基于短时统计分析的清浊音判别  

Voiced/Unvoiced Decision Based on Statistical Analysis

作　　者：程晓磊[1] 姜占才[1] 

机构地区：[1]青海师范大学计算机学院,西宁810008

出　　处：《电脑开发与应用》2014年第6期75-76,79,共3页Computer Development & Applications

摘　　要：在语音信号当中,清音和浊音的归属虽然没有明确的界限,但是在其短时能量的整体分布区间上存在很大差异。基于这种分布差异,提出了一种基于语音短时特性统计分析的方法进行清/浊音的判决。经过仿真实验证明,该方法具有较高的识别率。In the speech signal, although there is no clear boundary between the unvoiced and voiced , but there is a great difference in the overall distribution of short-time energy. So, the article presents a statis-tical analysis method based on short-speech characteristics about Voiced/Unvoiced Decision. The simulation experiments show that this method has high recognition rate.

关 键 词：清浊音判决 统计分析 短时能量 短时特性 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] TP391.9[自动化与计算机技术—计算机科学与技术]