Optimum source codec design in coded systems and its application for low-bit-rate speech transmission

A generalized algorithm for designing an optimum VQ source codec in systems with channel coding is presented. Based on an AWGN channel model, the algo rithm derives the distribution of the channel decoder soft-output and subst itutes it in the expression for the system end-to-end distortion. The VQ en coder/decoder pair is then optimized by minimizing this end-to-end distorti on. For a Gauss-Markov source, the proposed algorithm outperforms the conve ntional SOVQ source coding scheme by 5.0 dB in the decoded source SNR. Appl ication of this algorithm for designing optimum low-bit-rate speech codec i s given. A 4.0 kbps VQ based CELP codec is designed for performance? evalua tions, where all the CELP parameter encoder/decoder pairs are optimized by minimizing their end-to-end distortions, respectively. As a result, the spe ech distortion over the noisy channel is minimized. Subjective tests show t hat the proposed algorithm improves the decoded speech quality by 2.5 MOS r elative to a regular SOVQ CELP speech coding system. The performances of th e algorithm under channel mismatch conditions are also shown and discussed.