复杂噪声下自行高炮通信系统语音信号检测方法研究  

Voice Activity Detection of A Certain Self Propelled Anti-Aircraft Gun Communication System in the Environment with Complicated Noise

作　　者：程远增[1,2] 王渝[1] 马志强[2] 高庆[2] 

机构地区：[1]北京理工大学自动化学院,北京100081 [2]军械工程学院,河北石家庄050003

出　　处：《计算机测量与控制》2010年第12期2783-2785,共3页Computer Measurement &Control

摘　　要：研究了在噪声环境下自行高炮通信系统语音信号的检测方法,提出了在高信噪比环境下采用短时平均幅度(AM)算法与短时平均幅度差(AMDF)算法检测相结合的二级判断法,在低信噪比的情况下采用短时高低频能量比的算法,实现语音信号在不同噪声环境下的检测;实验结果表明,高信噪比情况下二级判别法具有良好的语音识别效果,低信噪比情况下短时高低频能量比算法可以有效地检测出语音端点,避免切音现象;与传统语音检测算法相比,此方法不同信噪比状态下都具有较好的检测效果,体现出算法一定的优越性。This paper researches voice activity detection of a certain self propelled anti-Aircraft gun communication system in the environment with complicated noise.It employs the two level detection between short-time average magnitude and short-time average magnitude difference function in the high SNR（Signal-Noise Ratio） environment and uses the ratio of the average energy in the low-bands to that in the high-bands in the low SNR environment.Experiments show that the two level detection has perfect capacity of voice discernment in the high SNR environment,the ratio of the average energy in the low-bands to that in the high-bands can detect the voice correctly and avoid missing voice in the low SNR environment.The proposed algorithm outperforms traditional algorithm for voice activity detection.

关 键 词：语音端点检测 短时平均幅度 短时平均幅度差 短时能量比 自行高炮通信系统 

分 类 号：TN912[电子电信—通信与信息系统]