Quantization theory and EC-CELP advantages at low bit rates

Abstract

The goal of this work is to analyze the advantages of the recently introduced entropy-constrained code-excited linear predictive (EC-CELP) quantization. The analysis is at low rates in comparison with other EC quantization schemes. Based on N-th order rate-distortion function (RDF), EC quantization theory, and empirical methods, RDF memory gain and empirical space-filling gain (dimensionality N) at low bit rates are defined and calculated. These gains categorize and help us analyze and compare the available coding gains for various EC coders for a given rate and delay (N).