Much has been written about theory and practice in the law, and the tension between practitioners and theorists. Judges do not cite theoretical articles often; they rarely "apply" theories to particular cases. These arguments are not revisited. Instead the Essay explores the working and interaction of theory and practice, practitioners and theorists. The Essay starts with a story about solving a legal issue using our intellectual tools - theory, practice, and their progenies: experience and "gut." Next the Essay elaborates on the nature of theory, practice, experience and "gut." The third part of the Essay discusses theories that are helpful to practitioners and those that are less helpful. The Essay concludes that practitioners theorize, and theorists practice. They use these intellectual tools differently because the goals and orientations of theorists and practitioners, and the constraints under which they act, differ. Theory, practice, experience and "gut" help us think, remember, decide and create. They complement each other like the two sides of the same coin: distinct but inseparable.

Of Theory and Practice

In the development process of noise-reduction algorithms, an objective machine-driven intelligibility measure which shows high correlation with speech intelligibility is of great interest. Besides reducing time and costs compared to real listening experiments, an objective intelligibility measure could also help provide answers on how to improve the intelligibility of noisy unprocessed speech. In this paper, a short-time objective intelligibility measure (STOI) is presented, which shows high correlation with the intelligibility of noisy and time-frequency weighted noisy speech (e.g., resulting from noise reduction) of three different listening experiments. In general, STOI showed better correlation with speech intelligibility compared to five other reference objective intelligibility models. In contrast to other conventional intelligibility models which tend to rely on global statistics across entire sentences, STOI is based on shorter time segments (386 ms). Experiments indeed show that it is beneficial to take segment lengths of this order into account. In addition, a free Matlab implementation is provided.

An Algorithm for Intelligibility Prediction of Time–Frequency Weighted Noisy Speech

Digital processing of speech signals

Formulation of speech separation as a supervised learning problem has shown considerable promise. In its simplest form, a supervised learning algorithm, typically a deep neural network, is trained to learn a mapping from noisy features to a time-frequency representation of the target of interest. Traditionally, the ideal binary mask (IBM) is used as the target because of its simplicity and large speech intelligibility gains. The supervised learning framework, however, is not restricted to the use of binary targets. In this study, we evaluate and compare separation results by using different training targets, including the IBM, the target binary mask, the ideal ratio mask (IRM), the short-time Fourier transform spectral magnitude and its corresponding mask (FFT-MASK), and the Gammatone frequency power spectrum. Our results in various test conditions reveal that the two ratio mask targets, the IRM and the FFT-MASK, outperform the other targets in terms of objective intelligibility and quality metrics. In addition, we find that masking based targets, in general, are significantly better than spectral envelope based targets. We also present comparisons with recent methods in non-negative matrix factorization and speech enhancement, which show clear performance advantages of supervised speech separation.

/pdf/on-training-targets-for-supervised-speech-separation-1vd9kby2kl.pdf

On training targets for supervised speech separation

▪ Abstract Turbulence is ubiquitous in atmospheric clouds, which have enormous turbulence Reynolds numbers owing to the large range of spatial scales present. Indeed, the ratio of energy-containing and dissipative length scales is on the order of 105 for a typical convective cloud, with a corresponding large-eddy Reynolds number on the order of 106 to 107. A characteristic trait of high-Reynolds-number turbulence is strong intermittency in energy dissipation, Lagrangian acceleration, and scalar gradients at small scales. Microscale properties of clouds are determined to a great extent by thermodynamic and fluid-mechanical interactions between droplets and the surrounding air, all of which take place at small spatial scales. Furthermore, these microscale properties of clouds affect the efficiency with which clouds produce rain as well as the nature of their interaction with atmospheric radiation and chemical species. It is expected, therefore, that fine-scale turbulence is of direct importance to the evolu...

/pdf/particle-turbulence-interactions-in-atmospheric-clouds-3g20038icm.pdf

Particle-turbulence interactions in atmospheric clouds

This paper considers techniques for single-channel speech enhancement based on the discrete Fourier transform (DFT). Specifically, we derive minimum mean-square error (MMSE) estimators of speech DFT coefficient magnitudes as well as of complex-valued DFT coefficients based on two classes of generalized gamma distributions, under an additive Gaussian noise assumption. The resulting generalized DFT magnitude estimator has as a special case the existing scheme based on a Rayleigh speech prior, while the complex DFT estimators generalize existing schemes based on Gaussian, Laplacian, and Gamma speech priors. Extensive simulation experiments with speech signals degraded by various additive noise sources verify that significant improvements are possible with the more recent estimators based on super-Gaussian priors. The increase in perceptual evaluation of speech quality (PESQ) over the noisy signals is about 0.5 points for street noise and about 1 point for white noise, nearly independent of input signal-to-noise ratio (SNR). The assumptions made for deriving the complex DFT estimators are less accurate than those for the magnitude estimators, leading to a higher maximum achievable speech quality with the magnitude estimators.

Minimum Mean-Square Error Estimation of Discrete Fourier Coefficients With Generalized Gamma Priors

This paper considers estimation of the noise spectral variance from speech signals contaminated by highly nonstationary noise sources. The method can accurately track fast changes in noise power level (up to about 10 dB/s). In each time frame, for each frequency bin, the noise variance estimate is updated recursively with the minimum mean-square error (mmse) estimate of the current noise power. A time- and frequency-dependent smoothing parameter is used, which is varied according to an estimate of speech presence probability. In this way, the amount of speech power leaking into the noise estimates is kept low. For the estimation of the noise power, a spectral gain function is used, which is found by an iterative data-driven training method. The proposed noise tracking method is tested on various stationary and nonstationary noise sources, for a wide range of signal-to-noise ratios, and compared with two state-of-the-art methods. When used in a speech enhancement system, improvements in segmental signal-to-noise ratio of more than 1 dB can be obtained for the most nonstationary noise sources at high noise levels.

/pdf/tracking-of-nonstationary-noise-based-on-data-driven-u75ggffsb3.pdf

Tracking of Nonstationary Noise Based on Data-Driven Recursive Noise Power Estimation

A data-driven approach to optimizing spectral speech enhancement methods for various error criteria

This paper considers suppression of late reverberation and additive noise in single-channel speech recordings. The reverberation introduces long-term correlation in the observed signal. In the first part of this work, we show how this correlation can be used to estimate the late reverberant spectral variance (LRSV) without having to assume a specific model for the room impulse responses (RIRs) while no explicit estimates of RIR model parameters are needed. That makes this correlation-based approach more robust against RIR modeling errors. However, the correlation-based method can follow only slow time variations in the RIRs. Existing model-based methods use statistical models for the RIRs, that depend on one or more parameters that have to be estimated blindly. The common statistical models lead to simple expressions for the LRSV that depend on past values of the spectral variance of the reverberant, noise-free, signal. All existing model-based LRSV estimators in the literature are derived assuming the RIRs to be time-invariant realizations of a stochastic process. In the second part of this paper, we go one step further and analyze time-varying RIRs. We show that in this case the reverberance tends to become decorrelated. We discuss the relations between different RIR models and their corresponding LRSV estimators. We show theoretically that similar simple estimators exist as in the time-invariant case, provided that the reverberation time T60 and direct-to-reverberation ratio (DRR) of the RIRs remain nearly constant during an interval of the order of a few frames. We show that the reverberation time can be taken frequency-bin independent in DFT-based enhancement algorithms. Experiments with time-varying RIRs validate the analysis. Experiments with additive nonstationary noise and time-invariant RIRs show the influence of blind estimation of the reverberation time and the DRR.

Correlation-Based and Model-Based Blind Single-Channel Late-Reverberation Suppression in Noisy Time-Varying Acoustical Environments

A method is presented to obtain droplet concentration for water clouds from ground-based remote sensing observations. It relies on observations of cloud thickness, liquid water path, and optical extinction near the cloud base. The method was tested for two case studies (19 April 1996 and 4 September 1996) during the Clouds And Radiation experiment (CLARA). The CLARA experiment was designed to observe clouds using a variety of remote sensing instruments near the city of Delft in the western part of the Netherlands. The measurement of cloud thickness is dependent on the detection of cloud base by lidar and cloud top by radar. It is shown that during CLARA it was possible to detect cloud base with an uncertainty of less than 30 m using current lidar techniques. The agreement between in situ and remote sensing observations of droplet concentration was reasonable. An error analysis indicates that this method is most sensitive to uncertainties in liquid water path and the unknown effects of multiple scattering on lidar signal returns. When the liquid water path is very small the relative error of the liquid water path increases to unacceptable levels, so that the retrieval of droplet concentration becomes very difficult. The estimated uncertainty in the strength of multiple scattering can explain differences between observations and retrievals of droplet concentration on one day, but not the other.

J.S. Erkelens

Papers

Minimum Mean-Square Error Estimation of Discrete Fourier Coefficients With Generalized Gamma Priors

Tracking of Nonstationary Noise Based on Data-Driven Recursive Noise Power Estimation

A data-driven approach to optimizing spectral speech enhancement methods for various error criteria

Correlation-Based and Model-Based Blind Single-Channel Late-Reverberation Suppression in Noisy Time-Varying Acoustical Environments

Ground-Based Remote Sensing of Stratocumulus Properties during CLARA, 1996