A Study on Speaker Adaptation of Large Continuous Spoken Language Using back-off bigram

Back-off bigram을 이랑한 대용량 연속어의 화자적응에 관한 연구

  • Published : 2003.09.01

Abstract

In this paper, we studied the speaker adaptation methods that improve the speaker independent recognition system. For the independent speakers, we compared the results between bigram and back-off bigram, MAP and MLLR. Cause back-off bigram applys unigram and back-off weighted value as bigram probability value, it has the effect adding little weighted value to bigram probability value. We did an experiment using total 39-feature vectors as featuring voice parameter with 12-MFCC, log energy and their delta and delta-delta parameter. For this recognition experiment, We constructed a system made by CHMM and tri-phones recognition unit and bigram and back-off bigrams language model.

본 논문에서는 화자 독립 시스템에서 필요한 화자 적응 방법에 관해 연구하였다. 훈련에 참여하지 않은 새로운 화자에 대해서 bigram과 back-off bigram, MAP와 MLLR의 결과를 비교해 보았다. back-off bigram은 훈련중 나타나지 않은 bigram 확률을 unigram과 back-off 가중치를 적용하므로 bigram 확률 값에 약간의 가중치를 더하는 효과를 가져온다. 음성의 특징 파라미터로는 12차의 MFCC와 log energy, 1차 미분, 2차 미분을 사용하여 총 39차의 특징 벡터를 사용하였다. 인식 실험을 위해 CHMM, 삼중음소(tri-phones)의 인식 단위, 그리고 bigram과 back-off bigram의 언어 모델을 사용한 시스템을 구성하였다.

Keywords

References

  1. Proc. ICASSP Speaker adaptation in continuous speech recognition via estimation of correlated mean vectors W.A.Rozzi;R.M.Stern
  2. Proc. ICASSP On speaker independent, speaker dependent, and speaker adaptive speech recognition X.D.Huang;K.F.Lee
  3. IEEE Trans. speech Audio Processing v.2 no.3 An acoustic-phonetic-based speaker adaptation technique for improving speaker independent continuous speech recognition Y.Zhao
  4. Proc. ICASSP A bayesian approach to speaker adaptation for the stochastic segment model B.F.Necioglu;M.Ostendorf;J.R.Rohlicek
  5. Fundamentals of Speech Recognition L.R.Rabiner;B.Juang
  6. Digital Signal Processing Oppenheim A.V.;Shafer R.W.
  7. Digital Coding of Waveforms Jayant N.O.S;Noll P.
  8. IEEE Tran. on ASSP v.29 no.2 Cepstral analysis techniques for automatic speaker verification Fruri S.
  9. Digital Processing of Speech Signal Rabiner L.R.;Shafer R.W.
  10. Computer Speech and Language v.8 On structuring Probabilstic dependencies in stochastic language modeling Ney H.;Essen U;Kneser R
  11. Technical Report LWMS-79 , Brown Unibersity Modelling and learning in speech recognition : the relationship between stochastic pattern chassigiera and neural networks Niles L.T.
  12. Handbook of Standards and Resources for Spoken Language Systems, Mouton de Gruyter Language models Ney H.
  13. The HTK Book Steve Young(et al.)
  14. Spoken Language Processing Xueding Huang;Alejandere Acero;Hsiao Wuen Hon
  15. Speech Recognition Claudio Becchetti;Lucio Prina Ricotti
  16. Proc. Eur. Conf. Speech Communication Technology Maximum a Posterior linear regression with elliptically symmetric matrix variate priors W.Chou