背景噪声下的端点检测算法的研究  

Endpoint Detection Methods Based on Background Noise

作　　者：周娜[1] 赵振东[1] 张莹[2] 

机构地区：[1]华北电力大学电子与通信工程系,河北保定071003 [2]东北财经大学津桥商学院,辽宁大连116600

出　　处：《通信技术》2008年第7期239-241,共3页Communications Technology

摘　　要：噪声环境下的端点检测在语音信号分析和识别中占有重要地位。文中将分形理论中的分形记盒维数应用到端点检测算法中,采用了基于分形记盒维数与短时能零比相结合的端点检测算法,以分形记盒维数为主要判决条件,并在判决门限的设定上采用了自适应机制。仿真结果表明,基于分形记盒维数和短时能零比相结合的端点检测算法在白噪声和粉红色噪声下具有很好的准确率和鲁棒性。Endpoint detection in noisy environment plays an important role in speech analysis and recognition.In this paper, the box-counting dimension in the fractal theory is used in the endpoint detection. The classical threshold method is improved, and the endpoint detection algorithm based on the box-counting dimensions and short-time energy-zero-ratio is employed, The box-counting dimension is taken as the main decision condition,and the se]f-adaptive mechanism is used to set the decision threshold. The simulation results show that this algorithm is of high accuracy and good robustness in the white noise and pink noise environment. Uowever, it is not very satisfying in the volvo and babble noise environment,

关 键 词：端点检测 能零比 分形记盒维数 自适应机制 

分 类 号：TN912.3[电子电信—通信与信息系统]