화자적응과 군집화를 이용한 화자식별 시스템의 성능 및 속도 향상 = Adaptation and Clustering Method for Speaker Identification with Small Training Data

다국어 초록 (Multilingual Abstract)

One key factor that hinders the widespread deployment of speaker identification technologies is the requirement of long enrollment utterances to guarantee low error rate during identification. To gain user acceptance of speaker identification technologies, adaptation algorithms that can enroll speakers with short utterances are highly essential. To this end, this paper applies MLLR speaker adaptation for speaker enrollment and compares its performance against other speaker modeling techniques: GMMs and HMM. Also, to speed up the computational procedure of identification, we apply speaker clustering method which uses principal component analysis (PCA) and weighted Euclidean distance as distance measurement. Experimental results show that MLLR adapted modeling method is most effective for short enrollment utterances and that the GMMs performs better when long utterances are available.

참고문헌 (Reference)

1 S. Young, "The HTK Book (for HTK version 3.1)" Cambridge University Engineering Department 2001

2 C. Huang, "Speaker selection training for large vocabulary continuous speech recognition" 1 : 609-612, 2002

3 P. C. Woodland, "Speaker adaptation: techniques and challenges" 1 : 85-90, 1999

4 D. A. Reynolds, "Robust text-independent speaker identification using Gaussian mixture speaker models" 3 (3): 72-83, 1995

5 R. Kuhn, "Rapid speaker adaptation in eigenvoice space" 6 : 695-707, 2000

6 C. J. Leggetter, "Maximum likelihood linear regression for speaker adaptation of continuous density hidden Markov models" 9 : 171-185, 1995

7 J. L. Gauvain, "Maximum a-posteriori estimation for multivariate Gaussian mixture observations of Markov chains" 2 (2): 291-298, 1994