The Orthogonal Decomposition and Implementation Algorithm for Speech Signals Based on Reproducing Kernel Space
-
摘要: 提出了一种在Hilbert空间W21[a,b]中对语音信号进行正交分解的方法及其实现算法.利用Hilbert空间W21[a,b]的再生核函数构造一组{φj*(x)}1n标准正交函数组,基于该函数组{φj*(x)}1n对语音信号实施正交分解,再根据W21[a,b]中再生核函数的性质给出了计算正交分解系数的快速算法.该方法将离散的问题影射到连续函数空间中进行处理,同时将Hilbert空间中的内积计算问题转化为函数在离散点的取值问题.实验结果表明,该方法可用于语音信号重建与特征抽取.Abstract: An new orthogonal decomposition method and implementation algorithm for speech signals is proposed in this paper.From the reproducing kernel function of Hilbert space W21 [a,b],a set of normalized orthogonal functions {φj*(x)}1n are generated.Based on {φj*(x)}1n,speech signals can be orthogonally decomposed,and the orthogonal decomposition coefficients can be computed by a fast algorithm based on the properties of reproducing kernel function.This method transforms the discrete problem to continuous function space and convert the inner product computation problem in Hilbert space into function evaluation problem in some discrete points.The experiment results indicate that it can be applied to the reconstruction and feature extraction of speech signals.
-
Keywords:
- speech signal /
- reproducing kernel space /
- orthogonal decomposition /
- signal analysis
-
-
[1] 崔明根,吴勃英.再生核空间数值分析[M].北京:科学出版社,2004. [2] WEI NERTH.Reproducing kernel Hilbert spaces:Application in statistical signal processing[M].New York:HutchinsonRoss Pub Co,1982.
[3] KAILATHT.An RKHS approach to detection and estimation problems-part v:parameter estimation[J].IEEE TransInformation Theory,1973,IT-19(1):29-37.
[4] CHAKRABATTY S,DENG Yun-bin,CAUWENBERGHS G.Robust speech feature extraction by growth transformation inreproducing kernel Hilbert space[C]∥Proc IEEE Int Conf Acoustics Speech and Signal Processing(ICASSP2'004).Canada:Montreal,2004.
[5] KOCSOR A,TOTH L.Kernel-based feature extraction with a speech technology application[J].IEEE Trans.SignalProcessing,2004,52(8):2250-2263.
[6] KOCSOR A,TOTHL.A comparative study of several feature transformation and learning methods for phoneme classification[J].Int Journal of Speech Technology,2000,3(3/4):263-276.
[7] KOCSOR A,TOTH L.Application of kernel-based feature space transformations and learning methods to phonemeclassification[J].Applied Intelligence,2004,21(2):129-142.
[8] TOTH L,KOCSOR A,GOSZTOLYA G.Telephone speech recognition via the combination of knowledge sources in asegmental speech model[J].Acta Cybernetica,2004,16:643-657.
[9] LARKI N F.Optimal approximation in Hilbert space with reproducing kernel function[J].Math Comp,1992,22(4):911-921.
[10] TOTHL,KOCSOR A.Harmonic aternatives to sine-wave speech[C]∥Proc Eurospeech.Switzerland:Geneva,2003.
[11] TOTH L,KOCSOR A.Explicit duration modelling in HMM/ANN hybrids[C]∥Proceedings of TSD 2005.New York:Springer,2005:310-317.
[12] BANHALMI A,KOVACS K,KOCSOR A,et al.Fundamental frequency estimation by least-squares harmonic model fitting[J]∥Proceedings of EuroSpeech.Portugal:Lisbon,2005.
[13] WAHBA G.Reproducing kernel hilbert spaces-two brief reviews[C]∥Proceedings of the 13th IFAC Symposium on SystemIdentification.The Netherlands:Rotterdam,2003.
[14] VAPNIK V N.The nature of statistical learning theory[M].New York:Springer-Verlag,1995.
计量
- 文章访问数: 8
- HTML全文浏览量: 0
- PDF下载量: 7