Efficient methods for evaluation and prediction of subjective speech quality in GSM mobile networks

Abstract

Describes efficient methods, which are able to predict the expected speech quality by deriving a measurement from the GSM standard measurement RXQUAL (coded bit error rate) and transform this measurement into a subjective speech quality score. The evaluation of subjective speech quality-which is the basis for scoring by transformation on a subjective speech quality scale-is obtained by scientifically carried out speech investigations on a wide variety of possible speech samples in the German D1 network. The theoretically-determined transformations were confirmed by extensive measurements, which consider many parameters, like different fading profiles or length of the speech material. Furthermore this work gives hints to the visualisation of the predicted speech quality.<>