摘要
将语音信号分为有音与无音段是语音信号处理的基础,但是常用的分段方法已无法将强噪声语音信号进行有效的分段,这对强噪声语音信号的进一步处理带来了困难,在基于输入输出信号的前提下,提出了利用短时互相关技术,通过合理的选择可变短时互相关的窗长度,将强噪声语音信号进行了分段处理。仿真实验结果表明此方法可行有效。
It is the most important progress to classify the speech into voiced and unvoiced regions in many speech progressing application. Several classification algorithm, such as zero-crossing rate or short-time energy, have been reported to classify the speech signal. But due to the strong noise signal existing, these algorithms can not classify the speech signal correctly. Based on the in-out signal, the cross correlation theory was used to analysis the speech with the strong noise through choosing the length of the window. Simulation results show that this method is very effectively.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2006年第z2期1381-1383,共3页
Chinese Journal of Scientific Instrument
关键词
强噪声语音信号
可变窗
短时互相关
speech signal with strong noise variational window short-time cross correlation