Speech Quality Measure in CDMA Networks Using PLP-CMS

Abstract

For set-up, management and repair of a mobile communication system, a continuous estimation of speech quality is required. Speech quality measurement can be conducted by listener's judgement in a subjective test such as MOS test. Since this method is, however, laborious, expensive and time-consuming, it is convenient to predict subjective speech quality via objective measure. There are several conventional objective speech quality measures such as BSD, PLPCD, PSQM and MNB. These measures are good at predicting the quality of speech, which is coded and decoded by vocoder. However, PLP-CD and PSQM show less correlation with subjective quality for the speech samples of different channel paths in a real mobile communication system because they are not prepared for channel effects such as spectral distortion. Therefore, this paper proposes a robust objective speech quality measure, PLP-CMS, which can predict subjective speech qualities in mobile communication systems. PLP-CMS has a high correlation with subjective qualities by using PLP analysis and shows a robust performance not being influenced by channel effects from PSTN by utilizing CMS. To verify the performance of the proposed algorithm, we carried out subjective and objective quality estimation on speech samples that are variously distorted in a real mobile communication system. From the results, PLP-CMS showed higher correlation with subjective quality than the conventional measures such as PLP-CD and PSQM.