Speaker Verification with the Constraint of Limited Data
1
조회
0
다운로드
1
인용
0
참고문헌
키워드
초록
Speaker verification system performance depends on the utterance of each speaker. To verify the speaker,important information has to be captured from the utterance. Nowadays under the constraints of limiteddata, speaker verification has become a challenging task. The testing and training data are in terms of fewseconds in limited data. The feature vectors extracted from single frame size and rate (SFSR) analysis is notsufficient for training and testing speakers in speaker verification. This leads to poor speaker modeling duringtraining and may not provide good decision during testing. The problem is to be resolved by increasingfeature vectors of training and testing data to the same duration. For that we are using multiple frame size(MFS), multiple frame rate (MFR), and multiple frame size and rate (MFSR) analysis techniques for speakerverification under limited data condition. These analysis techniques relatively extract more feature vectorduring training and testing and develop improved modeling and testing for limited data. To demonstrate thiswe have used mel-frequency cepstral coefficients (MFCC) and linear prediction cepstral coefficients (LPCC)as feature. Gaussian mixture model (GMM) and GMM-universal background model (GMM-UBM) are usedfor modeling the speaker. The database used is NIST-2003. The experimental results indicate that, improvedperformance of MFS, MFR, and MFSR analysis radically better compared with SFSR analysis. Theexperimental results show that LPCC based MFSR analysis perform better compared to other analysistechniques and feature extraction techniques.
Abstract
Speaker verification system performance depends on the utterance of each speaker. To verify the speaker,important information has to be captured from the utterance. Nowadays under the constraints of limiteddata, speaker verification has become a challenging task. The testing and training data are in terms of fewseconds in limited data. The feature vectors extracted from single frame size and rate (SFSR) analysis is notsufficient for training and testing speakers in speaker verification. This leads to poor speaker modeling duringtraining and may not provide good decision during testing. The problem is to be resolved by increasingfeature vectors of training and testing data to the same duration. For that we are using multiple frame size(MFS), multiple frame rate (MFR), and multiple frame size and rate (MFSR) analysis techniques for speakerverification under limited data condition. These analysis techniques relatively extract more feature vectorduring training and testing and develop improved modeling and testing for limited data. To demonstrate thiswe have used mel-frequency cepstral coefficients (MFCC) and linear prediction cepstral coefficients (LPCC)as feature. Gaussian mixture model (GMM) and GMM-universal background model (GMM-UBM) are usedfor modeling the speaker. The database used is NIST-2003. The experimental results indicate that, improvedperformance of MFS, MFR, and MFSR analysis radically better compared with SFSR analysis. Theexperimental results show that LPCC based MFSR analysis perform better compared to other analysistechniques and feature extraction techniques.
참고문헌 (32)
- A. K. Jain, "An introduction to biometric recognition", IEEE Transactions on Circuits and Systems for Video Technology, 14(1), pp.4~20 (2004)[DOI]
- S. Dey, "Speech biometric based attendance system", Proceedings of 2014 Twentieth National Conference on Communications (NCC),, pp.1~6 (2014)
- T. Kinnunen, "An overview of text-independent speaker recognition: from features to supervectors", Speech Communication, 52(1), pp.12~40 (2010)[DOI]
- G. Pradhan, "Speaker verification under degraded condition: a perceptual study", International Journal of Speech Technology, 14(4), pp.405~417 (2011)[DOI]
- A. E. Rosenberg, "Automatic speaker verification: a review", Proceedings of the IEEE, 64(4), pp.475~487 (1976)[DOI]
- A. Neustein, "Forensic Speaker Recognition" (2012), Springer
- H. S. Jayanna, "Analysis, feature extraction, modeling and testing techniques for speaker recognition", IETE Technical Review, 26(3), pp.181~190 (2009)[DOI]
- H. S. Jayanna, "Limited data speaker recognition", pp.2009
- D. Pati, "Subsegmental, segmental and suprasegmental processing of linear prediction residual for speaker information", International Journal of Speech Technology, 14(1), pp.49~64 (2011)[DOI]
- L. R. Rabiner, "Fundamentals of Speech Recognition" (1993), Prentice Hall
- S. M. Prasanna, "Extraction of speaker-specific excitation information from linear prediction residual of speech", Speech Communication, 48(10), pp.1243~1261 (2006)[DOI]
- B. Yegnanarayana, "Combining evidence from source, suprasegmental and spectral features for a fixed-text speaker verification system", IEEE Transactions on Speech and Audio Processing, 13(4), pp.575~582 (2005)[DOI]
- F. Farahani, "Speaker identification using supra-segmental pitch pattern dynamics", Proceedings of IEEE International Conference on Acoustics, Speech, and Signal Processing, pp.89~92 (2004)
- A. V. Jadhav, "Review of various approaches towards speech recognition", Proceedings of 2012 International Conference on Biomedical Engineering (ICoBE), pp.99~103 (2012)
- H. S. Jayanna, "Multiple frame size and rate analysis for speaker recognition under limited data condition", IET Signal Processing, 3(3), pp.189~204 (2009)[DOI]
- G. L. Sarada, "Multiple frame size and multiple frame rate feature extraction for speech recognition", Proceedings of 2004 International Conference on Signal Processing and Communications, pp.592~595 (2004)
- K. Samudravijaya, "Variable frame size analysis for speech recognition", Proceedings of the International Conference on Natural Language Processing (2004)
- Q. Zhu, "On the use of variable frame rate analysis in speech recognition", Proceedings of 2000 IEEE International Conference on Acoustics, Speech, and Signal Processing, pp.1783~1786 (2000)
- P. Le Cerf, "A new variable frame analysis method for speech recognition", IEEE Signal Processing Letters, 1(12), pp.185~187 (1994)[DOI]
- R. Pawar, "Analysis of FFSR, VFSR, MFSR techniques for feature extraction in speaker recognition: a review", International Journal of Computer Science, 7(4), pp.26~31 (2010)
- T. Nagarajan, "Implicit systems for spoken language identification", pp.2004
- G. S. Ghadiyaram, "Automatic transcription of continuous speech using unsupervised and incremental training", Proceedings of the 8th International Conference on Spoken Language Processing (2004)
- "The NIST Year 2003 speaker recognition evaluation plan" (2013)[URL]
- S. Nakagawa, "Speaker identification and verification by combining MFCC and phase information", IEEE Transactions on Audio, Speech, and Language Processing, 20(4), pp.1085~1095 (2012)[DOI]
- A. Salman, "Speaker verification using boosted cepstral features with Gaussian distributions", Proceedings of IEEE International Multitopic Conference, pp.1~5 (2007)
- D. Pat, "Processing of linear prediction residual in spectral and cepstral domains for speaker information", International Journal of Speech Technology, 18(3), pp.333~350 (2015)[DOI]
- W. C. Hsu, "Usefulness of residual-based features in speaker verification and their combination way with linear prediction coefficients", Proceedings of the 9th IEEE International Symposium on Multimedia Workshops, pp.246~251 (2007)
- S. Furui, "Comparison of speaker recognition methods using statistical features and dynamic features", IEEE Transactions on Acoustics, Speech, and Signal Processing, 29(3), pp.342~350 (1981)[DOI]
- V. Prakash, "In-set", out-of-set speaker recognition under sparse enrollment, 2007, pp.15(7), 2044~2052
- T. Hasan, "A study on universal background model training in speaker verification", IEEE Transactions on Audio, Speech, and Language Processing, 19(7), pp.1890~1899 (2011)[DOI]
- N. Dehak, "Front-end factor analysis for speaker verification", IEEE Transactions on Audio, Speech, and Language Processing, 19(4), pp.788~798 (2011)[DOI]
- E. Wong, "Comparison of linear prediction cepstrum coefficients and mel-frequency cepstrum coefficients for language identification", Proceedings of 2001 International Symposium on Intelligent Multimedia, Video and Speech Processing, pp.95~98 (2001)