Derivation of Equipment Impairment Factors from ITU-T P.863 for Fullband Speech Codecs
* Presenting author
Abstract:
Recently, the fullband version of the E-Model, a parametric tool for speech communication service planning, has been standardized by the International Telecommunication Union ITU-T. In order to use the E-model for a given codec, the equipment impairment factor Ie has to be defined for the speech codec under study. This procedure standardized in ITU-T Recommendation P.833 describes a methodology to derive Ie by conducting listening-only tests with that codec, together with other codecs for which the Ie is known. In ITU-T Rec. P.834, a similar methodology is described to derive the equipment impairment factor based on predictions of instrumental models. With the extension of the E-model to fullband communication scenarios, this methodology can be extended for fullband speech codecs.In this paper, we present our first efforts for this extension. We use the ITU-T P.863 instrumental model in fullband mode to estimate the quality of different fullband, wideband, and narrowband coded speech files to obtain instrumental model-based equipment impairment factors Ie,FB. We then apply a linear interpolation between the expected Ie,FB values, and the instrumental model-based Ie,FB predictions and use them to predict the Ie,FB for the Opus codec in fullband mode, for which currently no Ie,FB values are available.