Visual comparison of perceptual degradation indicators in two listening speech quality models

Recently, new objective speech quality evaluation methods which integrate perceptual dimensions have been developed in order to provide diagnostic information about specific aspects of the transmission systems' quality. In this paper, we present two approaches, one (DIAL) built explicitly on blocks corresponding to perceptual dimensions and the other one (POLQA) using quality indicators implicitly linked to these perceptual dimensions. The first goal of our study is to identify in POLQA the correspondence between perceptual dimensions and quality indicators. Such indicators can be either present in the model itself or derived from our own analysis of the model. In a second step, we present first results in comparing the indicators of these two approaches. We found that some indicators are equivalent when simple degradations are applied.