计算机工程与应用 ›› 2010, Vol. 46 ›› Issue (13): 233-235.DOI: 10.3778/j.issn.1002-8331.2010.13.069

• 工程与应用 • 上一篇    下一篇

藏语连续语音语料库设计与实现

李永宏1,于洪志1,孔江平2   

  1. 1.西北民族大学 中国民族语言文字信息技术重点实验室,兰州 730030
    2.北京大学 中文系汉语语言学研究中心,北京 100871

  • 收稿日期:2008-10-30 修回日期:2008-12-04 出版日期:2010-05-01 发布日期:2010-05-01
  • 通讯作者: 李永宏

Design and implementation of Tibetan continuous speech corpus

LI Yong-hong1,YU Hong-zhi1,KONG Jiang-ping2   

  1. 1.Key Lab of China’s National Linguistic Information Technology,Northwest University for Nationalities,Lanzhou 730030,China
    2.Department of Chinese Language and Literature,Peking University,Beijing 100871,China

  • Received:2008-10-30 Revised:2008-12-04 Online:2010-05-01 Published:2010-05-01
  • Contact: LI Yong-hong

摘要: 以藏语夏河话为研究对象,建立了基于三音子的藏语连续语音语料库。首先收集了10万句藏语文本语料库,并根据夏河话的实际发音,进行了国际音标转写;然后总结了夏河话的三音子音联结构形式,并用藏语文本处理平台对其组合类型和在原始文本语料库中的频度进行了详细的统计分析;最后在语音库的语料设计中综合考虑了三音子以及类三音子的覆盖率和稀疏度,设计并完成了语料抽取算法,实现了语料自动选取。

关键词: 藏语, 三音子, 语音库, Greed算法

Abstract: By taking Tibetan Xiahe dialect as the research object,continuous speech corpus based on triphone is built.At first,text corpus with 100 thousand sentences is collected and they are transformed to IPA according to pronunciation of Xiahe dialect,and then structure of triphone juncture is summarized and combination type and frequency of triphone in Corpus are statistically analyzed with text-processing platform in detail.At last by comprehensively considering coverage rate and sparseness of triphone and class-triphone the algorithm for extraction of corpus is designed and automatic selection to corpus is realized.

Key words: Tibetan, triphone, speech corpus, Greed algorithm

中图分类号: