| 基于格的汉语自然对话语音索引方法研究 |
| |
| 引用本文： | 孟莎,余鹏,刘加.基于格的汉语自然对话语音索引方法研究[J].自动化学报,2010,36(2):215-220. |
| |
| 作者姓名： | 孟莎 余鹏 刘加 |
| |
| 作者单位： | 1.清华大学电子工程系清华信息科学与技术国家实验室(筹) 北京 100084 |
| |
| 基金项目： | 国家高技术研究发展计划(863计划)(2006AA010101,2007AA04Z223);;国家自然科学基金委员会与微软亚洲研究院联合资助项目(60776800)资助~~ |
| |
| 摘 要： | 对汉语自然对话语音索引问题进行了研究. 比较了不同单元格的识别和检索性能, 提出不同单元格的转换方法、格间的融合方法以及格内节点与边的合并方法. 格转换实现了识别单元和索引单元的分离, 词格转换得到的无调音节格将品质因数(Figure of merit, FOM)从基线系统的69.2%提高到73.7%; 格间融合综合利用多个格的信息, 将FOM进一步提高到78.6%; 格内合并对格进行了有效的压缩, 使其可应用于海量语音检索.
|
| 关 键 词： | 语音检索 语音索引 后验概率格 索引单元 |
| 收稿时间： | 2008-11-14 |
| 修稿时间： | 2009-4-8 |
Lattice-based Indexing for Spontaneous Mandarin Speech |
| |
| MENG Sha YU Peng LIU Jia .Tsinghua National Laboratory for Information Science , Technology.Lattice-based Indexing for Spontaneous Mandarin Speech[J].Acta Automatica Sinica,2010,36(2):215-220. |
| |
| Authors: | MENG Sha YU Peng LIU Jia Tsinghua National Laboratory for Information Science Technology |
| |
| Affiliation: | 1.Tsinghua National Laboratory for Information Science and Technology (TNList), Department of Electronic Engineering, Tsinghua University, Beijing 100084;2.Microsoft Research Asia, Beijing 100190 |
| |
| Abstract: | We examine the task of spoken term detection in Chinese spontaneous speech with a lattice-based approach. We compare lattices generated with different units and lattices converted from one unit to another. We find that the best system is with toneless-syllable lattices converted from word lattices whose figure of merit (FOM) is 73.7% from the baseline 69.2%. By combining lattices from multiple systems into a single lattice and fully exploiting the redundant information in the combined lattice with a time-based node/arc merging, we achieve the result of a compact lattice index with the accuracy improved up to 79.2%. |
| |
| Keywords: | Speech retrieval speech indexing posterior lattice indexing units |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《自动化学报》浏览原始摘要信息 |
|
点击此处可从《自动化学报》下载全文 |
