语音检索中子词单元的构建算法

doi:10.3969/j.issn.1000-3428.2012.24.059

计算机工程 ›› 2012, Vol. 38 ›› Issue (24): 251-253.doi: 10.3969/j.issn.1000-3428.2012.24.059

语音检索中子词单元的构建算法

杨乐，吴及，吕萍

(清华大学电子工程系，北京 100084)

收稿日期:2012-03-01 修回日期:2012-04-27 出版日期:2012-12-20 发布日期:2012-12-18
作者简介:杨乐(1988－)，女，硕士研究生，主研方向：语音识别；吴及，副教授、博士、博士生导师；吕萍，副研究员、博士
基金项目:
国家自然科学基金资助项目(61170197)；清华大学自主科研计划基金资助项目(2011thz0)

Construction Algorithm of Sub-word Unit in Speech Retrieval

YANG Le, WU Ji, LV Ping

(Department of Electronic Engineering, Tsinghua University, Beijing 100084, China)

Received:2012-03-01 Revised:2012-04-27 Online:2012-12-20 Published:2012-12-18

摘要/Abstract

摘要： 针对语音关键词检索中的集外词问题，提出基于最大互信息-最小描述长度(MMI-MDL)的子词集构建算法。根据子词对的互信息挑选聚合对，通过MDL准则判断是否聚合成新的子词。使用该子词集把单词映射成子词的组合用于检索。实验结果表明，与已有的MDL子词集构建算法相比，由MMI-MDL方法得到的子词集对检索性能有较大提高，在相同精确率指标下，集外词的召回率相对MDL算法提高12.1%。

关键词: 集外词, 语音检索, 子词, 最小描述长度, 最大互信息, 词格网络

Abstract: In order to solve the Out-of-Vocabulary(OOV) problem in speech retrieval tasks, this paper presents a construction algorithm of sub-word units based on Maximum Mutual Information and Minimum Description Length(MMI-MDL). It selects candidate pairs according to the mutual information of sub-word pairs, judges whether combining the pairs to a new sub-word through MDL. After getting the sub-word set, map the word into sub-word for retrieval. Experimental results show that compared with the MDL algorithm, the proposed method has a better performance, and achieves a 12.1% relative improvement on the OOV recall rate.

Key words: Out-of-Vocabulary(OOV), speech retrieval, sub-word, Minimum Description Length(MDL), Max Mutual Information(MMI), word lattice network

中图分类号:

TP301.6

杨乐, 吴及, 吕萍. 语音检索中子词单元的构建算法[J]. 计算机工程, 2012, 38(24): 251-253.

YANG Le, TUN Ji, LV Ping. Construction Algorithm of Sub-word Unit in Speech Retrieval[J]. Computer Engineering, 2012, 38(24): 251-253.

参考文献

[1] NIST. The Spoken Term Detection(STD) 2006 Evaluation Plan[EB/OL]. (2006-09-16). http://www.itl.nist.gov/iad/mig/tests/ std/2006/docs/std06-evalplan-v10.pdf.
[2] Wang Dong, King S. Stochastic Pronunciation Modeling and Soft Match for Out-of-Vocabulary Spoken Term Detection[C]//Proc. of IEEE International Conference on Acoustics Speech and Signal Processing. [S. l.]: IEEE Press, 2010.
[3] 包叶波, 胡郁. 中文连续语音识别系统音素建模单元集的构建[J]. 清华大学学报: 自然科学版, 2011, 51(9): 1288-1292.
[4] Gouvea E. Subword Unit Approaches for Retrieval by Voice[C]// Proc. of IEEE International Conference on Acoustics Speech and Signal Processing. [S. l.]: IEEE Press, 2010.
[5] Hewlett D. Fully Unsupervised Word Segmentation with BVE and MDL[C]//Proc. of the 49th Annual Meeting of the Association for Computational Linguistics. Portland, USA: [s. n.], 2011.
[6] Bisani M, Ney H. Open Vocabulary Speech Recognition with Flat Hybrid Models[C]//Proc. of the European Conference on Speech Communication and Technology. Lisboa, Portugal: [s. n.], 2005.
[7] Akbacak M, Vergyri D, Stolcke A. Open-vocabulary Spoken Term Detection Using Graphone-based Hybrid Recognition Systems[C]// Proc. of IEEE International Conference on Acoustics, Speech and Signal Processing. Las Vegas, USA: [s. n.], 2008.
[8] Reddy S, Goldsmith J. An MDL-based Approach to Extracting Subword Units for Grapheme-to-Phoneme Conversion[C]//Proc. of Annual Conference of the North American Chapter of the Association for Computational Linguistics. Stroudsburg, USA: ACM Press, 2010.
[9] NIST. Nistg2p[EB/OL]. (2003-07-01). ftp://jaguar.ncsl.nist.gov// pub/addttp4-1.1.tar.
[10] Rissanen J. A Universal Prior for Integers and Estimation by Minimum Description Length[J]. The Annals of Statistics, 1983, 11(2): 416-431.
[11] 王鑫, 赵春晖. 基于MDL准则的MUSIC信号频率估计算法[J]. 数据采集与处理, 2008, 23(z1): 1-5.

[1]	谢彬,张琨,张云纯,蔡颖,蒋彤彤. 基于轨迹相似度的移动目标轨迹预测算法[J]. 计算机工程, 2018, 44(9): 177-183.
[2]	曹卫权,李智翔,魏强,褚衍杰. 基于区域分布概率密度估计的轨迹分类方法[J]. 计算机工程, 2018, 44(4): 262-267,286.
[3]	魏长宝,姚汝贤. 基于最小描述长度的图分割结构检测改进算法[J]. 计算机工程, 2016, 42(1): 231-236,242.
[4]	魏长宝,姚汝贤. 基于最小描述长度的图分割变化检测改进算法[J]. 计算机工程, 2015, 41(7): 274-279,284.
[5]	朱世磊, 任丙印, 王大鸣, 仵国锋. 一种基于子空间分解的认知MIMO传输机制[J]. 计算机工程, 2012, 38(20): 1-3.
[6]	贾可新, 何子述. 基于CSMDEM算法的GMM学习方法[J]. 计算机工程, 2011, 37(19): 153-156.
[7]	李东明, 张丽娟, 赵伟, 石晶. 基于MDL和LSC的语义优选方法[J]. 计算机工程, 2011, 37(17): 15-18.
[8]	张政伟. 模型未知的非双曲型非线性序列去噪算法[J]. 计算机工程, 2011, 37(15): 6-9.
[9]	杨辉;张玥杰;张涛. 基于词典的英汉双向跨语言信息检索方法[J]. 计算机工程, 2009, 35(16): 273-274.
[10]	孙成立;刘刚;郭军. 子词驻留特征在电话语音确认中的应用[J]. 计算机工程, 2009, 35(1): 27-29.
[11]	冯林;于孝航;孙焘;沈骁;潘晓雯. 基于最长公共子序列距离的主旨模式挖掘算法[J]. 计算机工程, 2008, 34(14): 47-48.
[12]	叶靓;王智斌;邵谦明. 基于相关反馈的语音检索引擎[J]. 计算机工程, 2007, 33(17): 228-230.
[13]	普鑫. 基于混合概率PCA模型高光谱图像本征维数确定[J]. 计算机工程, 2007, 33(09): 204-206.

语音检索中子词单元的构建算法

Construction Algorithm of Sub-word Unit in Speech Retrieval

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 13

Metrics

本文评价

推荐阅读 10