A PERFORMANCE STUDY OF SPEECH CODERS FOR TELEPHONE CONFERENCING IN DIGITAL MOBILE COMMUNICATION NETWORKS

  • Lee, M.S. (Department of Information and Communication Engineering, Korea Advanced Institute of Science and Technology) ;
  • Lee, G.C. (Department of Information and Communication Engineering, Korea Advanced Institute of Science and Technology) ;
  • Kim, K.C. (Department of Information and Communication Engineering, Korea Advanced Institute of Science and Technology) ;
  • Lee, H.S. (Department of Information and Communication Engineering, Korea Advanced Institute of Science and Technology) ;
  • Lyu, D.S. (Department of Information and Communication Engineering, Korea Advanced Institute of Science and Technology) ;
  • Shin, D.J. (Department of Information and Communication Engineering, Korea Advanced Institute of Science and Technology) ;
  • Lee, Hun (Department of Information and Communication Engineering, Korea Advanced Institute of Science and Technology)
  • Published : 1994.06.01

Abstract

This paper describes two methods to assess the output speech, quality of vocoders for telephone conferencing in digital mobile communication networks. The proposed methods are the sentence discrimiantion method and the modified degraded mean opinion score (MDMOS) test. We apply these two methods to Qualcomm code excited linear prediction (QCELP), vector sum excited linear prediction (VSELP) and regular pulse excited-long term predictin (RPE-LTD) voceders to evaluate which vocoding algorithm can process mixed voice signal from two speakers better for telephone conferencing. From the experiments we obtain that the VSELP vocoding algorithm reveals superior output speech quality to the other two.

Keywords