| 改善含噪语音说话人辨认系统性能的方法 |
| |
| 引用本文： | 韩春光,胡剑英,李华.改善含噪语音说话人辨认系统性能的方法[J].宁波大学学报(理工版),2007,20(3):297-300. |
| |
| 作者姓名： | 韩春光 胡剑英 李华 |
| |
| 作者单位： | 宁波大红鹰职业技术学院,浙江,宁波,315175 |
| |
| 摘 要： | 当对含噪语音进行说话人辨认时,系统的识别性能会明显变差,本文提出采用对倒谱参数非线性加权的方法,改善系统的噪声鲁棒性.通过对多种加权窗的正识率比较,发现对LPC倒谱低阶参数加权提升,对美尔倒谱高阶参数的加权提升,均提高了系统的识别性能.
|
| 关 键 词： | 含噪语音 说话人辨认 倒谱参数 非线性加权 改善 含噪语音 说话人辨认系统 识别性能 方法 Noise System Speaker Identification Performance Improve 参数加权 美尔倒谱 发现 比较 正识率 噪声鲁棒性 线性加权 倒谱参数 |
| 文章编号： | 1001-5132（2007）03-0297-04 |
| 修稿时间： | 2007-04-08 |
A Method to Improve Performance of Speaker Identification System with Noise |
| |
| HAN Chun-guang,HU Jian-ying,LI Hua.A Method to Improve Performance of Speaker Identification System with Noise[J].Journal of Ningbo University(Natural Science and Engineering Edition),2007,20(3):297-300. |
| |
| Authors: | HAN Chun-guang HU Jian-ying LI Hua |
| |
| Affiliation: | Ningbo Dahongying Vocational Technical College, Ningbo 315175, China |
| |
| Abstract: | The performance of the speaker recognition system tends to be deteriorated in the noisy speech cirsumstances. In this paper, a nonlinear weighting method is proposed for selecting the cepstral coefficient in an effort to improve the system robustness to noise. Having compared with a variety of recognition rates obtained in weighting windows, it is found that the system recognition performance is improved along with the increase of weighting of the low-order cepstral term of LPCC and the high-order cepstral term of MFCC. |
| |
| Keywords: | the speech with noise speaker identification cepstral coefficient nonlinear weighting |
| 本文献已被 维普 万方数据 等数据库收录！ |
|
