Stephen D. Voran

Objective quality assessment of digitally transmitted video

The authors discuss an automated, objective video quality measurement system being developed for use in measuring end-to-end user performance. The method is based on extraction and classification of features from sampled input and output video. The extracted features quantify many of the distortions present in modern digital compression and transmission systems. These distortions include video compression artifacts resulting from spatial compression (blocking, edge blurring/smearing, etc.) and temporal compression (image persistence, jerky motion, etc.). Since the quality of the motion and still portions of a video scene can differ dramatically, an algorithm that segments each video frame into motion and still parts has been developed.<<ETX>>

The development of objective video quality measures that emulate human perception

A method for deriving an objective video quality metric that emulates the human video quality metric is described. The objective metric comprises a family of measurements that quantify spatial, temporal, luminance, and chrominance distortions, followed by a prediction algorithm. The intent is that the derived metric will work well over a wide range of digital and analog video transmission and storage technologies. After describing the need for these measures, a detailed description of the derivation process and some preliminary results are given.<<ETX>>

Estimation of perceived speech quality using measuring normalizing blocks

We describe a new approach to the estimation of perceived speech quality. The approach uses a simple, but effective, perceptual transformation to emulate hearing and a hierarchy of measuring normalizing blocks (MNBs) to emulate auditory judgment. The resulting estimates were correlated with the results of seven subjective listening tests. Together, these seven tests include 182, 4-kHz bandwidth speech codecs, transmission systems, and reference conditions, with bit-rates ranging from 2.4 to 64 kbps. When compared with six other estimators, the MNB approach offers significant improvements in many cases, particularly at lower bit-rates, and when bit errors or frame erasures are present.

Listener ratings of speech passbands

We describe a listening experiment that measures the perceived speech quality of 19 speech passbands using 8 talkers and 28 listeners. Results are referenced to the traditional wide-band and narrow-band telephony passbands. Our findings may help those who wish to select new speech coding passbands that maximize perceived speech quality under bit-rate constraints. We identify several passbands that show particular promise.

User-oriented measures of telecommunication quality

With the deregulation and emergence of new telecommunication providers in many countries, national and international standards committees have assumed increased responsibility for the cooperative planning of new technology development and the matching of multivendor service offerings with user needs. One important focus of this effort has been the standardization of user-oriented, technology-independent measures of telecommunication service quality. The standardized measures will be used by providers to design and implement telecommunication systems and services, and by users to define telecommunication requirements and select the products that most effectively meet them. The authors discuss in particular call processing, data communication quality, video and voice quality measures.<<ETX>>

An objective technique for assessing video impairments

An objective video impairment assessment technique that appears to quantify the perceptual impact of video impairments in an accurate way is presented. The development uses 132 impaired video sequences that cover a remarkably wide range of motion, detail, impairment type, and impairment level. The approach discussed in based on digital image processing operations performed on digitized original and impaired video sequences. Measurements that quantify perceptual video attributes in both the spatial and the temporal domains are extracted from the digitized video. These measurements are then used to compute a single score that quantifies the perceptual impact of the impairments present in the video sequence. This objective impairment score is well-correlated (r=.92) with impairment assessments made by human viewers. Thus, it can be used to augment, or possibly to replace, the expensive and time-consuming subjective viewing tests that are typically used to evaluate video coding and transmission techniques.<<ETX>>

Perception-based bit-allocation algorithms for audio coding

We describe six algorithms for bit-allocation in audio coding. Each algorithm stems from the minimization of a different perceptually-motivated objective function. Three of these objective functions are extensions of existing ones, and three are new. Closed-form bit-allocation equations result in five cases, and an iterative approach is required in the sixth.

Observations on auditory excitation and masking patterns

Excitation patterns and masking patterns are used extensively in perceptual audio coders and quality assessment algorithms. Numerous algorithms for calculating these patterns have been proposed. This paper provides comparisons among the patterns generated by several of these algorithms. The comparisons are based on audio program material, rather than tones and noise. Explored areas include synthesis functions, spreading functions, masking indices, tonality measures, and the treatment of the absolute threshold of hearing. Several mathematical relations are provided to characterize observations in these areas. Patterns from simpler algorithms are considered as approximations to patterns from more complex algorithms, and the approximation error is characterized. Results may be useful to those who apply auditory excitation or masking patterns in their work.

Exploration of the additivity approximation for spectral magnitudes

The separation of acoustic signals is often accomplished through subtractive decompositions of frequency-domain representations. This is typically enabled by the zero phase approximation or the un-correlated signals approximation but both of these are very coarse approximations in the mathematical sense. We investigate this disconnect between what works in practice and what is mathematically correct. We conduct a broad search for a domain where the addi-tivity of spectral magnitudes is best satisfied. We apply objective estimators to time-domain reconstructions to characterize the true auditory impact of the magnitude additivity approximation. Our results show the auditory impacts of additivity approximations and allow comparison with the impact of using mixture phase and exact magnitudes in the time-domain reconstruction.

Using articulation index band correlations to objectively estimate speech intelligibility consistent with the modified rhyme test

We present an objective estimator of speech intelligibility that follows the paradigm of the Modified Rhyme Test (MRT). For each input, the estimator uses temporal correlations within articulation index bands to select one of six possible words from a list. The rate of successful word identification becomes the measure of speech intelligibility, as in the MRT. The estimator is called Articulation Band Correlation MRT (ABC-MRT). It consumes a tiny fraction of the resources required by MRT testing. ABC-MRT has been tested on a wide range of impaired speech recordings unseen during development. The resulting Pearson correlations between ABC-MRT and MRT results range from .95 to .99. These values exceed those of the other estimators tested.