摘要
针对激光侦测系统中传统去噪方法难以消除脉冲噪声,提出一种新的抑制方法。该方法通过分析脉冲噪声的产生原因将其分类,对由设备原因引起的脉冲噪声利用差分值确定脉冲噪声的位置,对由环境因素引起的脉冲噪声利用时频谱图检测能量值来判别语音是否被脉冲噪声所污染。对于被脉冲噪声污染的数据,利用线性预测分析方法估计得到正确的语音数据。仿真信号和实际信号实验表明,算法能显著地改善信噪比和抑制脉冲噪声。
This paper provides a newspeech enhancement method for the removal of impulsive noise in laser detector system. First, the method analyzes the causes of impulsive noise and classifies- them. Then the method detects the speech frame polluted by impulsive noise with ditterences be- tween adjacent elements of speech and energy detection. Furthermore, the linearprediction is used in estimating correct speech frame for impulsive noise removal. In some speechenhancement tests with simulation signal and measured speech signal, the results show that the methodachieves an im- proved Signal-to-NoiseRatio （SNR） andremoves impulsive noise even when the SNR of noisy speech is low.
作者
张文璐
彭华
ZHANG Wenlu;PENG Hua(Information Engineering University,Zhengzhou 450001,China)
出处
《信息工程大学学报》
2017年第6期652-656,共5页
Journal of Information Engineering University
基金
国家自然科学基金资助项目(61401511)
关键词
脉冲噪声
语音增强
能量检测
线性预测分析
impulsive noise
speech enhancement
energy detection
linear prediction
