Objective Modeling of Speech Quality with a Psychoacoustically Validated Auditory Model
A new objective measure for the transmission quality of low-bit-rate speech coding algorithms is described and tested. A quantitative psychoacoustical signal processing model is employed to measure the perceptually relevant deviations between the transmitted, degraded signal and the corresponding reference signal objectively. The processing is applied to transform the coded (distorted) signal and the corresponding original speech signal to an internal representation which is thought of as the information that is accessible to higher neural stages of perception. From a comparison of these internal representations a quality measure can be derived, which shows a high correlation with the subjective mean opinion score data of various test databases if a frequency-dependent weighting is applied that exhibits increasing weights for increasing center frequencies of the filter channels of the internal representation. The inherent parameters of the auditory processing model optimized from psychoacoustical data independent from the present study yielded also the best performance in predicting the subjective speech quality data compared to alterations of the auditory model. The performance of the new measure was investigated by comparing it with alternative models with regard to dynamic compression and with the PSQM/P.861 measure. The performance of the proposed model was on average comparable to that of the PSQM model.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is temporarily free for AES members.