The Journal of the Acoustical Society of Korea (한국음향학회지)

The Acoustical Society of Korea (한국음향학회)
권호 표지

State-Dependent Feature-Parameter Weighting By the Contribition of the feature parameter to the performance of the Speech Recongition

음성인식에 있어서 특징 파라미터의 기여도에 기반한 상태별 특징 파라미터 가중

  • 최환진 (자연어정보처리연구부 시스템공학연구소)
  • Published : 1998.01.01
Download PDF
⟨ Previous Next ⟩

Abstract

본 논문에서는 은닉 마르코프 모델에 기반한 음성인식에 있어서 특징 파라미터의 인식 성능에 미치는 영향의 차이를 인식 시스템에 반영하여 인식성능을 향상시키기 위한 방 법을 제안하였다. 특징 파라미터별 가중치를 유도하기 위해서 우선 상태별 특징 파라미터의 인식율에 대한 기여도를 가중치로 변환하고, 이를 특징 파라미터 각각의 상태에서의 출력확 률에 곱하여 상태별 출력확률을 재 추정하게 된다. 실험결과, "가변가중"방법이 "고정가중" 방법에 비해서 단어 인식의 경우 3.3%, 그리고 문장 인식율의 경우 5.3%의 성능향상을 보 임으로써 상태별 특징 파라미터의 가중이 인식 성능 향상에 유효함을 알 수 있었다.

Keywords

References

  1. Fundamentals of Speech Recognition L. R. Rabiner;B. H. Juang
  2. PhD Thesis, Computer Science Department, Carnegie Mellon University The acoustic-modeling problem in automatic speech recognition P. Brown
  3. IEEE trans. on ASSP v.38 no.1 An overview of the SPHNIX Speech Recognition System K. F. Lee(et al.)
  4. Journal of Phonetics v.16 A joint synchrony/mean-rate model of auditory speech processing S. Seneff
  5. IEEE trans. on Acoust. Speech and Signal Processing(ASSP) v.24 no.2 Speech Recognition Experiments with Linear Prediction, Bandpass Filtering, and Dynamic Programming G. M. White;R. B. Neely
  6. Computer Science Department, Carnegie Mellon University Technical Report Evaluation of LPC Spectral Matching Measures for Phonetic Unit Recognition K. Shikano
  7. IEEE trans. on ASSP v.34 Speaker-independent isolated word recognition using dynamic features of speech spectrum S. Furui
  8. AT & T Bell Lab. Technical Journal v.63 no.7 On the Performance of Isolated Word Speech Recognizers Using Vector Quantization and Temporal Energy Contours L. R. Rabiner;K. C. Pan(et al.)
  9. Computer Speech and Language v.3 no.1 On the use of some robust modeling techniques for speech recognition C. H. Lee
  10. Artificial Intelligence in Engineering (to be published) Robust Estimation of Discrete HMM parameters using Entropy based Feature-Parameter Weighting and the Source-Quantization Modeling H. J. Choi;S. J. Yun;Y. H. Oh
  11. Proc. of ICASSP Integration of Acoustic Information in a Large Vocabulary Word Recognizer V. N. Gupta;M. Lenning(et al.)
  12. Proc. of ICASSP Improved hidden Markov modeling of phonemes for continuous speech recognition R. Schwartz;Y. Chow(et al.)
  13. IEEE trans. on ASSP v.37 no.8 High Performance connected digit recognition using Hidden Markov Models L. R. Rabiner;J. G. Wilpon(et al.)
  14. Pattern Classification and Scene Analysis R. O. Duda;P. E. Hart
  15. IEEE trans. on Comm. v.28 An algorithm for vector quantizer design Y. Linde;A. Buzo;R. M. Gray
  16. AT & T Technical Journal v.65 A segmental k-means training procedure for hidden Markov models with continuous mixture densities L. R. Rabiner;J. G. Wilpon(et al.)
  17. IEEE trans. on ASSP v.32 no.2 The use of a one-stage Dynamic Progamming algorithm for connected word recognition H. Ney
  18. IEEE trans. on Informaiton Theory v.IT-21 Design of a linguistic statistical decoder for the recognition of continuous speech F. Jelinek;L. R. Bahl(et al.)
  19. Computer Speech and Language v.6 On the use of inter-word context-de-pendent units for word juncture modeling E. P. Gialchin(et al.)