M-LCELP speech coding at 4 kbps
K. Ozawa, Masahiro Serizawa, Toshiki Miyano, T. Nomura · 2002
This paper presents the M-LCELP (multi-mode learned code excited LPC) speech coder, which has been developed for the North American half-rate digital cellular systems. M-LCELP develops the following techniques to achieve high-quality synthetic speech at 4 kbps: (1) Multimode and multi-codebook coding, (2) Pitch lag differential coding with pitch tracking, (3) A two-stage joint design regular-pulse codebook with common phase structure in voiced frames, (4) An efficient vector quantization for LSP parameters, (5) An adaptive MA type comb filter to suppress excitation signal inter-harmonic noise. The MOS subjective test shows that 4.075 kbps M-LCELP synthetic speech quality is high, and that its quality is mostly equivalent to that for an 8 kbps North American full-rate VSELP coder.>