计算机工程与应用 ›› 2009, Vol. 45 ›› Issue (15): 28-30.DOI: 10.3778/j.issn.1002-8331.2009.15.008

• 博士论坛 • 上一篇    下一篇

汉语连续变调语音合成算法

康广玉1,郭世泽2,孙圣和1   

  1. 1.哈尔滨工业大学 自动化测试与控制系,哈尔滨 150001
    2.总参54所,北京 100093
  • 收稿日期:2008-12-22 修回日期:2009-03-02 出版日期:2009-05-21 发布日期:2009-05-21
  • 通讯作者: 康广玉

Speech synthesize using Chinese tone sandhi

KANG Guang-yu1,GUO Shi-ze2,SUN Sheng-he1   

  1. 1.Department of Automatic Test and Control,Harbin Institute of Technology,Harbin 150001,China
    2.The 54th Research Institute of General Staff,Beijing 100093,China
  • Received:2008-12-22 Revised:2009-03-02 Online:2009-05-21 Published:2009-05-21
  • Contact: KANG Guang-yu

摘要: 汉语是声调语言,同一个音节带上不同的声调可以表述不同的语义。发音时两个或两个以上的音节连在一起时,音节所属调类调值所发生变化的现象称为“连读变调”,目前的语音合成系统没有考虑连续变调,使得合成语音自然度不够。采用TD-PSOLA对具有连续变调现象的语音进行合成,实验表明合成语音的自然度较高,是适合小语料库语音合成的良好算法。

Abstract: Chinese is a tone language,and the different tone in one same syllable band represents the different meaning.The phenomena that the tone value of tone class in the syllable changes occurs when two or plural syllables of tone language connect each other,and the phenomena is called “tone sandhi”.Current speech synthesize systems do not consider the tone sandhi,which causes the naturalness of synthesized speech is not high.This paper synthesizes the speech with the tone sandhi using TD-PSOLA.The experimental results show that the naturalness of synthesized speech is high,and the proposed algorithm is an excellent algorithm of speech synthesizing for a small size of samples database.