首页 | 本学科首页   官方微博 | 高级检索  
     检索      

分形理论在语音信号端点检测及增强中的应用
引用本文:费珍福,王树勋,何凯.分形理论在语音信号端点检测及增强中的应用[J].吉林大学学报(信息科学版),2005,23(2):139-142.
作者姓名:费珍福  王树勋  何凯
作者单位:吉林大学,通信工程学院,长春,130025
摘    要:为了提高现有谱相减方法处理低信噪比语音信号的效果,利用时间序列信号的短时分形维数对低信噪比语音信号的端点检测方法进行了研究,提出了一种基于端点检测的谱相减语音增强方法,给出了其原理及具体算法.仿真实验结果表明,该方法简单可靠,有效可行,当语音信号信噪比降到-5 dB时仍然有效;与已有方法相比,具有更广泛的实用性.

关 键 词:分形维  语音端点检测  语音增强  谱相减  分形理论  语音信号  端点检测  增强方法  应用  Speech  Enhancement  Endpoint  Detection  Speech  Signal  Fractal  Theory  比降  结果  仿真实验  具体算法  原理  研究  检测方法  短时分形维数  时间序列信号  利用  效果
文章编号:1671-5896(2005)01-0139-04
修稿时间:2004年3月11日

Application of Fractal Theory in Speech Signal Endpoint Detection and Speech Enhancement
FEI Zhen-fu,WANG Shu-xun,HE Kai.Application of Fractal Theory in Speech Signal Endpoint Detection and Speech Enhancement[J].Journal of Jilin University:Information Sci Ed,2005,23(2):139-142.
Authors:FEI Zhen-fu  WANG Shu-xun  HE Kai
Abstract:To improve existing spectral subtraction methods on low SNR(Signal-to-Noise Ratio) speech signal processing, a kind of detection of noise-speech endpoint on low SNR speech signal through application of time sequence short time fraction dimension is proposed.A spectral subtraction speech enhancement method based on endpoint detection is put forward.Principle and detail method are given.By simulation to self-recorded speech signal,it turns out that the method is simple and workable.Compared with other methods,when speech signal SNR is reduced to -5 dB, the method is still workable.So the method has wider applicability.
Keywords:fraction dimension  speech endpoints detect  speech enhancement  spectral subtraction
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号