首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于谐振结构的线性预测编码
引用本文:胡剑凌,徐盛.基于谐振结构的线性预测编码[J].上海交通大学学报,2002,36(12):1770-1773.
作者姓名:胡剑凌  徐盛
作者单位:上海交通大学,电子工程系,上海,200030
摘    要:语音编码中一般用线性预测编码(LPC)建立信号的声道响应,如何提高LPC的预测增益并用尽可能少的比特数量化LPC参数将直接影响编码器的性能.文中提出了一种基于信号谐振结构的LPC技术(HLPC),利用HLPC可使LPC谱与信号谐振峰具有最佳的拟合,同时可获得0.36dB的额外预测增益.设计了LPC参数的线性预测多层分裂码本矢量量化(LPMSVQ)方法.数值实验表明,利用LPMSVQ可在21bit/帧下获得透明的量化质量.

关 键 词:语音编码  线性预测编码  谐振结构  矢量量化
文章编号:1006-2467(2002)12-1770-04
修稿时间:2001年9月16日

Harmonic Structure Based Linear Prediction Coding
HU Jian-ling,XU Sheng.Harmonic Structure Based Linear Prediction Coding[J].Journal of Shanghai Jiaotong University,2002,36(12):1770-1773.
Authors:HU Jian-ling  XU Sheng
Abstract:The response of vocal tract is mostly modeled by linear prediction (LP) in speech coding.So how to improve the accuracy of analysis and quantize the model coefficients will affect the performance of the codec directly. The harmonic structure based the linear prediction coding (HLPC) was proposed. By using HLPC, the LPC spectrum can match the harmonic amplitudes more accurately and the extra prediction gain of 0.36 dB can be obtained. Also presented is a method named linear prediction multi-split vector qunatization (LPMSVQ) which is designed to quantize the LPC coefficients and the transparent quality can be obtained at 21 bit/frame.
Keywords:speech coding  linear prediction coding (LPC)  harmonic structure  vector quantization (VQ)
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号