基于线谱区域量化技术的低速语音编码

Low-Rate Speech Coding Based on the Field Quantization Method of LSP

本文将线谱对 (L inear Spectrum Pair,L SP)参数用于语音编码 ,同时分析比较了 L SP与常用的线性预测系数 (L PC)和格滤波发射系统 (PARCOR)的特征分别 ,并由此引入更能有效量化线谱对参数的区域量化技术。因此在保证编码后语音 MOS指标的同时 ,可以达到进一步降低了码率的目的。

This paper describes how to apply parameters of Linear Spectrum Pair(LSP) to coding for speech. It compares the features of LSP with those of LPC and PARCOR. And it further introduces the more efficient method of field quantization for LSP parameters, which can lower the coding rate while still achieving the index of MOS.