Computer Engineering and Applications ›› 2009, Vol. 45 ›› Issue (18): 242-244.DOI: 10.3778/j.issn.1002-8331.2009.18.073

• 工程与应用 • Previous Articles     Next Articles

Uyghur sentence selection algorithm of thriphone model

Guljamal Mamateli,Askar Ruzi,Askar Hamdulla   

  1. School of Information Science & Engineering,Xinjiang University,Urumqi 830046,China
  • Received:2008-04-10 Revised:2008-06-23 Online:2009-06-21 Published:2009-06-21
  • Contact: Guljamal Mamateli

三音素模型的维吾尔语最佳文本选取算法

姑丽加玛丽·麦麦提艾力,艾斯卡尔·肉孜,艾斯卡尔·艾木都拉   

  1. 新疆大学 信息科学与工程学院,乌鲁木齐 830046
  • 通讯作者: 姑丽加玛丽·麦麦提艾力

Abstract: In this paper,the algorithm of select best sentence from large text corpus based on thriphone models is proposed by using the contextual idea,the different thriphone models of each sentence is considered and sentence redundancy is removed by using the greedy algorithm.Consequently,the purpose of minimizing the selected text has achieved.The algorithm is realized by using the C# programming language.Algorithm process and performance analysis is given.Experimental results prove the effectiveness and practicality of the algorithm.

Key words: Uyghur language, text corpus, greedy algorithm, thriphone, speech synthesis

摘要: 利用上下文关联的思想,提出了三音素模型的大型句子文本库中选取最佳句子文本的算法,充分考虑了每个句子涵盖的不同三音素模型,利用贪婪算法去除了众多句子之间的冗余度,从而达到了选择文本容量最小化的目标。通过C#语言实现了本算法,给出了算法流程和算法性能分析,结果表明此算法的有效性和实用性。

关键词: 维吾尔语, 文本库, 贪婪算法, 三音素, 语音合成