프레임간 상관관계의 모델링을 통한 강인한 음성인식
Robust Speech Recognition by Modeling Interframe Correlations
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
본 논문에서는 잡음 환경에서의 강인한 음성인식을 위해 음성의 동적 특성을 활용하는 기법을 제안한다. 음성의 동적특성은 강인한 인식을 위해 중요한 정보로써, 제안하는 기법은 은닉 마르코프 모델(HMM: hidden Markov model)에서 관측 프레임이 서로 독립이라는 가정을 완화하고 관측 프레임간의 상관관계를 가우시안 혼합모델로 모델링하여 동적특성을 반영한다. 제안하는 모델의 정의와 학습기법을 소개하며, 실험을 통해 기존의 HMM이나 특징추출 단계에서 동적특성을 반영하는 기존의 기법에 비해 우수한 성능을 나타냄을 보인다.
Abstract
In this paper, we propose a modeling method to utilize the information of dynamic characteristics of speech for achieving noise-robust speech recognition performance in noisy environments. The information of dynamic characteristics of speech is known to be important for robust recognition. In the proposed method, instead of independently observing each frame, the dependence between frames is explicitly modeled by using the Gaussian mixture model in the framework of hidden Markov models (HMMs). We present the formulation of the proposed model and its learning algorithm. Through the experimental results, we demonstrate that the proposed method shows more robust recognition performance in various noisy conditions than the conventional HMM and the conventional method considering the dynamic speech information at the feature extraction stage.
참고문헌 (17)
- A. L. Gorin, "Automated natural speech dialog,'' IEEE Computer", IEEE Computer, 35(4), pp.51~56 (2002)
- M. A. Walker, "''DARPA communicator evaluation: progress from 2000 to 2001", Proc. Int. Conf. Speech and Language Processing, Denver Colorado, 2002, pp.273~276 (2002)
- S. Goose, "Enhancing web accessibility via the vox portal and a web-hosted dynamic HTML↔VoxML converter", Computer Networks, 33(1), pp.583~592 (2000)
- Y. Gong, "Speech recognition in noisy environments: a survey", Speech Communication, 16, pp.261~291 (1995)
- A. W. F. Huggins, "Temporally segmented speech", Perception and Psychophysics, 18(2), pp.49~157 (1975)
- R. Drullman, "''Effect of temporal envelope smearing on speech recognition", J. Acoustical Society of America, 95(2), pp.1053~1064 (1994)
- L. R. Rabiner, "Fundamentals of Speech Recognition" (1993), Prentice Hall, Inc
- M. J. D. Powell, "A Fortran subroutine for solving systems of nonlinear algebraic equations, in Numerical Methods for Nonlinear Algebraic Equations, chap. 7" (1970), Gordon and Beach, London, UK
- J. E. Dennis, "''Nonlinear least squares, in State of the Art in Numerical Analysis", pp.269~312 (1977), Academic Press
- "The Mathworks, Optimization Toolbox User's Guide" (2005), The Mathworks, Inc.
- A. D. Belegundu, "Optimization Concepts and Applications in Engineering" (1999), Prentice Hall
- H. Hermansky, "RASTA processing of speech", IEEE Trans. Speech and Audio Processing, 2(4), pp.578~589 (1994)
- J.-S. Lee, "Training hidden Markov models by hybrid simulated annealing for visual speech recognition", Proc. IEEE Int. Conf. Systems, Man, and Cybernetics, Taipei, Taiwan, Oct. 2006. (2006)
- R. O. Duda, "Pattern Classification" (2001), John Wiley & Sons
- A. Varga, "Assessment for automatic speech recognition: II. NOISEX-92: A database and an experiment to study the effect of additive noise on speech recognition systems", Speech Communication, 12(3), pp.247~251 (1993)
- S. B. Davis, "Comparison of parametric representations for monosyllable word recognition in continuously spoken sentences", IEEE Trans. Acoustics Speech and Signal Processing, 28(4), pp.357~366 (1980)
- R. M. Stern, "Compensation for environmental degradation in automatic speech recognition", ESCA-NATO Tutorial and Research Workshop on Robust Speech Recognition using Unknown Communication Channels, Pont-a＇-Mousson, France, Apr. 1997, pp.33~42 (1997)