG data. Finally, MWF is trained to remove ocular artifacts from multiple channel EEG data adaptively. Extensive experiments have been carried out on semi-simulated EEG/EOG dataset and real eye blinking-contaminated EEG dataset to verify the effectiveness of our method when compared to two state-of-the-art algorithms. The results clearly demonstrate that ACCAMWF is a promising solution for removing EOG artifacts from emotional EEG data.

Objective evaluation of stimulation artefact removal techniques in the context of neural spike sorting

Objective. We present a framework to objectively test and compare stimulation artefact removal techniques in the context of neural spike sorting. Approach. To this end, we used realistic hybrid ground-truth spiking data, with superimposed artefacts from in vivo recordings. We used the framework to evaluate and compare several techniques: blanking, template subtraction by averaging, linear regression, and a multi-channel Wiener filter (MWF). Main results. Our study demonstrates that blanking and template subtraction result in a poorer spike sorting performance than linear regression and MWF, while the latter two perform similarly. Finally, to validate the conclusions found from the hybrid evaluation framework, we also performed a qualitative analysis on in vivo recordings without artificial manipulations. Significance. Our framework allows direct quantification of the impact of the residual artefact on the spike sorting accuracy, thereby allowing for a more objective and more relevant comparison compared to indirect signal quality metrics that are estimated from the signal statistics. Furthermore, the availability of a ground truth in the form of single-unit spiking activity also facilitates a better estimation of such signal quality metrics.

Robust parameter strategy for Wiener-based binaural noise reduction methods in hearing aids

This work presents a method for designing the weighting parameter required for Wiener-based binaural noise reduction methods. This parameter establishes the desired trade-off between noise reduction and binaural cue preservation in hearing aids. The proposed strategy was specially derived for the preservation of the interaural level difference, interaural time difference and interaural coherence binaural cues. It is defined as a function of the average input noise power at the microphones, providing robustness against the influence of joint changes in noise and speech power (Lombard effect), as well as signal-to-noise ratio variations. A theoretical framework based on the mathematical definition of the degree of homogeneity is presented and applied to a generic augmented Wiener-based cost function. The theoretical insights obtained are supported by computational simulations and psychoacoustic experiments using the multichannel Wiener filter with a noise interaural transfer function preservation technique (MWF-ITFv), as a case study. Statistical analysis indicates that, compared to a purely fixed weighting parameter, the proposed structure for the weighting parameter and the design method of its fixed part provide significant robustness against changes in the noise binaural cues at the cost of a small noise reduction performance decrease and speech binaural cue distortion.

Multi-sensory sound source enhancement for unmanned aerial vehicle recordings

A method to effectively perform sound source enhancement from an unmanned aerial vehicle (UAV)-mounted audio recording system is proposed. The objective of this study is to utilise audio recordings and non-acoustical UAV rotor characteristics to improve rotor noise power spectral density (PSD) estimation accuracy and robustness. The improved rotor noise PSD estimate in turn improves the effectiveness of the rotor noise postfilter, leading to improvement in source enhancement performance. The performance of the proposed source enhancement algorithm is evaluated via experiments, with recordings made in an outdoor environment with an in-flight UAV. Experiment results show PSD estimation accuracy to within 1.5 dB log spectral distortion regardless of the given input conditions, such as the presence of surrounding sound sources. The method also achieves a consistent ∼20–25 dB improvement in source enhancement performance, where the effects of rotor noise from the noisy microphone recordings are significantly reduced. The proposed method also outperforms existing state-of-the-art such as the beamformer with postfilter and speech distortion weighted multichannel Wiener filter frameworks.

Distributed Combined Acoustic Echo Cancellation and Noise Reduction in Wireless Acoustic Sensor and Actuator Networks

The paper presents distributed algorithms for combined acoustic echo cancellation (AEC) and noise reduction (NR) in a wireless acoustic sensor and actuator network (WASAN) where each node may have multiple microphones and multiple loudspeakers, and where the desired signal is a speech signal. A centralized integrated AEC and NR algorithm, i.e., multichannel Wiener filter (MWF), is used as starting point where echo signals are viewed as background noise signals and loudspeaker signals are used as additional input signals to the algorithm. By including prior knowledge (PK), namely that the loudspeaker signals do not contain any desired signal component, an alternative centralized cascade algorithm (PK-MWF) is obtained with an AEC stage first followed by an MWF-based NR stage which has a lower computational complexity. Distributed algorithms can then be obtained from the MWF and PK-MWF algorithm, i.e., the generalized eigenvalue decomposition (GEVD)-based distributed adaptive node-specific signal estimation (DANSE) and PK-GEVD-DANSE algorithm, respectively. In the former, each node performs a reduced dimensional integrated AEC and NR algorithm and broadcasts only 1 fused signal (instead of all its signals) to the other nodes. In the PK-GEVD-DANSE algorithm, each node performs a reduced dimensional cascade AEC and NR algorithm and broadcasts only 2 fused signals (instead of all its signals) to the other nodes. The distributed algorithms achieve the same performance, upon convergence, as the corresponding centralized integrated (MWF) and centralized cascade (PK-MWF) algorithm. It is observed, however, that the communication cost in the PK-GEVD-DANSE algorithm can also be reduced, where each node then broadcasts only 1 fused signal (instead of 2 signals) to the other nodes. The resulting algorithm, referred to as the pruned PK-GEVD-DANSE (pPK-GEVD-DANSE) algorithm, then effectively combines the lowest possible communication cost (as low as in the GEVD-DANSE algorithm) with a lowest possible computational complexity in each node (further reduced from the PK-GEVD-DANSE computational complexity), within the class of algorithms considered in this paper.

Convolutive Transfer Function-Based Multichannel Nonnegative Matrix Factorization for Overdetermined Blind Source Separation

Most multichannel blind source separation (BSS) approaches rely on a spatial model to encode the transfer functions from sources to microphones and a source model to encode the source power spectral density. The rank-1 spatial model has been widely exploited in independent component analysis (ICA), independent vector analysis (IVA), and independent low-rank matrix analysis (ILRMA). The full-rank spatial model is also considered in many BSS approaches, such as full-rank spatial covariance matrix analysis (FCA), multichannel nonnegative matrix factorization (MNMF), and FastMNMF, which can improve the separation performance in the case of long reverberation times. This paper proposes a new MNMF framework based on the convolutive transfer function (CTF) for overdetermined BSS. The time-domain convolutive mixture model is approximated by a frequency-wise convolutive mixture model instead of the widely adopted frequency-wise instantaneous mixture model. The iterative projection algorithm is adopted to estimate the demixing matrix, and the multiplicative update rule is employed to estimate nonnegative matrix factorization (NMF) parameters. Finally, the source image is reconstructed using a multichannel Wiener filter. The advantages of the proposed method are twofold. First, the CTF approximation enables us to use a short window to represent long impulse responses. Second, the full-rank spatial model can be derived based on the CTF approximation and slowly time-variant source variances, and close relationships between the proposed method and ILRMA, FCA, MNMF and FastMNMF are revealed. Extensive experiments show that the proposed algorithm achieves a higher separation performance than ILRMA and FastMNMF in reverberant environments.