| 一种基于噪声模型的语音/噪声分类方法 |
| |
| 引用本文： | 吴边,任晓林,刘重庆,张亚昕.一种基于噪声模型的语音/噪声分类方法[J].上海交通大学学报,2004,38(9):1528-1531. |
| |
| 作者姓名： | 吴边 任晓林 刘重庆 张亚昕 |
| |
| 作者单位： | 1. 上海交通大学,图像处理与模式识别研究所,上海,200030
2. 摩托罗拉中国研究中心,上海,200041 |
| |
| 摘 要： | 提出了一种可用于嵌入式ASR系统的语音/噪声分类的新方法，该方法利用一个噪声模型，对每帧信号的评价值进行语音/噪声分类．实验表明，该方法可以有效地区分语音和噪声，并表现出在各种噪声环境和不同信噪比条件下的鲁棒性．该算法已经集成进一个ASR系统，并在Corn-paq iPAQ上进行了测试，其计算代价不到整个系统代价的10％．
|
| 关 键 词： | 语音识别 语音／噪声分类 噪声鲁棒性 |
| 文章编号： | 1006-2467(2004)09-1528-04 |
| 修稿时间： | 2003年9月2日 |
A Noise Model Based Method for Speech/Noise Discrimination |
| |
| WU Bian,REN Xiao-lin,LIU Chong-qing,ZHANG Ya-xin.A Noise Model Based Method for Speech/Noise Discrimination[J].Journal of Shanghai Jiaotong University,2004,38(9):1528-1531. |
| |
| Authors: | WU Bian REN Xiao-lin LIU Chong-qing ZHANG Ya-xin |
| |
| Institution: | WU Bian~1,REN Xiao-lin~2,LIU Chong-qing~1,ZHANG Ya-xin~2 |
| |
| Abstract: | A robust method suitable for embedded automatic speech recognition(ASR) system was proposed to distinguish speech signals from noise signals. The method is based on a noise model, according to which a score is used to distinguish speech from noise. The experiments show that the method is effective in speech/noise discrimination and robust in various environments at any SNR. The algorithm has been employed as the front-end of an ASR system on Compaq iPAQ, and its cost is less than 10% of the ASR system. |
| |
| Keywords: | speech recognition speech/noise discrimination noise robustness |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
