Lucent Technologies Inc.

American Telephone & Telegraph Co., At&t Bell Laboratories

Oded Ghitza

Peter Kroon

Doh-Suk Kim

Jont B Allen

A mapping function is generated between subjective measures of audio signal quality, e.g., mean opinion score (MOS) or degradation MOS (DMOS) measures, and corresponding objective distortion measures, e.g., auditory speech quality measures (ASQMs) or perceptual speech quality measures (PSQMs), for known audio signals. The subjective measures and corresponding objective distortion measures are determined in accordance with modulated noise reference unit (MNRU) conditions or other suitable distortion conditions placed on the source speech, and a regression analysis is applied to the results to generate the mapping function. The mapping function may then be utilized, e.g., to evaluate speech quality of additional source speech from a particular speech coding system. In this case, the objective distortion measure is generated using the additional source speech, and the resulting objective measure is applied as an input to the mapping function to generate an estimate of the value of the subjective measure. Advantageously, the mapping function is database-independent, and can thus be used, e.g., to generate accurate estimates of subjective measures of speech quality for speech databases unrelated to those used in generating the mapping function.

Method and apparatus for estimating subjective audio signal quality from objective distortion measures

A sensory type pattern such as a speech or other sound pattern is analyzed to obtain the spectral distribution of the neural response thereto. A plurality of logarithmically related neural response intensity threshold signals is formed. The frequency spectrum of the sensory type pattern is divided into a plurality of overlapping spectral portions and the waveform of each prescribed spectral portion is partitioned into successive time segments. For the current time segment of each spectral portion waveform, the time intervals between crossings of the neural response intensity threshold level signals by the spectral portion waveform are detected and signals representative of the counts of inverse time intervals between the crossings of the plurality of levels are generated to form an inverse time interval histogram for the spectral portion. The inverse time interval histogram signals for the plurality of spectral portions are combined to produce a signal corresponding to the spectral distribution of the neural response to the sensory type pattern of the time segment.

Growing community of inventors

Westfield, NJ, United States of America

Oded Ghitza