摘要
主要研究用于分布式语音识别(DSR)的语音参数的提取方法以及参数性能分析。以前所用到的语音参数大部分是LPC倒谱参数,但其抗噪声性能较差。文中主要讨论了MEL倒谱参数,并在移动通信环境下,比较了两者的性能。
This paper focuses on the extraction and performance analysis of speech parameters for Distributed Speech Recognition(DSR).Due to the poor performances of LPC cepstrum coefficients(LPCC) in noise circumstances,we discuss Mel cepstrum parameters in this paper and also compare the performance of these two kinds of parameters in mobile communication.
出处
《南京邮电学院学报(自然科学版)》
2003年第2期31-34,共4页
Journal of Nanjing University of Posts and Telecommunications
关键词
分布式语音识别
DSR
Q-CELP语音编码
Mel倒谱参数
LPC倒谱参数
Distributed speech recognition
Qualcomm-code excited linear prediction speech code
Mel cepstrum
Linear prediction coding cepstrum