采用分块耦合广义分形维的语音端点检测技术研究  

A Researchin Speech Endpoint Detection Based on Boxes-coupling Generalization Dimension

作　　者：汪子梅[1] 杨翠容[1] 武薇[1] 范影乐[1] 

机构地区：[1]杭州电子科技大学生物医学工程与仪器研究所,杭州310018

出　　处：《生物医学工程学杂志》2008年第3期536-541,共6页Journal of Biomedical Engineering

基　　金：国家自然科学基金资助项目(60302027);浙江省教育厅科研计划项目资助(20030620)

摘　　要：为了克服传统广义分形维算法存在的边缘效应,提高低信噪比环境中语音信号起止点的检测能力,我们提出了一种分块耦合(Boxes-coupling)的广义维数计算方法,并将其应用于语音信号的端点检测中。首先确定各分块之间的耦合边长,利用耦合的分块覆盖待检测的语音信号,求得容量维数、信息维数和关联维数,以构成三维特征向量;然后根据特征距离和相似度之间的关系,用一般距离度量进行特征提取;最后采用双门限模式分类进行决策分类。实验结果表明,该方法适合多种噪声环境下的语音端点检测。与传统广义维数(Original generalized di-mension,OGD)以及谱熵(Spectral entropy,SE)算法相比,在低信噪比情形下,具有检测精度高、抗干扰能力强以及强系统稳定性(Robustness)等优点。In this paper,a new calculating method of generalized dimension, based on boxes-coupling principle, is proposed to overcome the edge effects and to improve the capability of the speech endpoint detection which is based on the original calculating method of generalized dimension. This new method has been applied to speech endpoint detection. Firstly, the length of overlapping border was determined, and through calculating the generalized dimension by covering the speech signal with overlapped boxes, three-di...

关 键 词：耦合 广义维数 端点检测 边缘效应 

分 类 号：TN912.3[电子电信—通信与信息系统]