非线性动力学方法在声纹分析中的应用
|
摘要
随着信息技术和网络通信的发展，信息安全成为一个信息时代的新课题日益被人们所关注，每个人所特有的声纹，可以作为鉴别身份的一种重要手段。如今，声纹识别技术研究虽然取得了飞速的发展，但是声纹识别技术还不理想，主要是由于声音信号是非线性时间序列，对于非线性系统人们还没有强有力的理论分析工具，复杂性的研究逐渐引起人们的关注，复杂性测度在非线性系统分析重取得了可喜的成果。
以前人们往往用线性的方法如频谱分析来分析语音信号，而这些线性方法只适用于平稳的、一致的、平衡的线性的时间序列，对于非平稳、不一致、非平衡的的非线性的语音时间序列，这些传统的线性方法就往往丢掉了许多蕴涵本质的重要信息。因此我们尝试了用复杂性方法来分析每个人讲话的语音特征，发现即使相同的一段话，不同受试者的语音特征遽然不同。
本文运用多种复杂性算法和统计的理论分析了大量的语音信号。对几种复杂性测度作了较详细的论述及分析比较，针对目前算法存在的缺陷，提出改进算法——分测度复杂性，并验证其有效性。我们认为，复杂性分析方法能够用于声纹的特征分析，具有宽广的应用前景，所以这是很有前途的研究课题。
Along with the development of information technology and network communications, information security which become a new problem of the Information Age is cared increasingly by people. Like dactylogram, speech character can be used to identify different speakers. Today speech recognition technology have been obtained great development but it is not ideal still. The mam reason is that speech signal is a nonlinear time serial and lacks a useful theoretic tool to analyze it .In recent years, the concept of complexity provoked interest; and applying complexity to nonlinear system analysis has been making gratifying achievements.
Usually linear techniques such as spectrum analysis were used to analyze the speech signal. As these traditional linear techniques are merely applicable to the steady, coherent and balanced linear time series, they generally ignored the most important information, which contained the essence of the unsteady, incoherent and unbalanced nonlinear time series of speech. So we try to analyze the speech character using complexity, we find the complexity for speech character of individual testee differs with others even if testees speak
the same sentences.
We have applied many complexity measures and theory of statistics to analyzing many speech signal. Had analyzed these algorithms, we bring forward a new complexity measure: partition measure complexity to overcome the disadvantage of those older methods. We had tested its advantage through the test. We conclude that this analysis method for tone texture can be applied to analyze speech character and this technique has wide application prospect.
以前人们往往用线性的方法如频谱分析来分析语音信号，而这些线性方法只适用于平稳的、一致的、平衡的线性的时间序列，对于非平稳、不一致、非平衡的的非线性的语音时间序列，这些传统的线性方法就往往丢掉了许多蕴涵本质的重要信息。因此我们尝试了用复杂性方法来分析每个人讲话的语音特征，发现即使相同的一段话，不同受试者的语音特征遽然不同。
本文运用多种复杂性算法和统计的理论分析了大量的语音信号。对几种复杂性测度作了较详细的论述及分析比较，针对目前算法存在的缺陷，提出改进算法——分测度复杂性，并验证其有效性。我们认为，复杂性分析方法能够用于声纹的特征分析，具有宽广的应用前景，所以这是很有前途的研究课题。
Along with the development of information technology and network communications, information security which become a new problem of the Information Age is cared increasingly by people. Like dactylogram, speech character can be used to identify different speakers. Today speech recognition technology have been obtained great development but it is not ideal still. The mam reason is that speech signal is a nonlinear time serial and lacks a useful theoretic tool to analyze it .In recent years, the concept of complexity provoked interest; and applying complexity to nonlinear system analysis has been making gratifying achievements.
Usually linear techniques such as spectrum analysis were used to analyze the speech signal. As these traditional linear techniques are merely applicable to the steady, coherent and balanced linear time series, they generally ignored the most important information, which contained the essence of the unsteady, incoherent and unbalanced nonlinear time series of speech. So we try to analyze the speech character using complexity, we find the complexity for speech character of individual testee differs with others even if testees speak
the same sentences.
We have applied many complexity measures and theory of statistics to analyzing many speech signal. Had analyzed these algorithms, we bring forward a new complexity measure: partition measure complexity to overcome the disadvantage of those older methods. We had tested its advantage through the test. We conclude that this analysis method for tone texture can be applied to analyze speech character and this technique has wide application prospect.
引文
1．约翰．霍甘，复杂性研究的发展趋势——从复杂性到困惑，科学(重庆)，1995(10)：42—47
2．郝柏林著，从抛物线谈起—混沌动力学引论，上海科技教育出版社，1993
3．谢惠民：复杂性与动力系统 上海科技教育出版社 1994
4．周作领：符号动力系统 上科技教育出版社 1997
5．张贤达 著，时间序列分析，清华大学出版社，1996．6
6．周迪伟等译，计算机语音处理，国防工业出版社，1987
7．胡航 编著，语音信号处理，哈尔滨工业大学出版社，2000．5
8．语音识别基本原理，清华大学出版社，1999
9．童勤业 等，脑电复杂性分析的新方法，中国生物医学工程学报，1998，Vol．17
10．曹业敏 等，说话人识别技术的现状与进展，河南科技．1998，No．9
11．赵铮 等，基于小波变换说话人识别技术的研究，西安电子科技大学学报．2000，Vol27，No．4
12．叶明等，LPC倒谱参数说话人特征分析，南京航空航天大学学报，1994 Vol．26，No．6
13．杨澄宇 等，基于高斯混合模型的说话人确认系统，计算机应用．2001，Vol．21，No．4
14．邓浩江 等，基于聚类统计与文本无关的说话认识别研究，电路与系统学报．2001，6(3)．-77-80
15．黄南华 等，基于特征空间轨迹多模板匹配说话认识别研究，计算机工程．2001，27(5)．-126-128
16．甄斌 等，语音识别和说话人识别中各倒谱分量的相对重要性，北京大学学报．2001，37(3)．-371-378
17．岳喜才 等，用神经阵列网络进行文本无关的说话人识别，声学学报，2000，25(3)．-230-234
18．周连山、庄显义著．液压系统的计算机仿真 国防工业出版社
19. M.Casdagli, Nonlinear prediction of chaotic time series ,physica D 35(1989):335-355,
20. Remo Badii "Complexity---Hierarchical structures and scaling in physics" Cambridge Nonlinear Science Series
21. F. Phan, et. al. "Speaker identification using neural networks and wavelets" IEEE Eng. in Med. & Bio. Magazine pp.92-101, Vol. 19, No. 1 Feb. 2000
22. R.Auckenthaler, M.J.Carey, J.S.D.MaSon. "Language dependency in text_independent speaker verification" IEEE Eng. In Med & Bio.Magazine pp441_444,Vol. 1, No. 4 2001
23. N. Pradhan and P. K. Sadasivan. Validity of Dimension complexity measures of EEG signals. Bifurcation and chaos 7(1997) 173-186
24. "Speaker Verification" Biometric Technology Today. Judy/August 2001
25. Gustavo Deco and Christian Schittenkopf et al, Information flow in chaotic symbolic dynamics for finite and infinitesimal resolution. Bifurcation and chaos 7(1997) 97-105
26. Ki H.Chon and Jorgen K.Kanters et al. Detection of chaotic determinism in time series from randomly forced maps, Physica D 99(1997)471-486
27. Lempel and J.Ziv, On the Complexity of finite sequences, IEEE trans inform theory 22(1976)75-88
28. F. Phan, et. al. "Speaker identification using neural networks and wavelets" IEEE Eng. in Med. & Bio. Magazine pp.92-101, Vol. 19, No. 1 Feb. 2000
29. W. S. Liming, et. al. ""Smoothness" in embedded time-series" Chaos in Communications,
SPIE. Vol.2038 pp.242-251, 1993
30. W. S. Liming, et. al. ""Smoothness" in embedded time-series" Chaos in Communications, SPIE. Vol.2038 pp.242-251, 1993
31. Peter Grassberger, Itamar Procaccia: Estimation of the Kolmogorov entropy from a chaotic signal; PHYSICAL REVIEW A,Vol.28,No.4,Oct. 1983(2591-2593)
28. Alan WOLF, et.: DETERMINING LYAPUNOV EXPONENTS FROM A TIME SERIES; Physica 16D(1985)285-317
29. Aviad Cohen and Itamar Procaccia: Computing the Kolmogorov entropy from time signals of dissipative and conservation dynamical systems; PHYSICAL REVIEW A,Vol.31, No.3, Mar. 1985 (1872-1882)
30. N.H.Packard and W. Lutzenberger et al Geometry from a time series. Phy. Rev .Let 45(1980) 721-715
31. N.H.Pachard, J.P. Crutchfield, et.: Germetry from a Time Series; PHYSICAL REVIEW LETTERS,Vol.45, No.9, 1980(712-716)
32. Peter Grassberger: Toward a Quantitative Theory of Self-Generated Complexity; Int. Journal of Theroerical Physics, Vol.25, No.9, 1986, p907-938
33. Steven M. Pincus- Approximate entropy as a measure of system complexity; Proc. Natl. Acad. Sci. USA,Vol.88, pp.2297-2301, Mar. 1991
34. D.S.Broomhead and Gregory P.King. Extracting qualitative dynamics from experimental data. Phys.D 20 (1986) 217-236
35. Soong A C K et al, Biol Cyber 62 (1989) 55-62
36. Rosechke j. et al, Biol Cybern 64(1991) 307-313
37. J.E.Bates, H.K. Sheperd, information fluctuation as a measure of complexity University of New Hampshire, Durham (1991)
38. J.j. wright Biol Cybeern 76 (1997) 181_194 EEG simulation: variation of spectral envelope, pulse synchrony and ≈40 HZ oscillation. Mental Health Research Institute of Victoria Parkville, Victoria 3052 Australia
2．郝柏林著，从抛物线谈起—混沌动力学引论，上海科技教育出版社，1993
3．谢惠民：复杂性与动力系统 上海科技教育出版社 1994
4．周作领：符号动力系统 上科技教育出版社 1997
5．张贤达 著，时间序列分析，清华大学出版社，1996．6
6．周迪伟等译，计算机语音处理，国防工业出版社，1987
7．胡航 编著，语音信号处理，哈尔滨工业大学出版社，2000．5
8．语音识别基本原理，清华大学出版社，1999
9．童勤业 等，脑电复杂性分析的新方法，中国生物医学工程学报，1998，Vol．17
10．曹业敏 等，说话人识别技术的现状与进展，河南科技．1998，No．9
11．赵铮 等，基于小波变换说话人识别技术的研究，西安电子科技大学学报．2000，Vol27，No．4
12．叶明等，LPC倒谱参数说话人特征分析，南京航空航天大学学报，1994 Vol．26，No．6
13．杨澄宇 等，基于高斯混合模型的说话人确认系统，计算机应用．2001，Vol．21，No．4
14．邓浩江 等，基于聚类统计与文本无关的说话认识别研究，电路与系统学报．2001，6(3)．-77-80
15．黄南华 等，基于特征空间轨迹多模板匹配说话认识别研究，计算机工程．2001，27(5)．-126-128
16．甄斌 等，语音识别和说话人识别中各倒谱分量的相对重要性，北京大学学报．2001，37(3)．-371-378
17．岳喜才 等，用神经阵列网络进行文本无关的说话人识别，声学学报，2000，25(3)．-230-234
18．周连山、庄显义著．液压系统的计算机仿真 国防工业出版社
19. M.Casdagli, Nonlinear prediction of chaotic time series ,physica D 35(1989):335-355,
20. Remo Badii "Complexity---Hierarchical structures and scaling in physics" Cambridge Nonlinear Science Series
21. F. Phan, et. al. "Speaker identification using neural networks and wavelets" IEEE Eng. in Med. & Bio. Magazine pp.92-101, Vol. 19, No. 1 Feb. 2000
22. R.Auckenthaler, M.J.Carey, J.S.D.MaSon. "Language dependency in text_independent speaker verification" IEEE Eng. In Med & Bio.Magazine pp441_444,Vol. 1, No. 4 2001
23. N. Pradhan and P. K. Sadasivan. Validity of Dimension complexity measures of EEG signals. Bifurcation and chaos 7(1997) 173-186
24. "Speaker Verification" Biometric Technology Today. Judy/August 2001
25. Gustavo Deco and Christian Schittenkopf et al, Information flow in chaotic symbolic dynamics for finite and infinitesimal resolution. Bifurcation and chaos 7(1997) 97-105
26. Ki H.Chon and Jorgen K.Kanters et al. Detection of chaotic determinism in time series from randomly forced maps, Physica D 99(1997)471-486
27. Lempel and J.Ziv, On the Complexity of finite sequences, IEEE trans inform theory 22(1976)75-88
28. F. Phan, et. al. "Speaker identification using neural networks and wavelets" IEEE Eng. in Med. & Bio. Magazine pp.92-101, Vol. 19, No. 1 Feb. 2000
29. W. S. Liming, et. al. ""Smoothness" in embedded time-series" Chaos in Communications,
SPIE. Vol.2038 pp.242-251, 1993
30. W. S. Liming, et. al. ""Smoothness" in embedded time-series" Chaos in Communications, SPIE. Vol.2038 pp.242-251, 1993
31. Peter Grassberger, Itamar Procaccia: Estimation of the Kolmogorov entropy from a chaotic signal; PHYSICAL REVIEW A,Vol.28,No.4,Oct. 1983(2591-2593)
28. Alan WOLF, et.: DETERMINING LYAPUNOV EXPONENTS FROM A TIME SERIES; Physica 16D(1985)285-317
29. Aviad Cohen and Itamar Procaccia: Computing the Kolmogorov entropy from time signals of dissipative and conservation dynamical systems; PHYSICAL REVIEW A,Vol.31, No.3, Mar. 1985 (1872-1882)
30. N.H.Packard and W. Lutzenberger et al Geometry from a time series. Phy. Rev .Let 45(1980) 721-715
31. N.H.Pachard, J.P. Crutchfield, et.: Germetry from a Time Series; PHYSICAL REVIEW LETTERS,Vol.45, No.9, 1980(712-716)
32. Peter Grassberger: Toward a Quantitative Theory of Self-Generated Complexity; Int. Journal of Theroerical Physics, Vol.25, No.9, 1986, p907-938
33. Steven M. Pincus- Approximate entropy as a measure of system complexity; Proc. Natl. Acad. Sci. USA,Vol.88, pp.2297-2301, Mar. 1991
34. D.S.Broomhead and Gregory P.King. Extracting qualitative dynamics from experimental data. Phys.D 20 (1986) 217-236
35. Soong A C K et al, Biol Cyber 62 (1989) 55-62
36. Rosechke j. et al, Biol Cybern 64(1991) 307-313
37. J.E.Bates, H.K. Sheperd, information fluctuation as a measure of complexity University of New Hampshire, Durham (1991)
38. J.j. wright Biol Cybeern 76 (1997) 181_194 EEG simulation: variation of spectral envelope, pulse synchrony and ≈40 HZ oscillation. Mental Health Research Institute of Victoria Parkville, Victoria 3052 Australia