Previous objective speech quality assessment models, such as bark spectral distortion (BSD), the perceptual speech quality measure (PSQM), and measuring normalizing blocks (MNB), have been found to be suitable for assessing only a limited range of distortions. A new model has therefore been developed for use across a wider range of network conditions, including analogue connections, codecs, packet loss and variable delay. Known as perceptual evaluation of speech quality (PESQ), it is the result of integration of the perceptual analysis measurement system (PAMS) and PSQM99, an enhanced version of PSQM. PESQ is expected to become a new ITU-T recommendation P.862, replacing P.861 which specified PSQM and MNB.

http://www.recursosvoip.com/docs/english/pap465.pdf

Perceptual evaluation of speech quality (PESQ)-a new method for speech quality assessment of telephone networks and codecs

In this paper, we evaluate the performance of several objective measures in terms of predicting the quality of noisy speech enhanced by noise suppression algorithms. The objective measures considered a wide range of distortions introduced by four types of real-world noise at two signal-to-noise ratio levels by four classes of speech enhancement algorithms: spectral subtractive, subspace, statistical-model based, and Wiener algorithms. The subjective quality ratings were obtained using the ITU-T P.835 methodology designed to evaluate the quality of enhanced speech along three dimensions: signal distortion, noise distortion, and overall quality. This paper reports on the evaluation of correlations of several objective measures with these three subjective rating scales. Several new composite objective measures are also proposed by combining the individual objective measures using nonparametric and parametric regression analysis techniques.

Evaluation of Objective Quality Measures for Speech Enhancement

Probabilistic inference is the problem of estimating the hidden variables (states or parameters) of a system in an optimal and consistent fashion as a set of noisy or incomplete observations of the system becomes available online. The optimal solution to this problem is given by the recursive Bayesian estimation algorithm which recursively updates the posterior density of the system state as new observations arrive. This posterior density constitutes the complete solution to the probabilistic inference problem, and allows us to calculate any “optimal” estimate of the state. Unfortunately, for most real-world problems, the optimal Bayesian recursion is intractable and approximate solutions must be used. Within the space of approximate solutions, the extended Kalman filter (EKF) has become one of the most widely used algorithms with applications in state, parameter and dual estimation. Unfortunately, the EKF is based on a sub-optimal implementation of the recursive Bayesian estimation framework applied to Gaussian random variables. This can seriously affect the accuracy or even lead to divergence of any inference system that is based on the EKF or that uses the EKF as a component part. Recently a number of related novel, more accurate and theoretically better motivated algorithmic alternatives to the EKF have surfaced in the literature, with specific application to state estimation for automatic control. We have extended these algorithms, all based on derivativeless deterministic sampling based approximations of the relevant Gaussian statistics, to a family of algorithms called Sigma-Point Kalman Filters (SPKF). Furthermore, we successfully expanded the use of this group of algorithms (SPKFs) within the general field of probabilistic inference and machine learning, both as stand-alone filters and as subcomponents of more powerful sequential Monte Carlo methods (particle filters). We have consistently shown that there are large performance benefits to be gained by applying Sigma-Point Kalman filters to areas where EKFs have been used as the de facto standard in the past, as well as in new areas where the use of the EKF is impossible.

/pdf/sigma-point-kalman-filters-for-probabilistic-inference-in-1affrc62j6.pdf

Sigma-point kalman filters for probabilistic inference in dynamic state-space models

Single-channel, speaker-independent speech separation methods have recently seen great progress. However, the accuracy, latency, and computational cost of such methods remain insufficient. The majority of the previous methods have formulated the separation problem through the time-frequency representation of the mixed signal, which has several drawbacks, including the decoupling of the phase and magnitude of the signal, the suboptimality of time-frequency representation for speech separation, and the long latency in calculating the spectrograms. To address these shortcomings, we propose a fully-convolutional time-domain audio separation network (Conv-TasNet), a deep learning framework for end-to-end time-domain speech separation. Conv-TasNet uses a linear encoder to generate a representation of the speech waveform optimized for separating individual speakers. Speaker separation is achieved by applying a set of weighting functions (masks) to the encoder output. The modified encoder representations are then inverted back to the waveforms using a linear decoder. The masks are found using a temporal convolutional network (TCN) consisting of stacked 1-D dilated convolutional blocks, which allows the network to model the long-term dependencies of the speech signal while maintaining a small model size. The proposed Conv-TasNet system significantly outperforms previous time-frequency masking methods in separating two- and three-speaker mixtures. Additionally, Conv-TasNet surpasses several ideal time-frequency magnitude masks in two-speaker speech separation as evaluated by both objective distortion measures and subjective quality assessment by human listeners. Finally, Conv-TasNet has a significantly smaller model size and a shorter minimum latency, making it a suitable solution for both offline and real-time speech separation applications.

/pdf/conv-tasnet-surpassing-ideal-time-frequency-magnitude-3fbpvuuk37.pdf

Conv-TasNet: Surpassing Ideal Time-Frequency Magnitude Masking for Speech Separation

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

A new model for the perceptual evaluation of speech quality (PESQ) was recently standardized by the International Telecommunications Union as Recommendation P.862. Unlike previous codec assessment models, such as PSQM and MNB (ITU-T P.861), PESQ is able to predict subjective quality with good correlation in a very wide range of conditions, which may include coding distortions, errors, noise, filtering, delay, and variable delay. The psychoacoustic model used in PESQ is introduced. Part I describes the time-delay identification technique that is used in combination with the PESQ psychoacoustic model to predict the end-to-end perceived speech quality.

/pdf/perceptual-evaluation-of-speech-quality-pesq-the-new-itu-4025ck0arp.pdf

Perceptual Evaluation of Speech Quality (PESQ) The New ITU Standard for End-to-End Speech Quality Assessment Part II: Psychoacoustic Model

A new model for the perceptual evaluation of speech quality (PESQ) was recently standardized by the International Telecommunications Union as Recommendation P.862. Unlike previous codec assessment models, such as PSQM and MNB (ITU-T P.861), PESQ is able to predict subjective quality with good correlation in a very wide range of conditions, which may include coding distortions, errors, noise, filtering, delay, and variable delay. In Part I time-delay identification techniques are introduced and some causes of variable delay are outlined before the processes that are integrated into PESQ and specified in P.862 are described. More information on the structure of PESQ as well as performance results will be given in Part II.

Perceptual evaluation of speech quality (PESQ) the new ITU standard for end-to-end speech quality assessment: Part I: Time-delay compensation

PESQ-The New ITU Standard for End-to-End Speech Quality Assessment

A test apparatus (1) comprises a test signal generator (21) which transmits a test stimulus comprising two signals RR, S. The signal S is transmitted to a customer equipment interface (2), for transmission over the system (5) under test to a complementary remote apparatus (1a). The test apparatus (1) further comprises two inputs which are to be processed by a number of functional elements. The first input is the reference signal RR. The other input is a corresponding test signal R' having been generated at the remote test apparatus (1a) as a test signal R, R' having been generated at the remote test apparatus (1a) as a test signal R, identical with the reference signal RR, and transmitted over the system (5) under tests, to be processed by the customer interface (2). These two versions RR, R' of the same test signal are compared in the test apparatus (1). Firstly, the signals RR, R' are processed by a psycho-acoustic model (22), the output of which is transmitted to a perceptual layer (23), and then to a conversion unit (24) which also receives a further input from a non-optimum loudness rating unit (25). The output from the conversion unit (24) is transmitted to a signal quality determination unit (26), which also receives an input from the customer equipment interface (2). The non-optimum loudness rating unit (25) processes inputs from the customer equipment interface (2) and signal generator (21). During the iterative process to be described, the signal generator (21) receives feedback control from the signal quality determination unit (26). The remote test apparatus (1a) similarly receives a modified version of signal S (viz S') from the network (5) and compares it with a locally-generated reference version SR. In an alternative arrangement (figure 11), the same test apparatus (1) is connected to both test points (4, 4a) of the network (5) and processes both signals S and R, either serially or in parallel.

Michael Peter Hollier

Papers

Perceptual evaluation of speech quality (PESQ)-a new method for speech quality assessment of telephone networks and codecs

Perceptual Evaluation of Speech Quality (PESQ) The New ITU Standard for End-to-End Speech Quality Assessment Part II: Psychoacoustic Model

Perceptual evaluation of speech quality (PESQ) the new ITU standard for end-to-end speech quality assessment: Part I: Time-delay compensation

PESQ-The New ITU Standard for End-to-End Speech Quality Assessment

Measurement of signal quality