摘要
本文应用时间序列信号的短时分形维维数的非线性理论，对低信噪比的语音信号端点检测和自适应滤波进行了研究。针对信噪比在０ｄＢ和－５ｄＢ的噪声语音信号的端点检测和滤波进行了计算机仿真实验。实验结果表明：短时分形维维数能较准确地进行语音端点的检测，并且。
This paper introduces the detection of the speech start point and end points and filtering on low SNR speech signal through application of time series short time fractal dimension. The simulation results by means of computer shows the short time fractal dimension of signals can be used to accurately detect the points of speech on the 0 dB and -5 dB noise speech signals. The adapting filter based on the short time fractal dimension of signals can filter the noise of low SNR speech signals.
出处
《浙江师大学报（自然科学版）》
1999年第1期16-21,共6页
Journal of Zhejiang Normal University(Natoral Sciences)
关键词
语音处理
分形维
低信噪比
语音端点检测
滤波
speech processing
fractal dimension
low SNR
speech point detect
adaptive filtering
