| 基于短时谱分析的语音增强改进算法 |
| |
| 引用本文： | 叶利剑,黄松华,邱小军.基于短时谱分析的语音增强改进算法[J].电声技术,2011,35(9):42-46. |
| |
| 作者姓名： | 叶利剑 黄松华 邱小军 |
| |
| 作者单位： | 1. 瑞声声学科技研发(南京)有限公司,江苏南京,210093
2. 南京大学声学研究所,江苏南京,210093 |
| |
| 摘 要： | 介绍了基本的语音增强短时谱分析算法的原理.基于经典的谱减算法进行改进,引入了先验信噪比估计,以及时间回归平均法对噪声进行更新,以及一系列后处理改进方法,提高了降噪效果.在保持语音可懂度的基础上,增强了噪声衰减量.主、客观实验结果表明,针对汽车噪声一类的平稳噪声,算法效果出色；而对于街道、人声一类非稳态噪声,该算法性能优...
|
| 关 键 词： | 谱减法 先验信噪比 时间回归平均法 |
Improved Speech Enhancement Algorithm Based on Short-Time Spectral Analysis |
| |
| YE Lijian,HUANG Songhua,QIU Xiaojun.Improved Speech Enhancement Algorithm Based on Short-Time Spectral Analysis[J].Audio Engineering,2011,35(9):42-46. |
| |
| Authors: | YE Lijian HUANG Songhua QIU Xiaojun |
| |
| Affiliation: | 1.AAC Acoustic R&D Centre(Nanjing),Nanjing 210093,China; 2.The Institute of Acoustics,Nanjing University,Nanjing 210093,China) |
| |
| Abstract: | The principle of speech enhancement algorithm based on short-time spectral analysis is presented.Improvements are made on the classical spectral subtractive method.The estimation of the a-priori-SNR and the time-recursive averaging algorithm for noise estimation are introduced and used.Several post-processing methods are used to further improve the results.The improved method is quite effective in speech enhancement as it can cancel a large amount of background noise with little distortion to speech signal.Objective and subjective experimental results indicate that this proposed method is superior to the known algorithms. |
| |
| Keywords: | spectral subtraction a-priori-SNR time-recursive averaging algorithm |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
