Channel codes that exploit the residual redundancy in CELP-encoded speech

We consider the problem of reliably transmitting CELP-encoded speech over noisy communication channels. Our objective is to design efficient coding/decoding schemes for the transmission of the CELP line spectral parameters (LSPs) over very noisy channels. We begin by quantifying the amount of "residual redundancy" inherent in the LSPs of Federal Standard 1016 CELP. This is done by modeling the LSPs as first- and second-order Markov chains. Two models for LSP generation are proposed; the first model characterizes the intraframe correlation exhibited by the LSPs, while the second model captures both intraframe and interframe correlation. By comparing the entropy rates of the models thus constructed with the CELP rates, it is shown that as many as one-third of the LSP bits in every frame of speech are redundant. We next consider methods by which this residual redundancy can be exploited by an appropriately designed channel decoder. Before transmission, the LSPs are encoded with a forward error control (FEC) code; we consider both block (Reed-Solomon) codes and convolutional codes. Soft-decision decoders that exploit the residual redundancy in the LSPs are implemented assuming additive white Gaussian noise (AWGN) and independent Rayleigh fading environments. Simulation results employing binary phase-shift keying (BPSK) indicate coding gains of 2-5 dB over soft-decision decoders that do not exploit the residual redundancy.