Evaluation of the G.729 speech coder with pathological voice samples

The performance of a speech coder is typically evaluated with a battery of objective and subjective tests. Objective measures of speech quality attempt to measure the "distance" between the original and coder-processed speech, while subjective testing involves obtaining mean opinion score (MOS) for the coder-processed speech. In general these tests are performed with speech samples collected from normal talkers, and no tests are conducted to determine how a speech coder interacts with abnormal speech samples. This paper focuses on the performance of the ITU standardized G.729 speech coder with speech and voice samples from disordered talkers. Standardized objective measures of speech coder quality and acoustic measures of voice are used to show that there is a significant drop in the performance of the G.729 speech coder when abnormal speech samples were used.

[1]  Li Deng,et al.  Interaction of speech disorders with speech coders: effects on speech intelligibility , 1996, Proceeding of Fourth International Conference on Spoken Language Processing. ICSLP '96.

[2]  Ian S. Burnett,et al.  On the effects of accent and language on low rate speech coders , 1996, Proceeding of Fourth International Conference on Spoken Language Processing. ICSLP '96.

[3]  J.E. Natvig,et al.  Speech processing in the pan-European digital mobile radio system (GSM)-system overview , 1989, IEEE Global Telecommunications Conference, 1989, and Exhibition. 'Communications Technology for the 1990s and Beyond.

[4]  D. Jamieson,et al.  Identification of pathological voices using glottal noise measures. , 2000, Journal of speech, language, and hearing research : JSLHR.

[5]  Stephen D. Voran,et al.  Objective estimation of perceived speech quality. I. Development of the measuring normalizing block technique , 1999, IEEE Trans. Speech Audio Process..

[6]  S. Voran.  Advances in objective estimation of perceived speech quality , 1999, 1999 IEEE Workshop on Speech Coding Proceedings. Model, Coders, and Error Criteria (Cat. No.99EX351).