The Pausing Method Based on Brown Clustering and Word Embedding

  • Arman Kaliyev
  • Sergey V. Rybin
  • Yuri Matveev
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 10458)


One of the most important parts of the synthesis of natural speech is the correct pause placement. Properly placed pauses in speech affect the perception of information. In this article, we consider the method of predicting pause positions for the synthesis of speech. For this purpose, two speech corpora were prepared in the Kazakh language. The input parameters were vector representations of words obtained from the cluster model and from the algorithm of the canonical correlations analysis. The support vector machine was used to predict the pauses within the sentence. Our results show F-1 = 0.781 for pause prediction.


Speech synthesis Pause Prosodic boundaries Statistical models 



This work was financially supported by the Government of the Russian Federation, Grant 074-U01.


  1. 1.
    Parlikar, A., Black, A.W.: Modeling pause-duration for style-specific speech synthesis. In: INTERSPEECH, pp. 446–449. ISCA (2012)Google Scholar
  2. 2.
    Norkevicius, G., Raskinis, G.: Modeling phone duration of Lithuanian by classification and regression trees, using very large speech corpus. Inf. Lith. Acad. Sci. 19(2), 271–284 (2008)Google Scholar
  3. 3.
    Bali, K., Nemala, S.K., Ramakrishnan, A.G., Talukdar, P.P.: Duration modeling for Hindi text-to-speech synthesis system. In: INTERSPEECH (2004)Google Scholar
  4. 4.
    Brown, P.F., et al.: Class-based n-gram models of natural language. Comput. Linguist. 18(4), 467–479 (1992)Google Scholar
  5. 5.
    Stratos, K. Kim, D., Collins, M., Hsu., D.: A spectral algorithm for learning class-based n-gram models of natural language. In: Zhang, N.L., Tian, J. (eds.) UAI, pp. 762–771. AUAI Press (2014)Google Scholar
  6. 6.
    Cortes, C., Vapnik, V.: Support vector networks. Mach. Learn. 20, 273–297 (1995)zbMATHGoogle Scholar
  7. 7.
    Sarkar, P., Sreenivasa, R.K.: Data-driven pause prediction for speech synthesis in storytelling style speech. In: Twenty First National Conference on Communications (2015)Google Scholar
  8. 8.
    Parlikar, A., Black, A.W.: A grammar based approach to style specific phrase prediction. In: INTERSPEECH, pp. 2149–2152. ISCA (2011)Google Scholar
  9. 9.
    Miller, S., Guinness, J., Zamanian, A.: Name tagging with word clusters and discriminative training. In: HLT-NAACL, pp. 337–342 (2005)Google Scholar
  10. 10.
    Koo, T., Carreras, X., Collins, M.: Simple semi-supervised dependency parsing. In: Proceedings of ACL 2008: HLT, pp. 595–603. Association for Computational Linguistics, Columbus (2008)Google Scholar
  11. 11.
    Stratos, K., Collins, M.: Simple semi-supervised POS tagging. In: Blunsom, P., et al. (eds.) VS@HLT-NAACL, pp. 79–87. The Association for Computational Linguistics (2015)Google Scholar
  12. 12.
    Loh, W.-Y.: Classification and regression tree methods. In: Encyclopedia of Statistics in Quality and Reliability, pp. 315–323. Wiley (2008)Google Scholar
  13. 13.
    Breiman, L.: Random forests. Mach. Learn. 45, 5–32 (2001)CrossRefzbMATHGoogle Scholar
  14. 14.
    van Rijsbergen, C.J.: Information Retrieval. Butterworths, London (1979)zbMATHGoogle Scholar
  15. 15.
    Chistikov, P., Khomitsevich, O.: Improving prosodic break detection in a Russian TTS system. In: Železný, M., Habernal, I., Ronzhin, A. (eds.) SPECOM 2013. LNCS (LNAI), vol. 8113, pp. 181–188. Springer, Cham (2013). doi: 10.1007/978-3-319-01931-4_24 CrossRefGoogle Scholar
  16. 16.
    Chistikov, P.G., Khomitsevich, O.G., Rybin, S.V.: Statistical methods for automatic prosodic break detection in a text-to-speech systems. J. Instrum. Eng. 57(2), 28–32 (2014). (in Russian)Google Scholar

Copyright information

© Springer International Publishing AG 2017

Authors and Affiliations

  1. 1.ITMO UniversitySaint PetersburgRussia

Personalised recommendations