| 一种基于后验信噪比频域语音增强方法的研究 |
| |
| 引用本文： | 陈紫强. 一种基于后验信噪比频域语音增强方法的研究[J]. 电声技术, 2007, 31(5): 56-59 |
| |
| 作者姓名： | 陈紫强 |
| |
| 作者单位： | 桂林电子工业学院,通信与信息工程系,广西,桂林,541004 |
| |
| 基金项目： | 国家自然科学基金项目(60272038)，广西自然基金项目(D20428) |
| |
| 摘 要： | 谱减法是常用的单通道语音降噪方法,传统谱减法在抑制背景噪声的同时引入了“音乐噪声”,影响听觉效果。为了抑制音乐噪声,提出了一种基于后验信噪比的频域语音增强新方法,当后验信噪比较高时,采用基于后验信噪比的谱减法增强语音信号;当后验信噪比较低时,采用基于后验信噪比的谱衰减方法对含噪语音信号谱线进行衰减,达到语音增强的目的。仿真结果表明,基于后验信噪比的频域语音增强法具有较好的背景噪声和音乐噪声抑制效果,并保持了较好语音可懂度。
|
| 关 键 词： | 后验信噪比 语音增强 谱衰减 |
| 文章编号： | 1002-8684（2007）05-0056-04 |
| 修稿时间： | 2007-01-08 |
Study on a New Spectral Domain Algorithm for Speech Enhancement Based on a Posteriori SNR |
| |
| CHEN Zi-qiang. Study on a New Spectral Domain Algorithm for Speech Enhancement Based on a Posteriori SNR[J]. Audio Engineering, 2007, 31(5): 56-59 |
| |
| Authors: | CHEN Zi-qiang |
| |
| Affiliation: | Dept. of Communication and Information Engineering, Guilin institute of Electronic Technology, Guilin Guangxi 541004, China |
| |
| Abstract: | Spectral subtraction algorithms are usually used in single-channel speech enhancement.The traditional spectral subtraction algorithms which can suppress the background noises will introduce the "music noise".The existence of the music noise strongly affects the hearing effect.To restrain the music noise,a new method based on a posteriori SNR is proposed.When the posteriori SNR is high,the spectral subtraction algorithm based on posteriori SNR is used to enhance the speech signal,otherwise the spectral attenuation algorithm is used.Simulating results show that the new method is excellent in the background noise and music noise cancellation while keeping little distortion to speech signals. |
| |
| Keywords: | posteriori SNR speech enhancement spectral attenuation |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
