informИнформатикаInformatics1816-03012617-6963UIIP NASBinform-3Research ArticleОБРАБОТКА СИГНАЛОВ, ИЗОБРАЖЕНИЙ, РЕЧИ, ТЕКСТА И РАСПОЗНАВАНИЕ ОБРАЗОВSIGNAL, IMAGE, SPEECH, TEXT PROCESSING AND PATTERN RECOGNITIONМЕТОД ОБЪЕДИНЕНИЯ РЕШЕНИЙ КЛАССИФИКАТОРОВ ДЛЯ ЗАДАЧИ РАСПОЗНАВАНИЯ ДИКТОРАMERGING CLASSIFIER DECISIONS FOR SPEAKER RECOGNITIONИмамвердиевЯ. Н.ImamverdiyevY. N.yadigar@lan.ab.azСухостатЛ. В.SukhostatL. V.lsuhostat@hotmail.comИнститут информационных технологий Национальной академии наук АзербайджанаRussian Federation201525092016011725Copyright © Имамвердиев Я.Н., Сухостат Л.В., 20162016Имамвердиев Я.Н., Сухостат Л.В.Imamverdiyev Y.N., Sukhostat L.V.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://inf.grid.by/jour/article/view/3

Предлагается использование нечетких интегралов для объединения решений классификаторовсистем распознавания диктора. В качестве набора признаков рассматриваются мгновенная частота и мгновенная амплитуда. Предлагаемый метод показывает значительно лучшие результаты по сравнению с применением единственного классификатора. Проводится сравнение предлагаемого метода с другими методами объединения решений классификаторов.

The paper proposes using fuzzy integrals for merging classifier decisions in speaker recognitionsystems. Instantaneous frequency and instantaneous amplitude are considered as the set of features. The approach shows significantly better results than a single classifier. A comparison of the proposed approach with the other methods for merging classifier decisions is provided.

ReferencesRoss, A.A. Handbook of Multibiometrics / A.A. Ross, K. Nandakumar, A.K. Jain. – London : Springer, 2006. – 198 p.Ross, A.A. Handbook of Multibiometrics / A.A. Ross, K. Nandakumar, A.K. Jain. – London : Springer, 2006. – 198 p.Solomonoff, A. Advances in channel compensation for SVM speaker recognition / A. Solomonoff, W. Campbell, I. Boardman // Proc. of ICASSP. – Philadelphia, PA, 2005. – P. 629– 632.Solomonoff, A. Advances in channel compensation for SVM speaker recognition / A. Solomonoff, W. Campbell, I. Boardman // Proc. of ICASSP. – Philadelphia, PA, 2005. – P. 629– 632.Gader, P.D. Fusion of handwritten word classifiers / P.D. Gader, M.A. Mohamed, J.M. Keller //Pattern Recognition Letters. – 1996. – № 17. – P. 577–584.4. Michel, G. The representation of importance and interaction of features by fuzzy measure /G. Michel // Pattern Recognition Letters. – 1996. – № 17. – P. 567–575.Gader, P.D. Fusion of handwritten word classifiers / P.D. Gader, M.A. Mohamed, J.M. Keller //Pattern Recognition Letters. – 1996. – № 17. – P. 577–584.4. Michel, G. The representation of importance and interaction of features by fuzzy measure /G. Michel // Pattern Recognition Letters. – 1996. – № 17. – P. 567–575.Kuncheva, L.I. Decision templates for multiple classifier fusion: an experimental comparison / L.I. Kuncheva, J.C. Bezdek, R.P.W. Duin // Pattern Recognition. – 2001. – № 34. – P. 299–314.Kuncheva, L.I. Decision templates for multiple classifier fusion: an experimental comparison / L.I. Kuncheva, J.C. Bezdek, R.P.W. Duin // Pattern Recognition. – 2001. – № 34. – P. 299–314.Mirhosseini, A.R. Human face image recognition: an evidence aggregation approach / A.R. Mirhosseini, H. Yan // Computer Vision and Image Understanding. – 1998. – № 71. – P. 213– 230.Mirhosseini, A.R. Human face image recognition: an evidence aggregation approach / A.R. Mirhosseini, H. Yan // Computer Vision and Image Understanding. – 1998. – № 71. – P. 213– 230.Pham, T.D. Color image segmentation using fuzzy integral and mountain clustering / T.D. Pham, H. Yan // Fuzzy sets and systems. – 1999. – № 107. – P. 121–130.Pham, T.D. Color image segmentation using fuzzy integral and mountain clustering / T.D. Pham, H. Yan // Fuzzy sets and systems. – 1999. – № 107. – P. 121–130.Kwak, K.-C. Face recognition using fuzzy integral and wavelet decomposition method / K.- C. Kwak, W. Pedrycz // IEEE Transactions on Systems, Man, and Cybernetics. – 2004. – № 34. – P. 1666–1675.Kwak, K.-C. Face recognition using fuzzy integral and wavelet decomposition method / K.- C. Kwak, W. Pedrycz // IEEE Transactions on Systems, Man, and Cybernetics. – 2004. – № 34. – P. 1666–1675.Auephanwiriyakul, S. Generalized Choquet fuzzy integral fusion / S. Auephanwiriyakul, M.K. James, P.D. Gader // Information Fusion. – 2002. – № 3. – P. 69–85.Auephanwiriyakul, S. Generalized Choquet fuzzy integral fusion / S. Auephanwiriyakul, M.K. James, P.D. Gader // Information Fusion. – 2002. – № 3. – P. 69–85.Wolf, J.J. Efficient acoustic parameters for speaker recognition / J.J. Wolf // J. Acoustical Society of America. – 1982. – Vol. 51, № 6. – P. 2044–2056.Wolf, J.J. Efficient acoustic parameters for speaker recognition / J.J. Wolf // J. Acoustical Society of America. – 1982. – Vol. 51, № 6. – P. 2044–2056.Kinnunen, T. An overview of text-independent speaker recognition: from features to supervectors / T. Kinnunen, H. Li // Speech Communication. – 2010. – Vol. 52, № 1. – P. 12– 40.Kinnunen, T. An overview of text-independent speaker recognition: from features to supervectors / T. Kinnunen, H. Li // Speech Communication. – 2010. – Vol. 52, № 1. – P. 12– 40.Rose, P. Forensic speaker identification. Taylor & Francis forensic science series / P. Rose. N.Y. : Taylor & Francis, 2002. – 380 p.Rose, P. Forensic speaker identification. Taylor & Francis forensic science series / P. Rose. N.Y. : Taylor & Francis, 2002. – 380 p.Kinnunen, T. Spectral features for automatic text-independent speaker recognition. Licentiate thesis / T. Kinnunen. – Finland : University of Joensuu, 2003.Kinnunen, T. Spectral features for automatic text-independent speaker recognition. Licentiate thesis / T. Kinnunen. – Finland : University of Joensuu, 2003.Маркел, Дж. Линейное предсказание речи / Дж. Маркел, А.X. Грей. – М. : Связь, 1980. – 308 с.Маркел, Дж. Линейное предсказание речи / Дж. Маркел, А.X. Грей. – М. : Связь, 1980. – 308 с.Furui, S. Cepstral analysis techniques for automatic speaker verification / S. Furui // IEEE tran. acoust., speech, signal processing. – 1981. – Vol. 27. – P. 254–272.Furui, S. Cepstral analysis techniques for automatic speaker verification / S. Furui // IEEE tran. acoust., speech, signal processing. – 1981. – Vol. 27. – P. 254–272.Reynolds, D. Channel robust speaker verification via feature mapping / D. Reynolds // Proc. of ICASSP. – Hong Kong, 2003. – Vol. 2. – P. 53–56.Reynolds, D. Channel robust speaker verification via feature mapping / D. Reynolds // Proc. of ICASSP. – Hong Kong, 2003. – Vol. 2. – P. 53–56.Doddington, G. Speaker recognition based on idiolectal differences between speakers / G. Doddington // Proc. of Eurospeech. – Aalborg, Denmark, 2001. – Vol. 4. – P. 2521–2524.Doddington, G. Speaker recognition based on idiolectal differences between speakers / G. Doddington // Proc. of Eurospeech. – Aalborg, Denmark, 2001. – Vol. 4. – P. 2521–2524.Hemant, A.P. Forensic Speaker Recognition / A.P. Hemant, Amy Neustein. – Heidelberg : Springer, 2012. – 540 p.Hemant, A.P. Forensic Speaker Recognition / A.P. Hemant, Amy Neustein. – Heidelberg : Springer, 2012. – 540 p.Benediktsson, J.A. Consensus theoretic classification methods / J.A. Benediktsson, P.H. Swain // IEEE Trans. Systems Man Cybernet. – 1992. – № 22. – P. 688–704.Benediktsson, J.A. Consensus theoretic classification methods / J.A. Benediktsson, P.H. Swain // IEEE Trans. Systems Man Cybernet. – 1992. – № 22. – P. 688–704.Ho, T.K. Decision combination in multiple classifier systems / T.K. Ho, J.J. Hull, S.N. Srihari // IEEE Trans. Pattern Anal. Machine Intelligence. – 1994. – № 16. – P. 66–75.Ho, T.K. Decision combination in multiple classifier systems / T.K. Ho, J.J. Hull, S.N. Srihari // IEEE Trans. Pattern Anal. Machine Intelligence. – 1994. – № 16. – P. 66–75.Xu, L. Methods of combining multiple classifiers and their applications to hand-written character recognition / L. Xu, A. Krzyzak, C.Y. Suen // IEEE Trans. Systems Man Cybernet. – 1992. – № 23. – P. 418–435.Xu, L. Methods of combining multiple classifiers and their applications to hand-written character recognition / L. Xu, A. Krzyzak, C.Y. Suen // IEEE Trans. Systems Man Cybernet. – 1992. – № 23. – P. 418–435.Soong, F.K. On the use of instantaneous and transitional spectral information in speaker recognition / F.K. Soong, A.E. Rosenberg // IEEE Trans. Acoust. Speech, Signal Process. – 1988. – ASSP-36. – P. 871–879.Soong, F.K. On the use of instantaneous and transitional spectral information in speaker recognition / F.K. Soong, A.E. Rosenberg // IEEE Trans. Acoust. Speech, Signal Process. – 1988. – ASSP-36. – P. 871–879.Farrell, K.R. Text-dependent speaker verification using data fusion / K.R. Farrell // IEEE Intern. Conf. on Acoustic, Speech and Signal Processing. – Detroit, Michigan, USA, 1995. – P. 349–352.Farrell, K.R. Text-dependent speaker verification using data fusion / K.R. Farrell // IEEE Intern. Conf. on Acoustic, Speech and Signal Processing. – Detroit, Michigan, USA, 1995. – P. 349–352.Sub-word speaker verification using data fusion methods / K.R. Farrell [et al.] // IEEE Workshop on Neural Networks for Signal Processing. – Amelia Island, Florida, 1997. – P. 531–540.Sub-word speaker verification using data fusion methods / K.R. Farrell [et al.] // IEEE Workshop on Neural Networks for Signal Processing. – Amelia Island, Florida, 1997. – P. 531–540.Farrell, K.R. An analysis of data fusion methods for speaker verification / K.R. Farrell, R.P. Ramachandran, R.J. Mammone // IEEE Intern. Conf. on Acoustic, Speech and Signal Processing. – Washington, USA, 1998. – P. 1129–1132.Farrell, K.R. An analysis of data fusion methods for speaker verification / K.R. Farrell, R.P. Ramachandran, R.J. Mammone // IEEE Intern. Conf. on Acoustic, Speech and Signal Processing. – Washington, USA, 1998. – P. 1129–1132.Schalkwyk, J. Speaker verification with low storage requirements / J. Schalkwyk, N. Jain, E. Barnard // IEEE Intern. Conf. on Acoustic, Speech and Signal Processing. – Georgia, USA, 996. – P. 693–696.Schalkwyk, J. Speaker verification with low storage requirements / J. Schalkwyk, N. Jain, E. Barnard // IEEE Intern. Conf. on Acoustic, Speech and Signal Processing. – Georgia, USA, 996. – P. 693–696.Zadeh, L.A. Fuzzy sets / L.A. Zadeh // Information and Control. – 1965. – № 8. – P. 338– 353.Zadeh, L.A. Fuzzy sets / L.A. Zadeh // Information and Control. – 1965. – № 8. – P. 338– 353.Gupta, M.M. Fuzzy measures and fuzzy integrals / M.M. Gupta, G.N. Saridis, B.R. Gaines. – N.Y. : Elsevier, 1977. – 510 p.29. Murofushi, T. A theory of fuzzy measures. Representation, the Choquet integral and null sets / T. Murofushi, M. Sugeno // J. Math. Anal. Appl. – 1991. – Vol. 159, № 2. – P. 532–549.Gupta, M.M. Fuzzy measures and fuzzy integrals / M.M. Gupta, G.N. Saridis, B.R. Gaines. – N.Y. : Elsevier, 1977. – 510 p.29. Murofushi, T. A theory of fuzzy measures. Representation, the Choquet integral and null sets / T. Murofushi, M. Sugeno // J. Math. Anal. Appl. – 1991. – Vol. 159, № 2. – P. 532–549.Maragos, P. On amplitude and frequency demodulation using energy operators / P. Maragos, J.F. Kaiser, T.F. Quatieri // IEEE Trans. on Signal Processing. – 1993. – Vol. 41, № 4. – P. 1532–1550.Maragos, P. On amplitude and frequency demodulation using energy operators / P. Maragos, J.F. Kaiser, T.F. Quatieri // IEEE Trans. on Signal Processing. – 1993. – Vol. 41, № 4. – P. 1532–1550.Zhang, W.D. A priori threshold determination for phrase-prompted speaker verification / W.D. Zhang [et al.] // Proc. Eurospeech’99. – Budapest, Hungary, 1999. – P. 1203–1206.Zhang, W.D. A priori threshold determination for phrase-prompted speaker verification / W.D. Zhang [et al.] // Proc. Eurospeech’99. – Budapest, Hungary, 1999. – P. 1203–1206.Hamid, L.A. Quality based Speaker Verification Systems using Fuzzy Inference Fusion Scheme / L.A. Hamid, D.A. Ramli // Proc. of the Intern. Conf. on Communications, Signal Processing and Computers. – Interlaken, Switzerland, 2014. – P. 96–103.Hamid, L.A. Quality based Speaker Verification Systems using Fuzzy Inference Fusion Scheme / L.A. Hamid, D.A. Ramli // Proc. of the Intern. Conf. on Communications, Signal Processing and Computers. – Interlaken, Switzerland, 2014. – P. 96–103.

The authors declare that there are no conflicts of interest present.