| 基于谱减法和均匀子带频带方差法的端点检测研究 |
| |
| 引用本文： | 王威,胡桂明,杨丽,黄东芳,周杨.基于谱减法和均匀子带频带方差法的端点检测研究[J].电声技术,2016,40(5). |
| |
| 作者姓名： | 王威 胡桂明 杨丽 黄东芳 周杨 |
| |
| 作者单位： | 广西大学电气工程学院,广西南宁,530004 |
| |
| 摘 要： | 传统的端点检测算法是基于短时能量和短时过零率的双门限检测法,该算法对于含噪语音信号检测效果不佳.因此,将谱减法与均匀子带频带方差算法相结合,结合算法分两个过程,首先用谱减法对带噪语音信号进行降噪,然后用均匀子带频带方差法进行端点检测.实验结果表明,结合算法对于含噪较多的语音信号检测效果良好.
|
| 关 键 词： | 端点检测 双门限检测 谱减法 频带方差 |
| 收稿时间： | 2015/5/8 0:00:00 |
| 修稿时间： | 2015/5/8 0:00:00 |
Research of Endpoint Detection Based on Spectral Subtraction and Uniform Sub-band Spectrum Variance |
| |
| WANG Wei,HU Gui-ming,YANG Li,HUANG Dong-fang and ZHOU Yang.Research of Endpoint Detection Based on Spectral Subtraction and Uniform Sub-band Spectrum Variance[J].Audio Engineering,2016,40(5). |
| |
| Authors: | WANG Wei HU Gui-ming YANG Li HUANG Dong-fang and ZHOU Yang |
| |
| Affiliation: | College of Electrical Engineering,Guangxi University,College of Electrical Engineering,Guangxi University,College of Electrical Engineering,Guangxi University,College of Electrical Engineering,Guangxi University,College of Electrical Engineering,Guangxi University |
| |
| Abstract: | The traditional endpoint detection algorithm is the double threshold detection algorithm based on short time energy and short time zero crossing rate. The noisy speech signal detection effect used the algorithm is not good. Considering the above problems, this paper combines spectral subtraction with uniform sub-band spectrum variance algorithm. The combination algorithm is divided into two steps. Firstly, spectral subtraction is used to reduce the noise of noisy speech signal. Then uniform sub-band spectrum variance is used to endpoint detection. The experimental results show that the more noisy speech signal detection effect used the combination algorithm is good. |
| |
| Keywords: | endpoint detection double threshold detection spectral subtraction spectrum variance |
| 本文献已被 万方数据 等数据库收录！ |
| 点击此处可从《电声技术》浏览原始摘要信息 |
|
点击此处可从《电声技术》下载全文 |
|
