Abstract

In the speaker recognition system with large numbers of models, the traditional computations of matching one by one can be very time-consuming. In order to solve the problem of fast recognition, this paper proposes a speaker model clustering method based on space position. The idea is: firstly, to divide the training models into multiple layers, then searches class representatives for every layer, then to cluster the training model by the gotten class representatives. This method can greatly reduce the number of models matching, and achieve the goal of fast recognition. The paper takes the Gaussian mixture model (GMM) as the speech model, the experiment shows the recognition average speed of 0.5s per person, and the correct recognition rate less than 1% loss is ensured.

Keyword

Speaker recognition, Space position, Stratification, Clustering model.

Cite this article

Refference

[1][1]Apsingekar VR, De Leon PL. Speaker model clustering for efficient speaker identification in large population applications. IEEE Transactions on Audio, Speech, and Language Processing. 2009; 17(4):848-53.

[2][2]De Leon PL, Apsingekar V. Reducing speaker model search space in speaker identification. In biometrics symposium 2007 (pp. 1-6). IEEE.

[3][3]Xiao WW, Zheng J, Hua J, Zhan E. Speaker identification based on classification sub-space Gaussian mixture model. In international conference on image analysis and signal processing 2011 (pp. 607-11). IEEE.

[4][4]Sun B, Liu W, Zhong Q. Hierarchical speaker identification using speaker clustering. In proceedings of international conference on natural language processing and knowledge engineering 2003 (pp. 299-304). IEEE.

[5][5]Than K, Ho TB, Nguyen DK. An effective framework for supervised dimension reduction. Neurocomputing. 2014; 139:397-407.

[6][6]Dong A, Chao-qun R, Dan Y, Jiao W. Speaker recognition method based on PSOA clustering and KMP algorithm. Chinese Journal of Scientific Instrument. 2013; 6: 015.

[7][7]Matza A, Bistritz Y. Speaker recognition with rival penalized EM training. In IEEE international workshop on machine learning for signal processing 2011(pp. 1-6). IEEE.

[8][8]Xing Y, Tan P. A novel SVM Kernel with GMM super-vector based on bhattacharyya distance clustering plus within class covariance normalization. In international conference on natural computation (ICNC) 2015 (pp. 47-51). IEEE.

[9][9]Shan ZY, Yang YC. Universal background model reduction based efficient speaker recognition. Journal of Zhejiang University (Engineering Science). 2009; 6: 003.

[10][10]Huaqiao X, Jianbin Z, Enqi Z, Yang W, Jia H. Speaker recognition based on speaker model clustering. Computer Engineering and Applications. 2014, 50(2):133-6.

[11][11]Patnaik M, Mathew A, Gill MS, Pradhan D. FastRec: a fast and robust text independent speaker recognition system for radio networks. In international conference on recent advances and innovations in engineering (ICRAIE) 2014 (pp. 1-7). IEEE.

[12][12]Wang HL, Han JQ, Zheng GB. K-L divergence based model clustering method for fast speaker identification. Pattern Recognition and Artificial Intelligence. 2010; 23(6): 856-61.