EFFICIENT ENCODING OF SPEECH LSF PARAMETERS USING THE KARHUNEN-LOEVE TRANSFORMATION

Abstract

Line Spectral Frequencies (LSF) provide an alternate parameterization of the analysis and synthesis filters used in Linear Predictive Coding (LPC) of speech. In this paper, a new method of quantization over the LSF vectors using the Karhunen-Loeve transform is presented. The new technique, which operates at about 1.8 kbits/s, has good quality and is not sensitive to quantization noise. When compared with other existing methods, this algorithm is better in terms of robustness and complexity.