Nicolas Côté

Speech Quality Estimation: Models and Trends

This article presents a tutorial overview of models for estimating the quality experienced by users of speech transmission and communication services. Such models can be classified as either parametric or signal based. Signal-based models use input speech signals measured at the electrical or acoustic interfaces of the transmission channel. Parametric models, on the other hand, depend on signal and system parameters estimated during network planning or at run time. This tutorial describes the underlying principles as well as advantages and limitations of existing models. It also presents new developments, thus serving as a guide to an appropriate usage of the multitude of current and emerging speech quality models.

Parameter-based prediction of speech quality in listening context—Towards a WB E-model

The paper presents a first version of a wideband E-model applicable to planning future telephone networks in a listening-only context. It allows to predict wideband speech quality under bandwidth restrictions and wideband and narrowband speech coding with and without transmission errors, including the quality impact due to noise. The paper presents the different model components and discusses their prediction performance for a large set of listening tests. Based on a discussion of the proposed listening model, an outlook on required future research including the conversational situation is provided.

Towards a universal scale for perceptual value

This paper addresses the problem of context effects in psychophysical measurement, and how these effects are reflected by MOS values obtained in auditory tests. Based on a simple model of the perception and judgment processes which are relevant in a subjective quality-judgment situation, it advocates for using a universal scale for the value which is related to the perceptual event, and which has a meaning beyond the specific test context. The E-models transmission rating scale [1] is considered to be a candidate for such a scale.

Instrumental estimation of E-model parameters for wideband speech codecs

A method is described for quantifying the quality of wideband speech codecs. Two parameters are derived from signal-based speech quality model estimations: (i) a wideband equipment impairment factor and (ii) a wideband packet-loss robustness factor . The equipment impairment factor can be combined with impairment factors for other quality degradations to form an estimate of the overall conversational quality of a wideband communication scenario, using a wideband extension of the E-model. The packet-loss robustness factor captures the robustness of the codec against packet-loss degradations. In contrast to past work, these parameters are no longer determined on the basis of auditory test results, but from signal-based speech quality models. We applied three intrusive models to several databases and compared the derived quality estimates and impairment factors to those obtained from auditory tests. The results show that when migrating from narrowband to wideband transmission—a quality improvement of roughly 30% can be obtained, which is very similar to the one observed in auditory tests. The estimated impairment factors show a high correlation to those derived from auditory scores. Congruences and discrepancies to auditory test results are discussed, and an outline of work necessary to set up a wideband or even superwideband E-model is given.

Optimization of acoustic absorption by green walls made of foliage and substrate

Green walls may absorb sound and contribute to noise reduction in urban areas. Recent experiments demonstrate that a foliage layer placed above a substrate layer may lead to a significant increase of acoustic absorption coefficient in a broad frequency range. However, the physical origin of this improvement remains unclear. In this work, measurements are carried out in an impedance tube on foliage, substrate, and foliage/substrate samples using the three-microphone two-load method. Acoustic absorption coefficient and surface specific impedance are measured in rigid backing condition between 100 and 1000 Hz. Effective speed of sound and characteristic impedance are also experimentally determined for foliage and substrate. For foliage/substrate samples, a good agreement is obtained between measured acoustic absorption coefficients and calculated ones using the effective properties of foliage and substrate layer and matrix manipulations. Analysis of results reveals that absorption coefficient spectrum is mai...

A comparison between two methods for speech transmission quality assessment with noise simulation at receiver's side

Two auditory experiments are presented that were designed in order to compare two different methods for the assessment of speech quality when noise at listeners side is present: (1) Headphone presentation with samples pre‐recorded over a head and torso simulator, and (2) test participants located in a noisy environment, rendered by a 4.1 loudspeaker system. Real‐world types of background noise were employed as well as a variety of user terminals. The correlation between corresponding overall quality scores amounts to r = 0.9. Hence, a headphone presentation can replace a loudspeaker noise simulation in principle, leading to the consequence that a simpler experimental set‐up is sufficient for assessing the speech quality when noise is present at the receivers side. However, there exist certain differences between both quality scores. For example, in most cases clean wideband (50‐7000 Hz) and standard narrow‐band (300‐3400 Hz) speech are rated slightly lower if the samples are presented over a headphone a...