On the Impact of Annotation Errors on Unit-Selection Speech Synthesis

  • Jindřich Matoušek
  • Daniel Tihelka
  • Luboš Šmídl
Part of the Lecture Notes in Computer Science book series (LNCS, volume 7499)

Abstract

Unit selection is a very popular approach to speech synthesis. It is known for its ability to produce nearly natural-sounding synthetic speech, but, at the same time, also for its need for very large speech corpora. In addition, unit selection is also known to be very sensitive to the quality of the source speech corpus the speech is synthesised from and its textual, phonetic and prosodic annotations and indexation. Given the enormous size of current speech corpora, manual annotation of the corpora is a lengthy process. Despite this fact, human annotators do make errors. In this paper, the impact of annotation errors on the quality of unit-selection-based synthetic speech is analysed. Firstly, an analysis and categorisation of annotation errors is presented. Then, a speech synthesis experiment, in which the same utterances were synthesised by unit-selection systems with and without annotation errors, is described. Results of the experiment and the options for fixing the annotation errors are discussed as well.

Keywords

speech synthesis unit selection annotation errors 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Tihelka, D., Kala, J., Matoušek, J.: Enhancements of Viterbi Search for Fast Unit Selection Synthesis. In: Proc. Interspeech, Makuhari, Japan, pp. 174–177 (2010)Google Scholar
  2. 2.
    Hanzlíček, Z.: Czech HMM-Based Speech Synthesis. In: Sojka, P., Horák, A., Kopeček, I., Pala, K. (eds.) TSD 2010. LNCS, vol. 6231, pp. 291–298. Springer, Heidelberg (2010)CrossRefGoogle Scholar
  3. 3.
    Cox, S., Brady, R., Jackson, P.: Techniques for Accurate Automatic Annotation of Speech Waveforms. In: Proc. ICSLP, Sydney, Australia (1998)Google Scholar
  4. 4.
    Tachibana, R., Nagano, T., Kurata, G., Nishimura, M., Babaguchi, N.: Preliminary Experiments Toward Automatic Generation of New TTS Voices from Recorded Speech Alone. In: Proc. Interspeech, Antwerp, Belgium, pp. 1917–1920 (2007)Google Scholar
  5. 5.
    Aylett, M.P., King, S., Yamagishi, J.: Speech Synthesis Without a Phone Inventory. In: Proc. Interspeech, Brighton, England, pp. 2087–2090 (2009)Google Scholar
  6. 6.
    Matoušek, J., Romportl, J.: Recording and Annotation of Speech Corpus for Czech Unit Selection Speech Synthesis. In: Matoušek, V., Mautner, P. (eds.) TSD 2007. LNCS (LNAI), vol. 4629, pp. 326–333. Springer, Heidelberg (2007)CrossRefGoogle Scholar
  7. 7.
    Matoušek, J., Tihelka, D., Psutka, J.V.: Experiments with Automatic Segmentation for Czech Speech Synthesis. In: Matoušek, V., Mautner, P. (eds.) TSD 2003. LNCS (LNAI), vol. 2807, pp. 287–294. Springer, Heidelberg (2003)CrossRefGoogle Scholar
  8. 8.
    Švec, J., Šmídl, L.: Prototype of Czech Spoken Dialog System with Mixed Initiative for Railway Information Service. In: Sojka, P., Horák, A., Kopeček, I., Pala, K. (eds.) TSD 2010. LNCS, vol. 6231, pp. 568–575. Springer, Heidelberg (2010)CrossRefGoogle Scholar
  9. 9.
    Železný, M., Krňoul, Z., Císař, P., Matoušek, J.: Design, Implementation and Evaluation of the Czech Realistic Audio-Visual Speech Synthesis. Signal Processing 12, 3657–3673 (2006)CrossRefGoogle Scholar
  10. 10.
    Müller, L., Psutka, J.V., Šmídl, L.: Design of Speech Recognition Engine. In: Sojka, P., Kopeček, I., Pala, K. (eds.) TSD 2000. LNCS (LNAI), vol. 1902, pp. 259–264. Springer, Heidelberg (2000)CrossRefGoogle Scholar
  11. 11.
    Šmídl, L., Trmal, J.: Keyword Spotting Result Post-processing to Reduce False Alarms. In: Recent Advances in Signals and Systems, vol. 9, pp. 49–52. WSEAS Press, Budapest (2009)Google Scholar
  12. 12.
    Malfrere, F., Deroo, O., Dutoit, T., Ris, C.: Phonetic Alignment: Speech Synthesis-Based Vs. Viterbi-Based. Speech Communication 40, 503–515 (2003)CrossRefGoogle Scholar
  13. 13.
    Lu, H., Wei, S., Dai, L., Wang, R.-H.: Automatic Error Detection for Unit Selection Speech Synthesis Using Log Likelihood Ratio Based SVM Classifier. In: Proc. Interspeech, Makuhari, Japan, pp. 162–165 (2010)Google Scholar
  14. 14.
    Grůber, M.: Acoustic Analysis of Czech Expressive Recordings from a Single Speaker in Terms of Various Communicative Functions. In: Proc. ISSPIT, Bilbao, Spain, pp. 267–272 (2011)Google Scholar
  15. 15.
    Přibil, J., Přibilová, A.: An Experiment with Evaluation of Emotional Speech Conversion by Spectrograms. Measurement Science Review 10, 72–77 (2010)CrossRefGoogle Scholar
  16. 16.
    Matoušek, J., Skarnitzl, R., Machač, P., Trmal, J.: Identification and Automatic Detection of Parasitic Speech Sounds. In: Proc. Interspeech, Brighton, England, pp. 876–879 (2009)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2012

Authors and Affiliations

  • Jindřich Matoušek
    • 1
  • Daniel Tihelka
    • 1
  • Luboš Šmídl
    • 1
  1. 1.Faculty of Applied Sciences, Dept. of CyberneticsUniversity of West BohemiaPlzeňCzech Republic

Personalised recommendations