| 混叠语音的场景分析方法 |
| |
| 引用本文： | 何映仙,许刚.混叠语音的场景分析方法[J].电声技术,2007,31(9):56-60. |
| |
| 作者姓名： | 何映仙 许刚 |
| |
| 作者单位： | 华北电力大学,电气与电子工程学院,北京,102206;华北电力大学,电气与电子工程学院,北京,102206 |
| |
| 摘 要： | 提出了1种混叠语音场景分析模型,模型以独立分量分析(ICA)方法作为语音分解的理论依据,以时频分布和距离测度作为语音辨认的实现工具,完成了混叠语音场景分析过程。阐述了ICA和时频分布的原理,采用改进快速ICA算法,结合优化时频分布和距离测度方法,有效完成了语音场景分析实验。通过对实验数据采用主客观评估,结果说明提出的模型具有较好的处理效果。
|
| 关 键 词： | 混叠语音 场景分析 独立分量分析 优化时频分布 距离测度 |
| 文章编号： | 1002-8684（2007）09-0056-05 |
| 修稿时间： | 2007-05-06 |
Scene Analysis Approach on Acoustically-Mixed Speech |
| |
| HE Ying-xian,XU Gang.Scene Analysis Approach on Acoustically-Mixed Speech[J].Audio Engineering,2007,31(9):56-60. |
| |
| Authors: | HE Ying-xian XU Gang |
| |
| Institution: | College of electrical and electronical engineering, North China Electronic Power University, Beijing 102206, China |
| |
| Abstract: | A scene analysis model of mixed-speech in natural environment is proposed. Improved FastICA algorithm is used to separate mixed speech, and optimized time-frequency distributions and distance measurement are combined to identify unmixed speech. Experimental results show that the proposed approach is effective in processing acoustically-mixed speech. |
| |
| Keywords: | mixed-speech scene analysis independent component analysis optimized time-frequency distribution distance measurement |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
