| 摘要: |
| 为了解决低速率语音编码中比特受限的问题，提出了一种基于二阶隐马尔可夫模型
的清浊音参数恢复算法。该算法采用二阶隐马尔可夫模型，通过归一化的能量参数和LPC倒
谱系数估计出序列中的全带清浊音判决和各个子带的清浊音度。解码器实现该算法后，编码
器就无需对清浊音参数进行量化传输，从而节约了比特数。实验结果表明，该算法比基于GM
M模型的算法能更好地恢复出清浊音信息，全带清浊音误判率减少了5%～20%，合成语音的MO
S分比用5 bit的矢量量化（VQ）算法提高了003左右，达到了在节约比特数的同时也提高
了语音质量的效果。 |
| 关键词: 低速率语音编码 二阶隐马尔可夫模型 全带V/U判决 BPVC恢
复 |
| DOI： |
|
| 基金项目:国家自然科学基金资助项目（60572081） |
|
| Voiced/Unvoiced Parameters Recovery Based on Second-Order Hidden Markov Model |
| HE Hong-hua,XU Jing-de,JI Zhe,CUI Hui-juan,TANG Kun |
| (Tsinghua National Laboratory for Information Science and Technology,Department of Electronic Engineering,Tsinghua University, Beijing 100084, China) |
| Abstract: |
| In order to solve the problem of limited number of bits in low bit rat
e speech coding, an algorithm using second-order Hidden Markov Model(HMM2) to r
ecover the voiced/unvoiced parameters is proposed in this paper. The algor
ithm uses the normalized energy and linear prediction coding(LPC) coefficients
to estimate the fullband V/U classification and the subband BPVC value. The
al
gorithm can be implemented in the decoder, saving the bits originally used by V/
U parameters and reducing the bit rate of speech coding. Experimental results sh
ow that the algorithm proposed can reduce the V/U classification error rate by 5
%～20% compared with the GMM algorithm, and improve the mean opinion score(MOS)
o
f the synthesized speech signal by about 0.03 compared with the 5bit vector quan
ti
zation（VQ）, thereby greatly improves the estimation performance of the V/U par
ameters. |
| Key words: low bit rate speech coding second order HMM V/U classification BPVC recover
y |
《电讯技术》编辑部 