효과적인 특징 벡터를 도입한 Support Vector Machine기반 음성 검출기
A Support Vector Machine-Based Voice Activity Detection Using Effective Feature Vectors
1
조회
0
다운로드
2
인용
0
참고문헌
키워드
초록
본 논문에서는 support vector machine (SVM)을 이용하여 잡음 환경에서의 음성 검출기 (voice activity detection) 성능 향상을 위한 효과적인 특징 벡터를 제안한다. SVM은 통계적 학습 이론으로 훈련 데이터 사이의 최적 분류 초평면을 찾아내 최적화된 이진 분류를 보여준다. 음성 검출을 위한 SVM의 특징 벡터는 음성의 통계 모델에 기반한 음성 검출기에서 주요 파라미터로 사용된 세 가지 특징 벡터인 a posteriori SNR, a priori SNR과 predicted SNR을 결합하여 제시하였다. 도입된 특징 벡터를 적용한 SVM을 이용하여 구현된 음성 검출기의 성능 평가를 위해 G.729B의 음성 검출기와 기존에 제안된 SVM 기반의 음성 검출기와 비교하였으며, 전반적으로 제안된 특징 벡터를 사용했을 때 다양한 잡음 구간에서 향상된 성능을 보였다.
Abstract
In this paper, we propose effective feature vectors to improve the performance of a voice activity detection (VAD) using a support vector machine (SVM) in noisy environments. The SVM incorporates an optimized non-linear decision rule over different sets of feature vectors. To extract novel feature vectors, we present a posteriori SNR, a priori SNR and predicted SNR adopted in a conventional statistical model-based VAD. As a result of experiments, the performance of the SVM-based VAD using novel feature vectors has shown better results compared with the ITU-T G.729B and other recently reported VADs.
참고문헌 (20)
- K. Srinivasant, "Voice activity detection for cellular networks", Proc. IEEE Speech Coding Workshop, pp.85~86 (1993)
- L. R. Rabiner, "Voiced-unvoicedsilence detection using Itakura LPC distance measure", Proc. IEEE Int. Conf. Acoust., Speech, Signal Process, 통권 : 25, pp.323~326 (1977)
- J. D. Hoyt, "Detection of human speech in structured noise", Proc. IEEE Int. Conf. Acoust., Speech, Signal Process, 통권 : 2, pp.237~240 (1994)
- J. C. Junqua, "A study of endpoint detection algorithms in adverse conditions: Incidence on a DTW and HMM recognize", Proc. Eurospeech, pp.1371~1374 (1991)
- J. A. Haigh, "''Robust voice activity detection using cepstral feature", Proc. IEEE TELCON, China, pp.321~324 (1993)
- N. B. Yoma, "Robust speech pulse-detection using adaptive noise modeling", Electron. Lett, 32(15), pp.1350~1352 (1996)
- R. Tucker, "Voice activity detection using a periodicity measure", Proc. Inst. Electr. Eng., 통권 : 139, pp.377~380 (1992)
- F. Beritelli, "a robust voice activity detector for wireless communications using soft computing", IEEE J. Sel. Areas Commun, 16(9), pp.1818~1829 (1998)
- E. Nemer, "Robust voice activity detection using higher-order statistics in the LPC Residual domain", IEEE Trans. Speech Audio Process, 9(3), pp.217~231 (2001)
- "ITU, ''A silence compression scheme for G.729 optimized for terminals conforming to ITU-T V.70,'' ITU-T Rec. G. 729, Annex B", ~ (1996)
- "3GPP2, ''Selectable mode vocoder service option for wide-band spread spectrum communication systems,'' 3GPP2 C.S0030-0 Ver. 1.0", ~ (2001)
- "ETSI, ''Voice activity detector (VAD) for adaptive multi-rate (AMR) speech traffic channels,'' ETSI EN 301 708 v7.1.1", ~ (1999)
- D. Enqing, "Applying support vector machines to voice activity detection", Proc. Int. Conf. Signal Process, 통권 : 2, pp.1124~1127 (2002)
- J. Rami ＇rez, "Speech", Non-speech discrimination combining advanced feature extraction and SVM learning, 2006, 1662~1665
- Y. Ephraim, "Speech enhancement using a minimum mean-square error short-time spectral amplitude estimator", IEEE Trans. Acoustics Speech Signal Process, 32(6), pp.1190~1121 (1984)
- J.-H. Chang, "Voice activity detection based on multiple statistical models", IEEE Trans. Signal Process, 54(6), pp.1965~1976 (2006)
- V. N Vapnik, "An overview of statistical learning theory", IEEE Trans. Neural Networks, 10(5), pp.988~999 (1999)
- N. Cristianini, "An introduction to support vector machines and other kernel-based learning methods" (2000), Cambridge Univ. Press
- J. Sohn, "A Statistical Model- Based Voice Activity Detection", IEEE Sig. Process. Lett, 6(1), pp.1~3 (1999)
- M. Marzinzik, "Speech pause detection for noise spectrum estimation by tracking power envelope dynamics", IEEE Trans. Speech Audio Process, 10(2), pp.341~351 (2002)