수정된 MAP 적응 기법을 이용한 음성 데이터 자동 군집화
Automatic Clustering of Speech Data Using Modified MAP Adaptation Technique
0
조회
0
다운로드
1
인용
0
참고문헌
키워드
초록
This paper proposes a speaker and environment clustering method in order to overcome the degradation of the speechrecognition performance caused by various noise and speaker characteristics. In this paper, instead of using the distancebetween Gaussian mixture model (GMM) weight vectors as in the Google’s approach, the distance between the adapted meanvectors based on the modified maximum a posteriori (MAP) adaptation is used as a distance measure for vector quantization(VQ) clustering. According to our experiments on the simulation data generated by adding noise to clean speech, theproposed clustering method yields error rate reduction of 10.6% compared with baseline speaker-independent (SI) model,which is slightly better performance than the Google's approach.
Abstract
This paper proposes a speaker and environment clustering method in order to overcome the degradation of the speechrecognition performance caused by various noise and speaker characteristics. In this paper, instead of using the distancebetween Gaussian mixture model (GMM) weight vectors as in the Google’s approach, the distance between the adapted meanvectors based on the modified maximum a posteriori (MAP) adaptation is used as a distance measure for vector quantization(VQ) clustering. According to our experiments on the simulation data generated by adding noise to clean speech, theproposed clustering method yields error rate reduction of 10.6% compared with baseline speaker-independent (SI) model,which is slightly better performance than the Google's approach.
참고문헌 (15)
- Hilgerk, F., "Quantile based histogram equalization for online applications", Proc. ICSLP, pp.237~240 (2002)
- Moreno, P. J., "A vector Taylor series approach for environment-independent speech recognition", Proc. ICASSP, pp.733~736 (1996)
- Gales, M. J. F., "Robust continuous speech recognition using parallel model combination", IEEE Trans. on Speech and Audio Process, 5(5), pp.352~359 (1996)
- Deng, L., "Recursive estimation of nonstationary noise using iterative stochastic approximation for robust speech recognition", IEEE Trans. Speech Audio Process, 11(6), pp.568~580 (2003)
- Maximum likelihood linear transformations for HMM based speech recognition, "Gales, M. J. F.", Cambridge Univ. Tech. Rep, ~, pp.TR 291 (1997)
- 송화전, "차량 잡음 환경에서 인위적 왜곡 음성을 이용한 Eigenspace-based MLLR에 기반한 고속 화자 적응", 말소리와 음성과학, 1(4), pp.119~125 (2009), 한국음성학회
- Beaufays, F., "Unsupervised discovery and training of maximally dissimilar cluster models", Proc. Interspeech, pp.66~69 (2010)
- Zhang, Y., "An i-vector based approach to training data clustering for improved speech recognition", Proc. Interspeech, pp.1247~1250 (2011)
- Tsao, Y., "An ensemble speaker and speaking environment modeling approach to robust speech recognition", IEEE Trans. Audio, Speech, and Language Processing, 17(5), pp.1025~1037 (2009)
- Lee, C. H., "A study on speaker adaptation of the parameters of continuous density hidden Markov models", IEEE Transactions on Signal Processing, 39(4), pp.806~814 (1991)
- Campbell, W. M., "SVM based speaker verification using a GMM supervector kernel and NAP variability compensation", Proc. ICASSP, 1, pp.97~100 (2006)
- Ban, S. M., "Automatic clustering of speech data using the distance between the cepstral mean vectors", Proc. 2012 Fall Conf. of the Korean Society of Speech Sciences, pp.35~36 (2012)
- Lim, Y., "Implementation of the POW (phonetically optimized words) algorithm for speech database", Proc. ICASSP, 1, pp.89~92 (1995)
- Lee, Y. J., "Some considerations for construction of PBW set", Proc. 12th Workshop on Speech Communications and Signal Processing, pp.310~314 (1995), Korean Association of Speech Sciences
- Sung Joo Lee, "Statistical Model-Based Noise Reduction Approach for Car Interior Applications to Speech Recognition", ETRI Journal, 32(5), pp.801~809 (2010), 한국전자통신연구원