C-TOBI-Based Pitch Accent Prediction Using Maximum-Entropy Model

  • Byeongchang Kim
  • Gary Geunbae Lee
Part of the Lecture Notes in Computer Science book series (LNCS, volume 3982)


We model Chinese pitch accent prediction as a classification problem with six C-ToBI pitch accent types, and apply conditional Maximum Entropy (ME) classification to this problem. We acquire multiple levels of linguistic knowledge from natural language processing to make well-integrated features for ME framework. Five kinds of features were used to represent various linguistic constraints including phonetic features, POS tag features, phrase break features, position features, and length features.


Training Corpus Position Feature Gaussian Smoothing Pitch Contour Pitch Accent 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Berger, A.L., Pietra, S.A.D., Pietra, V.J.D.: A maximum entropy approach to natural language processing. Computational Linguistics 22(1) (1996)Google Scholar
  2. 2.
    Black, A.W., Hunt, A.J.: Generating F0 contours from ToBI labels using linear regression. In: Proceeding of the international conference on spoken language processing(ICSLP), CSLI (1996)Google Scholar
  3. 3.
    Chen, S.F., Rosenfeld, R.: A Gaussian Prior for Smoothing Maximum Entropy Models. Technical Report CMU-CS-99-108 (1999)Google Scholar
  4. 4.
    Gregory, M.L., Altun, Y.: Using conditional random fields to predict pitch accents in conversational speech, ACL (2004)Google Scholar
  5. 5.
    Ha, J.-H., Zheng, Y., Lee, G.G.: Chinese segmentation and POS-tagging by automatic POS dictionary training. In: Proceedings of the 14th Conference of Korean and Korean Information Processing (2002)Google Scholar
  6. 6.
    Ha, J.-H., Zheng, Y., Lee, G.G.: High speed unknown word prediction using support vector machine for Chinese Text-to-Speech systems. In: Su, K.-Y., Tsujii, J., Lee, J.-H., Kwong, O.Y. (eds.) IJCNLP 2004. LNCS (LNAI), vol. 3248, pp. 509–517. Springer, Heidelberg (2005)CrossRefGoogle Scholar
  7. 7.
    Malouf, R.: A comparison of algorithms for maximum entropy parameter estimation. In: Proceedings of CoNLL 2002, Taipei, Taiwan, pp. 49–55 (2002)Google Scholar
  8. 8.
    Sun, X.: Pitch accent prediction using ensemble machine learning. In: ICSLP 2002 (2002)Google Scholar
  9. 9.
    Zhang, H., Yu, J.S., Zhan, W.D., Yu, S.W.: Disambiguation of Chinese polyphonic characters. In: International Workshop on Multimedia Annotation (2001)Google Scholar
  10. 10.
    Le, Z.: Maximum entropy modeling toolkit for python and C++ (2003),
  11. 11.
    Zheng, Y., Kim, B., Lee, G.G.: Using multiple linguistic features for Mandarin phrase break prediction in maximum-entropy classification framework. In: ICSLP 2004 (2004)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2006

Authors and Affiliations

  • Byeongchang Kim
    • 1
  • Gary Geunbae Lee
    • 2
  1. 1.School of Computer & Information Communications EngineeringCatholic University of DaeguSouth Korea
  2. 2.Department of Computer Science & EngineeringPohang University of Science & TechnologyPohangSouth Korea

Personalised recommendations