ITU-T P.833.2 PDF

The extension of the methodology described in this Recommendation is intended to derive the fullband equipment impairment factors Ie,FB, quantifying the impairment introduced by fullband (20-20,000 Hz) speech codecs, including or not the effects of transmission errors, such as random bit errors, random packet loss or discard, or bursty packet loss or discard. Such fullband equipment impairment factors have been introduced as a simplified measure of the degradation introduced by fullband speech codecs on the integral transmission quality from mouth to ear. They are in no way an exact description of the effects related to each individual codec or codec tandem, which may be very diverse in their perceptual nature. Instead, they represent the relative degradation in comparison to other impairments occurring in a connection.

In order to provide guidance on the quantitative amount of impairment introduced by such codecs, a framework of fullband equipment impairment factor values for several fullband speech codecs has been derived by ITU-T, see Appendix V of [ITU-T G.113]. The derivation of these values was based on subjective listening-only tests carried out in different test laboratories, so as to guarantee stable values for all the codecs under investigation so far, as well as their relations to each other. If new equipment impairment factor values for different codecs have to be derived, then the overall consistency with the established framework is of primary importance if results are to be obtained that are valid for network planning. The methodology described in this Recommendation was designed to fulfil this requirement.

So far, the fullband E-model defined in [ITU-T G.107.2] estimates fullband conversational speech quality considering the effects of background noise, pure delay, as well as random and bursty packet loss. The values of the fullband equipment impairment factor Ie,FB are intended to be used to estimate the Ie,FB,eff factor characterizing the effective impairment of the respective codec, considering the loss. Whether these formulae hold for all types of codecs, and all pattern loss probabilities and distributions, is an item for further study.

For asynchronous tandems of multiple codecs of the same type, or of multiple codecs of different types, it is assumed that individual equipment impairment factors are additive. The overall equipment impairment factor for the chain of codecs is then calculated as a simple sum of all the individual Ie,FBs. Experimental test data collected provide evidence that this simple additivity is not satisfied for all the potential combinations of codecs; these deviations from the pure additivity property are an item for further study.