Parametric speech coding-HVXC at 2.0-4.0 kbps

Abstract

MPEG-4 parametric speech coding, harmonic vector excitation coding (HVXC) algorithm, is described. New features of the coder includes a quantizer scheme capable of generating 2.0 and 4.0 kbps scalable bit-streams, where 2.0 kbps decoding is possible using a subset of 4.0 kbps bit-stream. Time scale modification of speech is also possible without changing pitch nor phoneme for fast and slow playback mode. Listening tests show that the proposed coding method at 2.0 kbps provides significantly better quality than that of FS1016 CELP at 4.8 kbps. In October 1998, the HVXC coder was adopted to the Final Draft International Standard (FDIS) of MPEG-4 standardization.