Quantizer Design in LSP Speech Analysis-Synthesis.

Abstract

The LSP speech analysis-synthesis method is known as one the most efficient vocoders. An important issue in encoding of the LSP parameters is that a certain ordering relationship between the LSP parameters is required to insure the stability of the synthesis filter. This requirement has an important impact on the design of quantizers for the LSP parameters. In this paper, the performance of several algorithms for the quantization of the LSP parameters is studied. A new adaptive method which utilizes the ordering property of the LSP parameters is presented. A combination of this adaptive algorithm with non-uniform step size quantization is shown to be a very effective method for encoding the LSP parameters. The performance of the different quantization schemes is studied on a long sequence of speech samples. For the spectral distortion measure, appropriate performance comparisons between the different quantization schemes are rendered.