摘要
在无线电侦听领域,侦听数据中常常含有纯噪音、误判为语音的定频数字调制信号等非语音信号段,造成侦听人员工作效率下降。针对该问题,提出一种基于子带能量分析的语音段识别算法。考虑到侦听数据在时域上存在幅度跳变现象,首先利用直方图分析对侦听数据进行分段,得到若干个子数据段,然后对每个子数据段进行子带能量分析,最后提取特征参数实现对语音段的识别。仿真结果表明,该方法能够准确地对侦听数据进行分段,并识别出其中的语音段。
The intercepted data contains pure noise,frequency-fixed with erroneous judgement of speech as well as others non-speech signal sections in the field of radio interception.Hence,it causes a decline in staff's efficiency.Aiming at this problem,a speech recognition algorithm based on the sub-band energy analysis is proposed.In view of the fact that a phenomenon of intercepted data that exists range hop on the time domain.Firstly,the histogram analysis is made to get sub-data segments.Secondly,each sub-data segment is analyzed by means of sub-band energy analysis.Finally,to pick up the characteristic parameter is to achieve recognition of speech signal segments.As the simulation result shows that the method can not only section interception data with much accuracy,but also recognize speech segments.
作者
黄均安
詹毅
HUANG Jun-an;ZHAN Yi(College of Communication Engineering,Hangzhou Dianzi University,Hangzhou 310018,China;No.36 Research Institute,China Electronic Technology Corporation,Jiaxing 314033,China)
出处
《软件导刊》
2019年第1期162-165,共4页
Software Guide
关键词
纯噪音
定频数字调制信号
子带能量
语音段识别
无线电侦听
pure noise
frequency-fixed digital modulation signal
sub-band energy
speech segments recognition
radio interception
