Measurement and visualization of speech quality with VIPER
Harald Mundt, Peter Daniel · The Journal of the Acoustical Society of America · 2000
In this presentation the new VIPER speech quality measurement (VSQM) will be presented. Audible differences between two versions of a sound can be visually inspected in VIPER through the help of auditory distance spectrograms. The objective instrumental measures provided by VIPER will be validated with the results of listening experiments on speech quality (MOS) and compared with the performance of other speech quality measures (PAMS, TOSQA, MNB, PSQM+, PSQM). The comparison is based on the ITU-T Series P. supplement 23 database. The database has been developed in order to characterize the performance of the ITU-T speech codec G.729. Therefore, a wide range of different speech codes, in noisy and clean conditions, in various tandem conditions, with bit and frame errors of various intensity, was tested for different languages (English, French, Japanese). In order to quantify the assessment performances, the linear correlation coefficient and the standard deviation based on the mapped objective data and the subjective data are used.