The analysis of speech codecs using psychoacoustic measures

Authors

M. Raad, University of WollongongFollow
C. H. Ritz, University of WollongongFollow
I. Burnett, University of WollongongFollow
Alfred Mertins, University of WollongongFollow

RIS ID

7071

Publication Details

This paper originally appeared as: Raad, M, Ritz, CH, Burnett, I and Mertins, A, The analysis of speech codecs using psychoacoustic measures, IEEE Workshop Proceedings. Speech Coding, 6-9 October 2002, 108-110. Copyright IEEE 2002.

Abstract

This paper analyses two narrowband speech codecs, the 4.8 kbit/s FS1016 coder and the 8 kbit/s G729 coder, using objective psychoacoustic measures. Four measures are used: loudness, sharpness, roughness and tonality. The results show sharpness and roughness as the two major contributing factors to the subjective difference between the two coders.