Search

Numerous advanced and lightweight signal processing methods have been presented for single-channel speech enhancement (SE). It is imperative to carefully explore how to efficiently combine, integrate, and balance these methods. This paper proposes a more effective and less resource-intensive SE system, focused on the integration and adaptation of several approaches, especially the temporal cepstrum smoothing (TCS). First, a more robust fundamental frequency estimator is employed within TCS, mitigating the performance limitations caused by the inaccuracy of the original estimator. Additionally, a harmonic enhancement mechanism is introduced, effectively recovering the weak harmonic components. By incorporation of the modified TCS in the a posteriori speech presence probability estimation, the unbiased minimum mean square error noise power spectral density estimator can be refined. The modified TCS is also utilized for the a priori signal-to-noise ratio estimation. Moreover, this paper enhances the log-spectral amplitude estimator by applying both super-Gaussian speech priors and speech presence uncertainty for further improvement. Experimental evaluations demonstrate that the proposed method yields an improvement in speech quality while maintaining modest computational and storage requirements. Furthermore, the proposed system exhibits comparable performance to several baseline systems based on lightweight deep neural networks. [ABSTRACT FROM AUTHOR]

Concentric circular microphone arrays have been used in a wide range of applications, such as teleconferencing systems and smarthome devices for speech signal acquisition. Such arrays are generally designed with omnidirectional sensors, and the associated beamformers are fully steerable but only in the sensors' plane. If operated in the three-dimensional space, the performance of those arrays would suffer from significant degradation if the sound sources are out of the sensors' plane, which happens due to the incomplete spatial sampling of the sound field. This paper addresses this issue by presenting a new method to design concentric circular microphone arrays using both omnidirectional microphones and bidirectional microphones (directional sensors with dipole-shaped patterns). Such arrays are referred to as superarrays as they are able to achieve higher array gain as compared to their traditional counterparts with omnidirectional sensors. It is shown that, with the use of bidirectional microphones, the spatial harmonic components that are missing in the traditional arrays are compensated back. A beamforming method is then presented to design beamformers that can achieve frequency-invariant beampatterns with high directivity and are fully steerable in the three-dimensional space. Simulations and real experiments validate the effectiveness and good properties of the presented method. [ABSTRACT FROM AUTHOR]

Spatial information is important for human perception of speech and sound signals. However, this information is often either distorted or completely neglected in noise reduction because it is challenging, to say the least, to achieve optimal noise reduction and accurate spatial information preservation at the same time. This paper studies the problem of binaural speech enhancement. By jointly diagonalizing the speech and noise correlation matrices, we present a method to construct the noise reduction filter as a linear combination of different eigenvectors, which span a certain subspace of the entire space. A different dimension of the subspace gives a different trade-off between noise reduction and speech/noise spatial information preservation. On the one side, if the dimension is equal to 1, maximum noise reduction is achieved but at the price of significant spatial information distortion. On the other extreme, if the dimension of the subspace is equal to that of the entire space, spatial information is accurately preserved but at the cost of no noise reduction. Therefore, one can achieve different levels of compromises between the amount of noise reduction and the level of speech/noise spatial information preservation by adjusting the dimension of the used subspace. [ABSTRACT FROM AUTHOR]

This paper studies signal models for microphone array beamforming in the short-time-Fourier-transform (STFT) domain with long acoustic impulse responses. The major contributions are as follows. First, the signal modeling problem is investigated in the STFT domain and a general decomposition is proposed for the convolved source signal. Second, new insights into the array manifold are presented: the STFT of the windowed acoustic impulse response from the source to the sensors. Third, the structure of the reference signal is analyzed: it can be viewed as the output of a beamformer without considering the noise in the observation signal. Fourth, based on the new perspectives and decomposition, a signal model is derived based on the use of the superdirective beamformer. Finally, three performance measures are defined, based on which three optimal/suboptimal signal models are derived and their performance is assessed under different acoustic environments and analysis window lengths. The performance of the well-known minimum variance distortionless response (MVDR) beamformer is evaluated, which justifies the properties of the developed signal models. [ABSTRACT FROM AUTHOR]

Differential beamforming combined with microphone arrays can be used in a wide range of applications related to acoustic and speech signal acquisition and recovery. A practical and useful method for designing differential beamformers is the so-called null-constrained method, which was developed based on linear arrays and requires only the nulls' information from the target directivity pattern. While it is effective and easy to use, this method is found not suitable for designing steerable differential beamformers with circular arrays. This paper reexamines this technique in the context of circular differential microphone arrays. By analyzing the properties of the circular array topology, the null-constrained method is extended to include symmetric constraints, which is inherent in the design of circular arrays. This extension yields a design method for fully steerable differential beamformers that require only minimum information from the target beampattern. Simulations justify the theoretical analysis and demonstrate the good properties of the developed method. [ABSTRACT FROM AUTHOR]

The knowledge of speech presence probability (SPP) plays an essential role in noise estimation and speech enhancement. Single channel SPP estimation and centralized multi-channel SPP estimation have been well studied. However, how to estimate SPP in wireless acoustic sensor networks (WASNs) remains a great challenge and few efforts can be found in this topic, particularly for WASN applications with multiple speakers. Accordingly, this paper is devoted to the problem of SPP estimation in WASNs and it presents a distributed model-based SPP estimation method for multi-speaker detection, which does not need any fusion center. A distributed k-means clustering method is first used to cluster the nodes into subnetworks, which detect different speakers. For each node in the subnetwork, the speech and noise power spectral densities are estimated locally by using a model-based method, then a distributed SPP estimator is developed and applied in every subnetwork. A distributed consensus method is used to obtain the distributed clustering and the distributed SPP estimation. Simulation results show that the proposed distributed clustering method can assign nodes into subnetworks based on their noisy observations. Moreover, the proposed distributed SPP estimator achieves robust speech detection performance under different noise conditions. [ABSTRACT FROM AUTHOR]

Microphone arrays are typically used in room acoustic environments to acquire high fidelity audio and speech signals while suppressing noise, interference, and reverberation. In many application scenarios, interference and reverberation may mainly come from a certain region, and it is therefore necessary to develop beamformers that can preserve signals of interest while minimizing the power of signals coming from the region where interference and reverberation dominate. For this purpose, this paper first reexamines the so-called front-to-back ratio and the classical supercardioid beamformer. To deal with the white noise amplification problem and the limited directivity factor associated with the supercardioid beamformer, a set of reduced-rank beamformers are deduced by using the well-known joint diagonalization technique, which can make compromises between the front-to-back ratio and the amount of white noise amplification or the directivity factor. Then, the definition of the front-to-back ratio is extended to a generalized version, from which another set of reduced-rank beamformers and their regularized versions are developed. Simulations are conducted to illustrate the properties and advantages of the proposed beamformers. [ABSTRACT FROM AUTHOR]

This letter deals with the problem of differential beamforming with microphone arrays of arbitrary planar geometry. By approximating the beampattern with the Jacobi-Anger expansion, it develops an algorithm that can form any specified frequency-invariant beampattern with a microphone array of any planar geometry as long as the sensors' coordinates are given and the spacing between neighboring sensors is smaller than the smallest wavelength. This method is rather general and it can be used to design differential beamformers with linear, circular, and concentric circular differential microphone arrays as well as differential arrays of arbitrary planar geometry where sensors are placed in any specified positions. [ABSTRACT FROM AUTHOR]

The maximum directivity (MD) beamformer with spherical microphone arrays has many salient features in processing broadband acoustic and speech signals while suppressing noise and reverberation; but it is sensitive to sensors' self-noise and mismatch among these sensors. One effective way to deal with this sensitivity is by increasing the number of microphones, thereby improving the so-called white noise gain (WNG), but this increase may lead to many other design issues in terms of cost, array aperture, and possibly other performance degradation. This paper is tackling this sensitivity problem and presents a flexible high directivity (HD) beamforming algorithm. By approximating the ideal directivity pattern and the beamformer's beampattern with spherical harmonic series, the relationship between the two is obtained. This relationship is subsequently used to deduce a flexible HD beamformer, which can improve the WNG while achieving a directivity factor (DF) between the DF of an Nth-order MD beamformer and that of an ( N − 1 ) th-order MD one. Also derived is the analytical link between the DF and a tuning parameter and the link between the WNG and this parameter. Based on these links, one can easily determine the optimal value of the tuning parameter once the value of the DF or the WNG is specified. [ABSTRACT FROM AUTHOR]

This paper develops an approach to beamforming with small spacing uniform linear microphone arrays based on the null-steering (NS) principle. It first formulates the beamforming problem from the conventional mean-squared error (MSE) criterion and its normalized version. Several NS algorithms are then derived for beamforming with the constraint of placing nulls to either a single direction or multiple angles. The difference and relationships between different algorithms are discussed and their performances are evaluated. These algorithms can be used to design either fixed or adaptive beamformers. In the former case, the resulting beamformers work as differential microphone arrays (DMAs) since they exhibit frequency-invariant beampatterns and their directivity factors are high with a given number of sensors. In the latter case, the resulting beamformers can be viewed as a combination of DMAs and single-channel noise reduction since they do not only exhibit frequency-invariant beampatterns but also can achieve noise reduction based on the noise statistics. [ABSTRACT FROM AUTHOR]

This paper studies the problem of single-channel noise reduction in the time domain and presents a block-based approach where a vector of the desired speech signal is recovered by filtering a frame of the noisy signal with a rectangular filtering matrix. With this formulation, the noise reduction problem becomes one of estimating an optimal filtering matrix. To achieve such estimation, a method is introduced to decompose a frame of the clean speech signal into two orthogonal components: One correlated and the other uncorrelated with the current desired speech vector to be estimated. Different optimization cost functions are then formulated from which non-causal optimal filtering matrices are derived. The relationships among these optimal filtering matrices are discussed. In comparison with the classical sample-based technique that uses only forward prediction, the block-based method presented in this paper exploits both the forward and backward prediction as well as the temporal interpolation and, therefore, can improve the noise reduction performance by fully taking advantage of the speech property of self correlation. There is also a side advantage of this block-based method as compared to the sample-based technique, i.e., it is computationally more efficient and, as a result, more suitable for practical implementation. [ABSTRACT FROM AUTHOR]

This paper addresses the problem of noise reduction in the time domain where the clean speech sample at every time instant is estimated by filtering a vector of the noisy speech signal. Such a clean speech estimate consists of both the filtered speech and residual noise (filtered noise) as the noisy vector is the sum of the clean speech and noise vectors. Traditionally, the filtered speech is treated as the desired signal after noise reduction. This paper proposes to decompose the clean speech vector into two orthogonal components: one is correlated and the other is uncorrelated with the current clean speech sample. While the correlated component helps estimate the clean speech, it is shown that the uncorrelated component interferes with the estimation, just as the additive noise. Based on this orthogonal decomposition, the paper presents a way to define the error signal and cost functions and addresses the issue of how to design different optimal noise reduction filters by optimizing these cost functions. Specifically, it discusses how to design the maximum SNR filter, the Wiener filter, the minimum variance distortionless response (MVDR) filter, the tradeoff filter, and the linearly constrained minimum variance (LCMV) filter. It demonstrates that the maximum SNR, Wiener, MVDR, and tradeoff filters are identical up to a scaling factor. It also shows from the orthogonal decomposition that many performance measures can be defined, which seem to be more appropriate than the traditional ones for the evaluation of the noise reduction filters. [ABSTRACT FROM AUTHOR]

We proposed a multi-frequency sandwich type vibration energy harvester to widen the effective frequency range of vibration energy harvester. The harvester is composed of three resonant structures with different natural frequencies. The resonant structures are two cantilevers each with bi-layer coils and a plane spring with a magnet. The maximum peak-peak voltages of the three different frequencies are 172 mV, 104 mV, and 112 mV at the frequencies of 235 Hz, 330 Hz, and 430 Hz, respectively. The first maximum voltage is much higher than the others, because the coils in both cantilevers can produce voltages. [ABSTRACT FROM AUTHOR]

Blood analysis plays a major role in medical and science applications and white blood cells (WBCs) are an important target of analysis. We proposed an integrated microfluidic chip for direct and rapid trapping WBCs from whole blood. The microfluidic chip consists of two basic functional units: a winding channel to mix and arrays of two-layer trapping structures to trap WBCs. Red blood cells (RBCs) were eliminated through moving the winding channel and then WBCs were trapped by the arrays of trapping structures. We fabricated the PDMS (polydimethylsiloxane) chip using soft lithography and determined the critical flow velocities of tartrazine and brilliant blue water mixing and whole blood and red blood cell lysis buffer mixing in the winding channel. They are 0.25 μl/min and 0.05 μl/min, respectively. The critical flow velocity of the whole blood and red blood cell lysis buffer is lower due to larger volume of the RBCs and higher kinematic viscosity of the whole blood. The time taken for complete lysis of whole blood was about 85 s under the flow velocity 0.05 μl/min. The RBCs were lysed completely by mixing and the WBCs were trapped by the trapping structures. The chip trapped about 2.0 × 103 from 3.3 × 103 WBCs. [ABSTRACT FROM AUTHOR]

Circular differential microphone arrays (CDMAs) have been extensively studied in speech and audio applications for their steering flexibility, potential to achieve frequency-invariant directivity patterns, and high directivity factors (DFs). However, CDMAs suffer from both white noise amplification and deep nulls in the DF and in the white noise gain (WNG) due to spatial aliasing, which considerably restricts their use in practical systems. The minimum-norm filter can improve the WNG by using more microphones than required for a given differential array order; but this filter increases the array aperture (radius), which exacerbates the spatial aliasing problem and worsens the nulls problem in the DF. Through theoretical analysis, this research finds that the nulls of the CDMAs are caused by the zeros in the denominators of the filters' coefficients, i.e., the zeros of the Bessel function. To deal with both the white noise amplification and deep nulls problems, this paper develops an approach that combines different rings of microphones together with appropriate radii. The resulting robust concentric circular differential microphone arrays (CCDMAs) can mitigate both problems. Simulation results justify the superiority of the robust CCDMA approach over the traditional CDMAs and robust CDMAs.

This paper presents an approach to the direction-of-arrival (DOA) estimation problem in acoustic environments using microphone arrays. It works in the short-time Fourier transform (STFT) domain. It first transforms the noisy speech signals received at the array into the STFT domain. A Householder transformation is then constructed and applied to the multichannel STFT coefficients in each subband. This transformation converts the multichannel STFT coefficients into two components: one is a single coefficient that is dominated by the signal of interest and the other consists of the M - 1 coefficient that is dominated by noise (or even consists of noise-only if there is no reverberation), where M is the number of sensors. A cost function is then formed from the outputs of the Householder transformation and the DOA information can subsequently be obtained by searching the extremum value of this cost function in the angle range between 0° and 180°. Simulation results are provided to illustrate the performance of this approach.

Differential microphone arrays have the potential to be widely deployed in hands-free communication systems thanks to their frequency-invariant beampatterns, high directivity factors, and small apertures. Traditionally, they are designed and implemented in a multistage way with uniform linear geometries. This paper presents an approach to the design of differential microphone arrays with orthogonal polynomials, more specifically with Jacobi polynomials. It first shows how to express the beampatterns as a function of orthogonal polynomials. Then several differential beamformers are derived and their performance depends on the parameters of the Jacobi polynomials. Simulations show the great flexibility of the proposed method in terms of designing any order differential microphone arrays with different beampatterns and controlling white noise gain.

This paper develops a multistage approach to the implementation of the minimum variance distortionless response (MVDR) beamformer. It first divides the microphone array of M sensors into M/2 subarrays with each subarray having only two microphones, and a two-channel MVDR beamformer is performed with each subarray. The M/2 subarrays' outputs are then treated as the inputs of M/4 subarrays of two channels in the next stage. Similarly, a two-channel MVDR beamformer is performed with each subarray in the second stage. This process is repeated till the last stage that has only a single output. This multistage MVDR beamformer has the following properties: (1) Its performance is identical to that of the conventional MVDR beamformer in spatially uncorrelated noise; (2) it is much more robust than the conventional MVDR beamformer in diffuse noise, i.e., it has a significantly higher white noise gain as compared to the traditional MVDR beamformer; and (3) its complexity is an order of magnitude smaller than that of the traditional MVDR beamformer. This basic principle can also be easily generalized to the case where every subarray has more than two microphones.

This paper proposes a sparse linear prediction based algorithm to estimate time difference of arrival. This algorithm unifies the cross correlation method without prewhitening and that with prewhitening via an ℓ2/ℓ1 optimization process, which is solved by an augmented Lagrangian alternating direction method. It also forms a set of time delay estimators that make a tradeoff between prewhitening and non-prewhitening through adjusting a regularization parameter. The effectiveness of the proposed algorithm is demonstrated in noisy and reverberant environments.

Differential microphone array (DMA), a particular kind of sensor array that is responsive to the differential sound pressure field, has a broad range of applications in sound recording, noise reduction, signal separation, dereverberation, etc. Traditionally, an Nth-order DMA is formed by combining, in a linear manner, the outputs of a number of DMAs up to (including) the order of N - 1. This method, though simple and easy to implement, suffers from a number of drawbacks and practical limitations. This paper presents an approach to the design of linear DMAs. The proposed technique first transforms the microphone array signals into the short-time Fourier transform (STFT) domain and then converts the DMA beamforming design to simple linear systems to solve. It is shown that this approach is much more flexible as compared to the traditional methods in the design of different directivity patterns. Methods are also presented to deal with the white noise amplification problem that is considered to be the biggest hurdle for DMAs, particularly higher-order implementations.

Blind multichannel identification is generally sensitive to background noise. Although there have been some efforts in the literature devoted to improving the robustness of blind multichannel identification with respect to noise, most of those works assume that the noise is Gaussian distributed, which is often not valid in real room acoustic environments. This paper deals with the more practical scenario where the noise is not Gaussian. To improve the robustness of blind multichannel identification to non-Gaussian noise, a robust normalized multichannel frequency-domain least-mean M-estimate algorithm is developed. Unlike the traditional approaches that use the squared error as the cost function, the proposed algorithm uses an M-estimator to form the cost function, which is shown to be immune to non-Gaussian noise with a symmetric α-stable distribution. Experiments based on the identification of a single-input/multiple-output acoustic system demonstrate the robustness of the proposed algorithm.