The evaluation of the speech quality is based on a multidimensional approach which consists in selecting some perceptive attributes according to degradations caused by the type of telecommunication. The paper presents a model assessing speech quality as a combination of three perceptive features: noisiness, coloration and continuity. Each one of these dimensions is estimated without reference by hybrid indicators which are a combination of parametric (codec, network statistics) and signal based indicators, depending on the type of information available at the measuring point.
展开▼
