Sample-accurate synchronization of nodes is required to enable the full potential of acoustic sensor networks for cooperative and enhanced signal acquisition. While metrics of spatio-temporal sensor utility are key to successful aggregation of sensor nodes, for instance, to perform sound localization or beamforming, the same is true for waveform-based assessment and compensation of sampling-rate offset (SRO). This paper therefore proposes an acoustic coherence state (ACS) metric to support systems for SRO estimation to integrate estimations of various utility due to nonpersistent acoustic activity and geometrical diversity. Specifically, we consider systems with SRO estimation and compensation in open- and closed-loop structures and outline the architectures for embedding ACS-based sensor utility. It is demonstrated in both cases that the acoustic coherence metric is more appropriate in terms of end-to-end synchronization performance than voice or sound activity detectors.

Distributed hardware of acoustic sensor networks bears inconsistency of local sampling frequencies, which is detrimental to signal processing. Fundamentally, sampling rate offset (SRO) nonlinearly relates the discrete-time signals acquired by different sensor nodes. As such, retrieval of SRO from the available signals requires non-linear estimation, like double-cross-correlation processing (DXCP), and frequently results in biased estimation. SRO compensation by asynchronous sampling rate conversion (ASRC) on the signals then leaves an unacceptable residual. As a remedy to this problem, multi-stage procedures have been devised to diminish the SRO residual with multiple iterations of SRO estimation and ASRC over the entire signal. This paper converts the mechanism of offline multi-stage processing into a continuous feedback-control loop comprising a controlled ASRC unit followed by an online implementation of DXCP-based SRO estimation. To support the design of an optimum internal model control unit for this closed-loop system, the paper deploys an analytical dynamical model of the proposed online DXCP. The resulting control architecture then merely applies a single treatment of each signal frame, while efficiently diminishing SRO bias with time. Evaluations with both speech and Gaussian input demonstrate that the high accuracy of multi-stage processing is maintained at the low complexity of single-stage (open-loop) processing.

Coherent processing of signals captured by a wireless acoustic sensor network (WASN) requires an estimation of unknown parameters such as the sampling-rate and sampling-time offset (SRO and STO) of asynchronous sensor clocks. Although some sophisticated techniques for blind parameter estimation have become available in this young field of research, further development of new methods is required especially regarding environmental robustness, online estimation, and computational efficiency. As the main contribution of this work, we therefore introduce a novel approach for blind SRO estimation in the spirit of both a recently available time-domain double-cross-correlation processor (DXCP) and the well-known generalized cross-correlation with phase transform (GCC-PhaT). For the proposed approach, called DXCP-PhaT, we specifically introduce secondary cross-quantities to restore ergodicity of the otherwise drifting cross-correlation function or cross-spectrum of asynchronous input signals, which is an important property for estimation. Based upon this theoretical contribution of the paper, we then derive the DXCP-PhaT algorithm in the STFT domain, which brings along a number of improvements over state-of-the-art: advanced accuracy in terms of SRO assessment, environmental robustness to larger microphone distances as of real sensor networks, realtime-applicability in terms of online architecture and reduced complexity, and eventually an extension with online STO estimation (up to the inherent ambiguity of digital STO and acoustic TDOA) reliant on SRO compensation. Those claims are confirmed by comprehensive experiments, both offline and online, on simulated data and real recordings from a Rasberry-Pi-based WASN.

Signal synchronization in wireless acoustic sensor networks requires an accurate estimation of the sampling rate offset (SRO) inevitably present in signals acquired by sensors of ad-hoc networks. Although some sophisticated methods for blind SRO estimation have been recently proposed in this very young field of research, there is still a need for the development of new ideas and concepts especially regarding robust approaches with low computational complexity. We therefore propose a novel time-domain method based on the calculation of a double-cross-correlation function in this contribution. Experimental evaluation of the introduced approach in a challenging acoustic environment and comparison with a state-of-the-art frequency-domain method confirms the high accuracy and low computational load of the proposed technique.

The benefits of both a logarithmic spectral amplitude (LSA) estimation and a modeling in a generalized spectral domain (where short-time amplitudes are raised to a generalized power exponent, not restricted to magnitude or power spectrum) are combined in this contribution to achieve a better tradeoff between speech quality and noise suppression in single-channel speech enhancement. A novel gain function is derived to enhance the logarithmic generalized spectral amplitudes of noisy speech. Experiments on the CHiME-3 dataset show that it outperforms the famous minimum mean squared error (MMSE) LSA gain function of Ephraim and Malah in terms of noise suppression by 1.4 dB, while the good speech quality of the MMSE-LSA estimator is maintained.

In this contribution we investigate a priori signal-to-noise ratio (SNR) estimation, a crucial component of a single-channel speech enhancement system based on spectral subtraction. The majority of the state-of-the art a priori SNR estimators work in the power spectral domain, which is, however, not confirmed to be the optimal domain for the estimation. Motivated by the generalized spectral subtraction rule, we show how the estimation of the a priori SNR can be formulated in the so called generalized SNR domain. This formulation allows to generalize the widely used decision directed (DD) approach. An experimental investigation with different noise types reveals the superiority of the generalized DD approach over the conventional DD approach in terms of both the mean opinion score - listening quality objective measure and the output global SNR in the medium to high input SNR regime, while we show that the power spectrum is the optimal domain for low SNR. We further develop a parameterization which adjusts the domain of estimation automatically according to the estimated input global SNR. Index Terms: single-channel speech enhancement, a priori SNR estimation, generalized spectral subtraction

We present a new beamformer front-end for Automatic Speech Recognition and apply it to the 3rd-CHiME Speech Separation and Recognition Challenge. Without any further modification of the back-end, we achieve a 53% relative reduction of the word error rate over the best baseline enhancement system for the relevant test data set. Our approach leverages the power of a bi-directional Long Short-Term Memory network to robustly estimate soft masks for a subsequent beamforming step. The utilized Generalized Eigenvalue beamforming operation with an optional Blind Analytic Normalization does not rely on a Direction-of-Arrival estimate and can cope with multi-path sound propagation, while at the same time only introducing very limited speech distortions. Our quite simple setup exploits the possibilities provided by simulated training data while still being able to generalize well to the fairly different real data. Finally, combining our front-end with data augmentation and another language model nearly yields a 64 % reduction of the word error rate on the real data test set.

Noise tracking is an important component of speech enhance-ment algorithms. Of the many noise trackers proposed, Min-imum Statistics (MS) is a particularly popular one due to its simple parameterization and at the same time excellent perfor-mance. In this paper we propose to further reduce the number of MS parameters by giving an alternative derivation of an optimal smoothing constant. At the same time the noise tracking perfor-mance is improved as is demonstrated by experiments employ-ing speech degraded by various noise types and at different SNR values. Index Terms: speech enhancement, noise tracking, optimal smoothing

In this contribution we derive a variational EM (VEM) algorithm for model selection in complex Watson mixture models, which have been recently proposed as a model of the distribution of normalized microphone array signals in the short-time Fourier transform domain. The VEM algorithm is applied to count the number of active sources in a speech mixture by iteratively estimating the mode vectors of the Watson distributions and suppressing the signals from the corresponding directions. A key theoretical contribution is the derivation of the MMSE estimate of a quadratic form involving the mode vector of the Watson distribution. The experimental results demonstrate the effectiveness of the source counting approach at moderately low SNR. It is further shown that the VEM algorithm is more robust with respect to used threshold values.

In this contribution we derive the Maximum A-Posteriori (MAP) estimates of the parameters of a Gaussian Mixture Model (GMM) in the presence of noisy observations. We assume the distortion to be white Gaussian noise of known mean and variance. An approximate conjugate prior of the GMM parameters is derived allowing for a computationally efficient implementation in a sequential estimation framework. Simulations on artificially generated data demonstrate the superiority of the proposed method compared to the Maximum Likelihood technique and to the ordinary MAP approach, whose estimates are corrected by the known statistics of the distortion in a straightforward manner.

In this paper we present an improved version of the recently proposed Maximum A-Posteriori (MAP) based noise power spectral density estimator. An empirical bias compensation and bandwidth adjustment reduce bias and variance of the noise variance estimates. The main advantage of the MAP-based postprocessor is its low estimation variance. The estimator is employed in the second stage of a two-stage single-channel speech enhancement system, where eight different state-of-the-art noise tracking algorithms were tested in the first stage. While the postprocessor hardly affects the results in stationary noise scenarios, it becomes the more effective the more nonstationary the noise is. The proposed postprocessor was able to improve all systems in babble noise w.r.t. the perceptual evaluation of speech quality performance.

In this paper we present a novel noise power spectral density tracking algorithm and its use in single-channel speech enhancement. It has the unique feature that it is able to track the noise statistics even if speech is dominant in a given time-frequency bin. As a consequence it can follow non-stationary noise superposed by speech, even in the critical case of rising noise power. The algorithm requires an initial estimate of the power spectrum of speech and is thus meant to be used as a postprocessor to a first speech enhancement stage. An experimental comparison with a state-of-the-art noise tracking algorithm demonstrates lower estimation errors under low SNR conditions and smaller fluctuations of the estimated values, resulting in improved speech quality as measured by PESQ scores.