一种新的鉴别汉语语音信号起止点的有效算法  

A New Valid Algorithm to Distinguish the Start-Stop Points of Chinese Speech Signal

作　　者：寻纯阳 杨红[1] 

机构地区：[1]湖南大学电气工程系

出　　处：《湖南大学学报（自然科学版）》1996年第2期95-99,共5页Journal of Hunan University:Natural Sciences

摘　　要：提出短时能频值（ＥＦＶ）的新概念，设计了用短时能频值的参量来估计语音起止点的新的有效算法，提高了语音信号与背景噪声的分辨力．运用基于短时能频值ＥＦＶ估计语音起止点的算法，研究汉语普通话音素的区别特征，上机计算分析证明了ＥＦＶ门限能够很好地从噪声中分辨出语音信号，有着较好的稳定性和较高的语音－噪声分辨力．A new concept of short time Energy-Frequency-Value(EFV) is proposed, and a new valid algorithm using short time EFV parameter is designed to evaluate the start-stop points of chinese speech signal. Resolution power is improved between speech signal and background noise. Based on the algorithm, we carry out the study of the common spoken chinese phoneme' s distinctive features. The computer calculation and analysis confirm that EFV threshold can distinguish speech signal from noise,achieve good stability and speech signal-noise resolution

关 键 词：语音 噪声 分辨力 元音识别 汉语 语音起止点 

分 类 号：TN912.34[电子电信—通信与信息系统]