Interframe LSF quantization for noisy channels

In linear predictive speech coding algorithms, transmission of linear predi ctive coding (LPC) parameters-often transformed to the line spectrum freque ncies (LSF) representation-consumes' a large part of the total bit rate of the coder. Typically, the LSF parameters are highly correlated from one fra me to the next, and a considerable reduction in bit rate can be achieved by exploiting this interframe correlation. However, interframe coding leads t o error propagation if the channel is noisy, which possibly cancels the ach ievable gain. In this paper, several algorithms for exploiting interframe c orrelation of LSF parameters are compared. Especially, performance for tran smission over noisy channels is examined, and methods to improve noisy chan nel performance are proposed, By combining an interframe quantizer and a me moryless "safety-net" quantizer, we demonstrate that the advantages of both quantization strategies can be utilized, and the performance for both nois eless and noisy channels improves. The results indicate that the best inter frame method performs as good as a memoryless quantizing scheme, with 4 bit s less per frame. Subjective listening tests have been employed that verify the results from the objective measurements.