$\require{mediawiki-texvc}$

연합인증

연합인증 가입 기관의 연구자들은 소속기관의 인증정보(ID와 암호)를 이용해 다른 대학, 연구기관, 서비스 공급자의 다양한 온라인 자원과 연구 데이터를 이용할 수 있습니다.

이는 여행자가 자국에서 발행 받은 여권으로 세계 각국을 자유롭게 여행할 수 있는 것과 같습니다.

연합인증으로 이용이 가능한 서비스는 NTIS, DataON, Edison, Kafe, Webinar 등이 있습니다.

한번의 인증절차만으로 연합인증 가입 서비스에 추가 로그인 없이 이용이 가능합니다.

다만, 연합인증을 위해서는 최초 1회만 인증 절차가 필요합니다. (회원이 아닐 경우 회원 가입이 필요합니다.)

연합인증 절차는 다음과 같습니다.

최초이용시에는
ScienceON에 로그인 → 연합인증 서비스 접속 → 로그인 (본인 확인 또는 회원가입) → 서비스 이용

그 이후에는
ScienceON 로그인 → 연합인증 서비스 접속 → 서비스 이용

연합인증을 활용하시면 KISTI가 제공하는 다양한 서비스를 편리하게 이용하실 수 있습니다.

Recognition Performance Improvement of Unsupervised Limabeam Algorithm using Post Filtering Technique 원문보기

대한임베디드공학회논문지 = IEMEK Journal of embedded systems and applications, v.8 no.4, 2013년, pp.185 - 194  

Nguyen, Dinh Cuong (Department of information and communication engineering, Yeungnam university) ,  Choi, Suk-Nam (Department of information and communication engineering, Yeungnam university) ,  Chung, Hyun-Yeol (Department of the information and communication, Yeungnam University)

Abstract ▼ AI-Helper 아이콘AI-Helper

Abstract- In distant-talking environments, speech recognition performance degrades significantly due to noise and reverberation. Recent work of Michael L. Selzer shows that in microphone array speech recognition, the word error rate can be significantly reduced by adapting the beamformer weights to ...

주제어

#Robust speech recognition   #Microphone array processing   #Beamforming   #Limabeam   #hidden markov model   #HM-Net   #Post filtering   #Feature Weighted Malahanobis Distance  

AI 본문요약
AI-Helper 아이콘 AI-Helper

* AI 자동 식별 결과로 적합하지 않은 문장이 있을 수 있으니, 이용에 유의하시기 바랍니다.

제안 방법

  • HM-Net speech recognition system was used for all experiment in this paper. 1004 states (8 Gaussians/state) HM-Net system were trained using Trade database, a speaker-independent database consisting of 8892 utterances uttered by 90 speakers.
  • Originally, Limabeam was investigated with Sphinx3, an HMM-based large-vocabulary speech recognition system [7], and English database. In this paper, we present the results of investigation the performance of speech recognition using microphone array and the way to implement Limabeam algorithm with Hidden Markov Network (HM-Net) speech recognition system and Korean database.
  • In this paper, we proposed an alternative way to implement Limabeam algorithm in Hidden Markov Network speech recognizer for efficient implementation and we proposed to add a post filter technique with Feature Weighted Mahalanobis Distance to Limabeam algorithm in order to improve recognition performance. From our prior investigation for the unsupervised Limabeam, we could see that because the performance of optimization depended strongly on the transcription output of the first recognition step, the output became unstable and that caused to lead lower performance.
  • 1004 states (8 Gaussians/state) HM-Net system were trained using Trade database, a speaker-independent database consisting of 8892 utterances uttered by 90 speakers. The system was trained using 39-dimensional feature vectors consisting of 13 MFCC parameters, along with their delta and delta-delta parameters. A 25-ms window length and a 10-ms frame shift were used.

대상 데이터

  • In order to investigate the performance of speech recognition with microphone array, we employed two microphone array databases recorded at Yeungnam University. In the first database, YUM4-6, we play backed Trade6 databases (596 utterances uttered by 6 speakers) through an Harman/Kardon loudspeaker and used a linear B&K microphone array with 4 elements spaced 20cm apart for recording.

이론/모형

  • In the experiment, the channels were aligned based on time delays estimated by GCC-PHAT method. The aligned channels were then averaged to generate the delay-and-sum beamforming output signal.
  • Delay-and-Sum (D&S) is the most popular and simplest method in microphone array processing. To process, one channel is chosen as a reference and the Time-Difference Of Arrival (TDOA) for the rest of the channels is estimated using Generalized Cross-Correlation (GCC) Phase Transform (PHAT) [12] or any other Time Delay Estimation (TDE) techniques. Next the time aligned speech signals are summed up as
본문요약 정보가 도움이 되었나요?

참고문헌 (19)

  1. M.F. Font, "Multi-microphone signal processing for automatic speech recognition in meeting rooms," Master thesis, Berkeley, California, 2005. 

  2. P.J. Moreno, "Speech recognition in noisy environments," Doctoral dissertation, Carnegie Mellon University, Pittsburgh, PA, 1996. 

  3. F.H. Liu, "Environmental adaptation for robust speech recognition," Doctoral dissertation, Carnegie Mellon University, Pittsburgh, PA, 1994. 

  4. A. Acero, "Acoustical and environmental robustness in automatic speech recognition," Doctoral dissertation, Carnegie Mellon University, Pittsburgh, PA, 1990. 

  5. L.J. Griffiths, C.W. Jim, "An alternative approach to linearly constrained adaptive beamforming," IEEE Transaction on Antennas and Propagation, Vol. AP-30, No. 1, pp.27-34, 1982. 

  6. M. Seltzer, "Microphone Array Processing for Robust Speech Recognition," Doctoral dissertation, Carnegie Mellon University, Pittsburgh, PA, 2003. 

  7. P. Placeway, S. Chen, M. Eskenazi, U. Jain, V. Parikh, B. Raj, M. Ravishankar, R. Ronsenfeld, K. Seymore, M. Siegler, R. Stern, E. Thayer, "The 1996 hub-4 sphinx-3 system," Proceedings on the DARPA Speech Recognition workshop, Vol. 1, pp.243-252, 1997. 

  8. S.J. Oh, C.J. Hwang, H.Y. Jung, H.Y. Chung, "A study on statistical language models for large vocabulary continuous speech recognition system," Proceedings on ICSP, Vol. 1, pp.113-119, 1999. 

  9. W.H. Press, B.P. Flannery, S.A. Teukolsky, W.T. Vetterling, "Numerical Recipes in C: The Art of Scientific Computin," New York: Cambridge University Press, 1998. 

  10. L. Rabiner, B.H. Juang, "Fundamentals of Speech Recognition," New Jersey: Prentice Hall, 1993. 

  11. A.J. Viterbi, "Error bounds for convolutional codes and an asymptotically optimum decoding algorithm," IEEE Transactions on Information Theory, Vol. 13, No. 2, pp.260-269, 1967. 

    상세보기 crossref

  12. P. Morento, B. Raj, R.M. Stern, "A Unified Approach for Robust Speech Recognition," Proceedings of Eurospeech, Vol. 1, pp.481-485, 1995. 

  13. R. Zenlinski, "A microphone array with adaptive post-filtering for noise reduction in reverberant room," Proceedings on International Conference of Acoustics, Speech, and Signal Processing, Vol. 5, pp.2578-2581, 1988. 

  14. I.A. McCowan, "Robust speech recognition using microphone arrays," Doctoral dissertation, Queen land University of Technology, Australia, 2001. 

  15. C.H. Knapp, C. Carter, "The generalized correlation method for estimation of time delay," IEEE Transactions on Acoustics, Speech and Signal Processing, Vol. 24, No. 4, pp.320-327, 1976. 

    상세보기 crossref

  16. D.C. Nguyen, H.Y. Chung, "Performance Improvement of Microphone Array Speech Recognition using Feature Weighted Mahalanobis Distance," The Journal of the Acoustical Society of Korea, Vol. 29, No. 1E, pp.45-53, 2010. 

  17. N.D. Cuong, S. Guanghu, J.H. Youl, C.H. Yeol, "Performace improvement of speech recognition system using microphone array," Proceedings on IEEE International Conference of Research, Innovation and Vision for the Future, pp.91-95, 2008. 

  18. L. Brayda, C. Wellekens, M. Omologo, "N-Best Parallel Maximum Likelihood Beamformers for Robust Speech Recognition," Proceedings of European Signal Processing Conference, 2006. 

  19. K. Bahram, B. Hamidreza, R. Farbod, "Improvement in speech recognition using phone-based filter and sum parameter optimuization," IEICE Electronics Express, Vol. 6, No. 8, pp.437-442, 2009. 

    상세보기 crossref

저자의 다른 논문 :

관련 콘텐츠

오픈액세스(OA) 유형

BRONZE

출판사/학술단체 등이 한시적으로 특별한 프로모션 또는 일정기간 경과 후 접근을 허용하여, 출판사/학술단체 등의 사이트에서 이용 가능한 논문

섹션별 컨텐츠 바로가기

AI-Helper ※ AI-Helper는 오픈소스 모델을 사용합니다.

AI-Helper 아이콘
AI-Helper
안녕하세요, AI-Helper입니다. 좌측 "선택된 텍스트"에서 텍스트를 선택하여 요약, 번역, 용어설명을 실행하세요.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.

선택된 텍스트

맨위로