一种结合局部线性嵌入与支持向量机的语音识别方法
|
摘要
语音识别是模式识别领域的重要应用之一。本文提出一种结合局部线性嵌入SLLE(supervised Locally Linear Embedding)与支持向量机support vector machine (SVM)的新型语音识别方法SVM-SLLE算法。SVM-SLLE参考了带监督的局部线性嵌入降维算法SLLE的优点,采用改进的非线性监督距离公式,运用一个常数参数因子α来控制不同类的数据点的距离;通过支持向量机的方式计算最优的局部重构的权重向量w*,使得SVM-SLLE具有最优的归纳学习能力。使用标准的自然语音情感特征数据作为样本数据进行实验,测试结果表明, SVM-SLLE算法降维分类效果明显,语音正确识别率高于常见的SLLE,LLE等方法。
Speech recognition is one of the most important applications in pattern recognition fields. A new locally linear embedding and support vector machine based speech recognition method called SVM-SLLE was presented and described in this paper. SVM-SLLE referred the SLLE and the enhenced supervised distance was used for interchanging the SLLE's supervised distance. A α constant factor was used to control the distance of different class data points and it is close to or less than that of the same class data points.With the aim to possess the optimal generalization ability and optimum weight vector w*, support vector machine(SVM) algorithm was also adopted in SVM-SLLE. A seriers of experiments were done for testing SVM-SLLE using the standard natural Speech emotion database. The good dimensionality reduction effect had been also obtained and the recognition accuracy is better than other dimensionality reduction algorithm such as SLLE, LLE and so on.
Speech recognition is one of the most important applications in pattern recognition fields. A new locally linear embedding and support vector machine based speech recognition method called SVM-SLLE was presented and described in this paper. SVM-SLLE referred the SLLE and the enhenced supervised distance was used for interchanging the SLLE's supervised distance. A α constant factor was used to control the distance of different class data points and it is close to or less than that of the same class data points.With the aim to possess the optimal generalization ability and optimum weight vector w*, support vector machine(SVM) algorithm was also adopted in SVM-SLLE. A seriers of experiments were done for testing SVM-SLLE using the standard natural Speech emotion database. The good dimensionality reduction effect had been also obtained and the recognition accuracy is better than other dimensionality reduction algorithm such as SLLE, LLE and so on.
引文
[1] 文仕学,孙磊,杜俊. 渐进学习语音增强方法在语音识别中的应用[J]. 小型微型计算机系统, 2018, 39(1):1-6.
[2] TABATABAEI T S, KRISHNAN S. Towards robust speech-based emotion recognition[C]. IEEE International Conference on Systems Man & cybernetics, 2010:608-611.
[3] YU Q, LI Y, JIA P.Speech emotion recognition using supervised manifold learning based on all-class and pairwise-class feature extraction[C]. 2011 IEEE International Conference on Intelligent Computing and Integrated Systems (ICISS), 2013:1-8.
[4] KORENO Y, CARMEL L. Robust linear dimensionality reduction[J]. IEEE Transactions on Visualization and Computer Graphics, 2012, 10(4):459-470.
[5] JOLLIFFE I T. Principal component analysis[M]. Springer-Verlag, New York, 2002.
[6] BELHUMEOUR P, HESPANHA J, KRIEGMAN D. Eigenfaces versus fisherfaces: recognition using class specific linear projection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007, 19(7):711-720.
[7] SCHOLKOPF B. SMOLA A J, MULLER K R. Nonlinear component analysis as a kernel eigenvalue problem[J]. Neural Computation, 1998, 10(5):1299-1319.
[8] ROWEIS S T, SAUL L K. Nonlinear dimensionality reduction by locally linear embedding[J]. Science 2000, 290(5500):2323-2326.
[9] TENENBAUM J B, SILVA V DE, LANGFPR J C. A global intrinsic framework for nonlinear dimensionality reduction[J]. Science, 2000, 290(5500):2319-2323.
[10] BELKIN M, NIYOGI P. Laplacian eigenmaps for dimensionality reduction and representation[J]. Neural Computation, 2003, 15:1373-1396.
[11] HE X, NIYOGI P. Locality preserving projections[J]. Advances in Neural Information Processing System, 2004, 16:153-160.
[12] ZHANG Z, ZHA H. Principal manifolds and nonlinear dimension reduction via local tangent space alignment, SIAM Journal of Scientific Computing, 2005, 26(1):313-338.
[13] DICK D R, OLGA KOUROPTEVA, OLEG OKUN, et al. Supervised Locally Linear Embedding[C]. Artificial Neural Networks and Neural Information Processing-ICANN/ICONIP2003 Springer, 2003, Lecture Notes in computer science, 2714, 333-341.
[14] 张石清,李乐民,赵知劲. 基于一种改进的监督流形学习算法的语音情感识别[J]. 电子与信息学报, 2010, 32(11):2724-2729.
[15] D L DONOHO, C GRIMES. Hessian eigenmaps: Locally linear embedding, techniques for high-dimensional data[J]. PNAS, 2003, 100(10):5591-5596.
[16] 刘方原,夏克文, 牛文佳. 改进重构权值的局部线性嵌入算法[J]. 中国图象图形学报, 2018, 23(1):52-60.
[17] C ZHAO, C WANG, L HUA, et al. Recognition of Control Chart Pattern Using Improved Supervised Locally Linear Embedding and Support Vector Machine[J]. Procedia Engineering, 2017, 174:281-288.
[18] LIN F, YEH C C, LEE M Y. A hybrid business failure prediction model using locally linear embedding and support vector machines[J]. Journal for Economic Forecasting, 2013, 16(1):82-97.
[2] TABATABAEI T S, KRISHNAN S. Towards robust speech-based emotion recognition[C]. IEEE International Conference on Systems Man & cybernetics, 2010:608-611.
[3] YU Q, LI Y, JIA P.Speech emotion recognition using supervised manifold learning based on all-class and pairwise-class feature extraction[C]. 2011 IEEE International Conference on Intelligent Computing and Integrated Systems (ICISS), 2013:1-8.
[4] KORENO Y, CARMEL L. Robust linear dimensionality reduction[J]. IEEE Transactions on Visualization and Computer Graphics, 2012, 10(4):459-470.
[5] JOLLIFFE I T. Principal component analysis[M]. Springer-Verlag, New York, 2002.
[6] BELHUMEOUR P, HESPANHA J, KRIEGMAN D. Eigenfaces versus fisherfaces: recognition using class specific linear projection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007, 19(7):711-720.
[7] SCHOLKOPF B. SMOLA A J, MULLER K R. Nonlinear component analysis as a kernel eigenvalue problem[J]. Neural Computation, 1998, 10(5):1299-1319.
[8] ROWEIS S T, SAUL L K. Nonlinear dimensionality reduction by locally linear embedding[J]. Science 2000, 290(5500):2323-2326.
[9] TENENBAUM J B, SILVA V DE, LANGFPR J C. A global intrinsic framework for nonlinear dimensionality reduction[J]. Science, 2000, 290(5500):2319-2323.
[10] BELKIN M, NIYOGI P. Laplacian eigenmaps for dimensionality reduction and representation[J]. Neural Computation, 2003, 15:1373-1396.
[11] HE X, NIYOGI P. Locality preserving projections[J]. Advances in Neural Information Processing System, 2004, 16:153-160.
[12] ZHANG Z, ZHA H. Principal manifolds and nonlinear dimension reduction via local tangent space alignment, SIAM Journal of Scientific Computing, 2005, 26(1):313-338.
[13] DICK D R, OLGA KOUROPTEVA, OLEG OKUN, et al. Supervised Locally Linear Embedding[C]. Artificial Neural Networks and Neural Information Processing-ICANN/ICONIP2003 Springer, 2003, Lecture Notes in computer science, 2714, 333-341.
[14] 张石清,李乐民,赵知劲. 基于一种改进的监督流形学习算法的语音情感识别[J]. 电子与信息学报, 2010, 32(11):2724-2729.
[15] D L DONOHO, C GRIMES. Hessian eigenmaps: Locally linear embedding, techniques for high-dimensional data[J]. PNAS, 2003, 100(10):5591-5596.
[16] 刘方原,夏克文, 牛文佳. 改进重构权值的局部线性嵌入算法[J]. 中国图象图形学报, 2018, 23(1):52-60.
[17] C ZHAO, C WANG, L HUA, et al. Recognition of Control Chart Pattern Using Improved Supervised Locally Linear Embedding and Support Vector Machine[J]. Procedia Engineering, 2017, 174:281-288.
[18] LIN F, YEH C C, LEE M Y. A hybrid business failure prediction model using locally linear embedding and support vector machines[J]. Journal for Economic Forecasting, 2013, 16(1):82-97.