基于DSP的语音编码中的模糊增量调制算法的实现
摘要
随着多媒体通信不断地融入到了我们的生活中,语音编码扮演着越来越重要的角色。目前应用比较广泛的有波形编码方式以及参数编码和混合编码方式,虽然参数编码和混合编码在编码速率上有一定的优势,但因为其解码恢复出来的语音自然度较低,且算法复杂度较高,对实时处理有一定的局限,而波形编码以其硬件实现电路简单,具有较强的抗干扰能力及极强的保密性仍广泛应用于军事通信等领域。目前广泛采用的算法有差分编码调制(DPCM),自适应差分编码调制(adaptive DPCM),增量调制(△M)。其主要原理都是对预测器的预测值与实际值之差编码,其预测器中Δ的变化都是依赖于精确的数学模型,这使得对于一些幅度变化特别大的信号来说仍然会带来严重的颗粒噪声或斜率过载失真。
为解决上述问题,在本设计中将模糊控制理论引入到增量调制中,提出了一种模糊增量调制算法,它具有无需确定其准确数学模型的特点,即将预测值与实际值之差e(n)及e(n)与e(n-1)之差Δe(n)作为输入,经过模糊控制器一系列推理输出精确的控制量δ。最后通过合理选取参数初始值,在matlab上对此算法进行仿真并分析其性能,为了进一步考查此算法的实时处理能力,将此算法移植到具有强大的数字信号处理能力的DSP芯片TMS320C5402的开发板上,结果表明在DSP上有较好的编解码效果:量阶δ的大小能够随着语音信号幅度的变化快慢自动地改变,使预测器输出能更好地跟踪快速变化的信号波形,既具备了ADM的优点,又比它有更好的SNR性能和更大的动态范围,基本实现了“透明”编码,达到了设计要求。
With the integration of multimedia communications to keep our life, speech coding plays an increasingly important role. At present, there is a broader application of waveform coding and the parameter coding and hybrid coding, although parameters of coding and hybrid coding have on the advantages at coding rates, but for its decoding restored voice's nature is low and high complexity algorithm for real-time processing so they have limitations, for waveform coding circuit has easy hardware implementation, strong interference capability and Security, so it is still very widely used in military communications and other fields, now there is a widely used algorithms such as DPCM, ADPCM,ΔM. the main principle of it is to code with the difference between forcasting value and the actual value .the changes ofΔin the predictor are dependent on the precise mathematical model . which still causes severe slope overload noise or distortion for some particularly high rate of change of the signal.
To address these issues, come up with an algorithm—fuzzy delta modulation algorithm using fuzzy control theory, which needn't to determine the accurate mathematical model of the process, that is put the difference between Predictive value and the actual value as input , then by application of fuzzy control theory to predict the difference (e(n))between the value of prediction and the actual value and the difference (Δe (n)) between e (n) and e (n-1) in fuzzy way , get a a series of fuzzy control through fuzzy inference based on fuzzy control rules .then get precise value of control output ofδthrough fuzzy theory. Select the parameters of the initial adoption of a reasonable value, simulate and analysis of its performance on matlab, in order to further test this algorithm in real-time processing capability, at last migrate this algorithm to a powerful digital signal processing capability of DSP chip TMS320C5402 development board as the core, the results show that this algorithm in the DSP codec better effect: the volume ofδsize changes automatically with the voice signal amplitude changes which make the forecast better track the rapid changes in the signal waveform, it has not only the advantage of ADM, but also a better SNR performance and more dynamic the scope, fulfill the basic realization of the 'transparent' coding, then achieve the design requirements.
为解决上述问题,在本设计中将模糊控制理论引入到增量调制中,提出了一种模糊增量调制算法,它具有无需确定其准确数学模型的特点,即将预测值与实际值之差e(n)及e(n)与e(n-1)之差Δe(n)作为输入,经过模糊控制器一系列推理输出精确的控制量δ。最后通过合理选取参数初始值,在matlab上对此算法进行仿真并分析其性能,为了进一步考查此算法的实时处理能力,将此算法移植到具有强大的数字信号处理能力的DSP芯片TMS320C5402的开发板上,结果表明在DSP上有较好的编解码效果:量阶δ的大小能够随着语音信号幅度的变化快慢自动地改变,使预测器输出能更好地跟踪快速变化的信号波形,既具备了ADM的优点,又比它有更好的SNR性能和更大的动态范围,基本实现了“透明”编码,达到了设计要求。
With the integration of multimedia communications to keep our life, speech coding plays an increasingly important role. At present, there is a broader application of waveform coding and the parameter coding and hybrid coding, although parameters of coding and hybrid coding have on the advantages at coding rates, but for its decoding restored voice's nature is low and high complexity algorithm for real-time processing so they have limitations, for waveform coding circuit has easy hardware implementation, strong interference capability and Security, so it is still very widely used in military communications and other fields, now there is a widely used algorithms such as DPCM, ADPCM,ΔM. the main principle of it is to code with the difference between forcasting value and the actual value .the changes ofΔin the predictor are dependent on the precise mathematical model . which still causes severe slope overload noise or distortion for some particularly high rate of change of the signal.
To address these issues, come up with an algorithm—fuzzy delta modulation algorithm using fuzzy control theory, which needn't to determine the accurate mathematical model of the process, that is put the difference between Predictive value and the actual value as input , then by application of fuzzy control theory to predict the difference (e(n))between the value of prediction and the actual value and the difference (Δe (n)) between e (n) and e (n-1) in fuzzy way , get a a series of fuzzy control through fuzzy inference based on fuzzy control rules .then get precise value of control output ofδthrough fuzzy theory. Select the parameters of the initial adoption of a reasonable value, simulate and analysis of its performance on matlab, in order to further test this algorithm in real-time processing capability, at last migrate this algorithm to a powerful digital signal processing capability of DSP chip TMS320C5402 development board as the core, the results show that this algorithm in the DSP codec better effect: the volume ofδsize changes automatically with the voice signal amplitude changes which make the forecast better track the rapid changes in the signal waveform, it has not only the advantage of ADM, but also a better SNR performance and more dynamic the scope, fulfill the basic realization of the 'transparent' coding, then achieve the design requirements.
引文
[1]JAYANT N S.(1974).Digital Coding Of Speech Waveforms:PCM,DPCM,and DM quantizers[A].Proc.IEEE[C].1974,62:611-632.
[2]樊昌信,詹道庸,徐炳祥等.通信原理[M].北京:国防工业出版社,2000.2
[3]王勇涛,邵丙铣.连续可变斜率增量调制编解码器的设计.微电子学.1999,6,29(3):1-5
[4]TI Instruments.TMS320C54x DSP Reference Set Volume 1:CPU and Peripherals,1997.
[5]徐素妍,曹坤梅.基于Matlab语言的现代通信仿真分析.计算机应用.2001,第21卷第8期:52-53
[6]V.Cuperman,A.Gersho,Adaptive Differential Vector Coding of Speech.IEEE GLOBECOM,1982:1092-1096
[7]ITU-T Recommendation G711,Pulse code modulation(PCM) of voice frequencies,Nov 1988.
[8]V.Cuperman and A.Gersho,Vector predictive coding of speech at 16 kbits/s,IEEE Trans On Commun.1985(33):685-696
[9]M.R.Schroeder,B.S.Atal.Code Excited Linear Prediction(CELP) High Quality Speech at very LowBit rates.ICASSP,1985:937-940.
[10]韩剑坡,王林斗,郝晓鹏,张书军,海春辉.基于DSP的ACELP语音编码系统.电子测量技术.2007.1
[11]李振玉,卢玉民.现代通信中的编码技术.中国铁道出版社,1996
[12]S.Mallat.A theory for multiresolution signal decomposition:the wavelet representation,IEEE Trans.Patt.Anal.And Mach.Intell.1989,11(7):674-693
[13]R.D.Coifman,D.L.Donoho.Translation-invariant de-noising[J].Wavelets and Statistics,New York,CA:Springer-verlag,1995
[14]ITU-T Recommendation G.723.1 Dual Rate Speech Coder for Multimedia Communications Transmitting at 5.3 and 6.3 kbit/s,Mar 1996
[15]韩剑坡,王林斗,郝晓鹏,张书军,海春辉.基于DSP的ACELP语音编码系统.电子测量技术.2007.1
[16]R.R.Coifman and M.V Wickerhauser.Entropy-Based Algorithms For Best Basis Selection,IEEE Trans On Information Theory,vol.38,1992:713-718
[17]韩峻峰,李玉惠等编著.模糊逻辑控制技术[M].重庆.重庆大学出版社.2003年.p.29-56
[18]丛爽著.神经网络、模糊控制及其在运动控制中的应用[M].合肥.中国科学技术大学出版社.2001.5.p.110-133
[19]楼天顺,胡昌华,张伟编著.基于MATLAB的系统分析与设计-模糊控制[M].西安.西安电子科技大学出版社.1999年.p.70-90
[20]WANG zhao xiang,XIA hai lun,DING wei.A Fuzzy Logic Cooperative Mac For Manet.The Jounal Of China Universities Of Posts and Telecommunication.2008.3
[21]王炳锡.语音编码.西安电子科技大学出版社,2002.
[22]N.S.Jayant.Digital Coding of Speech Waveforms:PCM,DPCM,and DM Quantizers,Proc.IEEE,1974,Vol.62,No.5:611-633
[23]李昌立,吴善培.数字语音语音编码使用教程[北京人民邮电出版社,2004.11
[24]李振玉,卢玉民.现代通信中的编码技术.中国铁道出版社,1996:45-78
[25]Chong Kwan Un,Hwang Soo Lee.A Study of the Comparative Performance of Adaptive Delta ModulationSystems[J].IEEE Transactions On Communi cations,1980,Com-28(1):96-102.
[26]Chong Un,Hwang Lee.A Study of the Comparative Performance of Adaptive Delta Modulation Systems.IEEE Trans on Comm,1980,Vol.28,Issue 1:96-101
[27]]王炳锡.语音编码.西安电子科技大学出版社,2002.
[28]周捷.石念*.基于CVSD编码的无限语音系统方案的设计与研究.微计算机信息.2006.2
[29]丛爽著.神经网络、模糊控制及其在运动控制中的应用[M].合肥.中国科 学技术大学出版社.2001.5.p.110-133
[30]贺今朝,一种基于FPGA的模糊控制器的研究.2002.3.p.12-16
[31]王炳锡.语音编码.西安电子科技大学出版社,2002:119-145
[32]赵静,密度自适应的无线传感器网络簇头功率控制技术.2005.12.p.33-34
[33]WANG zhao xiang,XIA hal lun,DING wei.A Fuzzy Logic Cooperative Mac For Manet.The Jounal Of China Universities Of Posts and Telecommunication.2008.3
[34]Liu Weiyan,Zhang Shunyi,Zhang Mu,Liu TaoA Fuzzy-Logic Control Algorithm For Active Queue Management In IP Networks.Journal Of Electronics.January 2008
[35]YU Yang,WEI Xue-xia,ZHANG Yong-fa.Fuzzy Logic Control for Suspension Systems of Tracked Vehicle.Journal of Beijing Institute of Technolog.2009.1
[36]贺今朝,一种基于FPGA的模糊控制器的研究.2002.3.p.12-16
[37]丛爽著.神经网络、模糊控制及其在运动控制中的应用[M].合肥.中国科学技术大学出版社.2001.5.p.110-133
[38]孙庆,吴宏鑫.自适应模糊控制的仿真研究[J].计算机仿真.Vol 14 No.3 July 1997 p3-9
[39]TI Instruments.TMS320C54x DSP Reference Set Volume 2:Mnemonic Instuction Set,1997.
[40]TI Instruments.TMS320C54x DSP Reference Set Volume 4:Applicatons
[41]罗雪松,基于数字对讲机的语音处理方案的研究及DSP实现,2008.3.p.21-23
[42]刘益成.TM5320C54X DSP应用程序设计与开发.北京航空航天大学出版社,2002..p.234-279
[43]张波,周煦林等.TMS320C3X DSP引导装载系统的设计与实现[J].微电子学与计算机2004.21(10).p110-116
[44]阳华,刘海涛.新一代TI集成开发环境Code Composer Studio.微处理机.2001.5.p.33-35
[45]郭森林.颜允圣著,数字信号处理器--体系结构、实现与应用,清华大学出版社,2005.6.p.495-501
[46]周德卿.CCS集成环境下C与汇编的混合编程实现.微电子技术.2002.4.p.45-47
[47]张勇.C/C++语言硬件程序设计-基于TMS320C5000系列DSP.西安电子科技大学出版社,2003.5.p.127-156
[48]彭启琮,李玉柏,管庆DSP技术的发展与应用,高等教育出版社,2002.p.221-256
[49]张雄伟,陈亮,徐光辉.DSP集成开发与应用实例.电子工业出版社,2002.
[50]Thomas.F.quatieri.离散时间语音信号处理.北京电子工业出版社.2004.8:1-30
[51]杨行峻,迟惠生.语音信号数字处理.电子工业出版社,1995:60-72
[52]张雄伟,曹铁勇.DSP芯片的原理与开发应用(第2版).电子工业出版社,2000..p.178-201
[2]樊昌信,詹道庸,徐炳祥等.通信原理[M].北京:国防工业出版社,2000.2
[3]王勇涛,邵丙铣.连续可变斜率增量调制编解码器的设计.微电子学.1999,6,29(3):1-5
[4]TI Instruments.TMS320C54x DSP Reference Set Volume 1:CPU and Peripherals,1997.
[5]徐素妍,曹坤梅.基于Matlab语言的现代通信仿真分析.计算机应用.2001,第21卷第8期:52-53
[6]V.Cuperman,A.Gersho,Adaptive Differential Vector Coding of Speech.IEEE GLOBECOM,1982:1092-1096
[7]ITU-T Recommendation G711,Pulse code modulation(PCM) of voice frequencies,Nov 1988.
[8]V.Cuperman and A.Gersho,Vector predictive coding of speech at 16 kbits/s,IEEE Trans On Commun.1985(33):685-696
[9]M.R.Schroeder,B.S.Atal.Code Excited Linear Prediction(CELP) High Quality Speech at very LowBit rates.ICASSP,1985:937-940.
[10]韩剑坡,王林斗,郝晓鹏,张书军,海春辉.基于DSP的ACELP语音编码系统.电子测量技术.2007.1
[11]李振玉,卢玉民.现代通信中的编码技术.中国铁道出版社,1996
[12]S.Mallat.A theory for multiresolution signal decomposition:the wavelet representation,IEEE Trans.Patt.Anal.And Mach.Intell.1989,11(7):674-693
[13]R.D.Coifman,D.L.Donoho.Translation-invariant de-noising[J].Wavelets and Statistics,New York,CA:Springer-verlag,1995
[14]ITU-T Recommendation G.723.1 Dual Rate Speech Coder for Multimedia Communications Transmitting at 5.3 and 6.3 kbit/s,Mar 1996
[15]韩剑坡,王林斗,郝晓鹏,张书军,海春辉.基于DSP的ACELP语音编码系统.电子测量技术.2007.1
[16]R.R.Coifman and M.V Wickerhauser.Entropy-Based Algorithms For Best Basis Selection,IEEE Trans On Information Theory,vol.38,1992:713-718
[17]韩峻峰,李玉惠等编著.模糊逻辑控制技术[M].重庆.重庆大学出版社.2003年.p.29-56
[18]丛爽著.神经网络、模糊控制及其在运动控制中的应用[M].合肥.中国科学技术大学出版社.2001.5.p.110-133
[19]楼天顺,胡昌华,张伟编著.基于MATLAB的系统分析与设计-模糊控制[M].西安.西安电子科技大学出版社.1999年.p.70-90
[20]WANG zhao xiang,XIA hai lun,DING wei.A Fuzzy Logic Cooperative Mac For Manet.The Jounal Of China Universities Of Posts and Telecommunication.2008.3
[21]王炳锡.语音编码.西安电子科技大学出版社,2002.
[22]N.S.Jayant.Digital Coding of Speech Waveforms:PCM,DPCM,and DM Quantizers,Proc.IEEE,1974,Vol.62,No.5:611-633
[23]李昌立,吴善培.数字语音语音编码使用教程[北京人民邮电出版社,2004.11
[24]李振玉,卢玉民.现代通信中的编码技术.中国铁道出版社,1996:45-78
[25]Chong Kwan Un,Hwang Soo Lee.A Study of the Comparative Performance of Adaptive Delta ModulationSystems[J].IEEE Transactions On Communi cations,1980,Com-28(1):96-102.
[26]Chong Un,Hwang Lee.A Study of the Comparative Performance of Adaptive Delta Modulation Systems.IEEE Trans on Comm,1980,Vol.28,Issue 1:96-101
[27]]王炳锡.语音编码.西安电子科技大学出版社,2002.
[28]周捷.石念*.基于CVSD编码的无限语音系统方案的设计与研究.微计算机信息.2006.2
[29]丛爽著.神经网络、模糊控制及其在运动控制中的应用[M].合肥.中国科 学技术大学出版社.2001.5.p.110-133
[30]贺今朝,一种基于FPGA的模糊控制器的研究.2002.3.p.12-16
[31]王炳锡.语音编码.西安电子科技大学出版社,2002:119-145
[32]赵静,密度自适应的无线传感器网络簇头功率控制技术.2005.12.p.33-34
[33]WANG zhao xiang,XIA hal lun,DING wei.A Fuzzy Logic Cooperative Mac For Manet.The Jounal Of China Universities Of Posts and Telecommunication.2008.3
[34]Liu Weiyan,Zhang Shunyi,Zhang Mu,Liu TaoA Fuzzy-Logic Control Algorithm For Active Queue Management In IP Networks.Journal Of Electronics.January 2008
[35]YU Yang,WEI Xue-xia,ZHANG Yong-fa.Fuzzy Logic Control for Suspension Systems of Tracked Vehicle.Journal of Beijing Institute of Technolog.2009.1
[36]贺今朝,一种基于FPGA的模糊控制器的研究.2002.3.p.12-16
[37]丛爽著.神经网络、模糊控制及其在运动控制中的应用[M].合肥.中国科学技术大学出版社.2001.5.p.110-133
[38]孙庆,吴宏鑫.自适应模糊控制的仿真研究[J].计算机仿真.Vol 14 No.3 July 1997 p3-9
[39]TI Instruments.TMS320C54x DSP Reference Set Volume 2:Mnemonic Instuction Set,1997.
[40]TI Instruments.TMS320C54x DSP Reference Set Volume 4:Applicatons
[41]罗雪松,基于数字对讲机的语音处理方案的研究及DSP实现,2008.3.p.21-23
[42]刘益成.TM5320C54X DSP应用程序设计与开发.北京航空航天大学出版社,2002..p.234-279
[43]张波,周煦林等.TMS320C3X DSP引导装载系统的设计与实现[J].微电子学与计算机2004.21(10).p110-116
[44]阳华,刘海涛.新一代TI集成开发环境Code Composer Studio.微处理机.2001.5.p.33-35
[45]郭森林.颜允圣著,数字信号处理器--体系结构、实现与应用,清华大学出版社,2005.6.p.495-501
[46]周德卿.CCS集成环境下C与汇编的混合编程实现.微电子技术.2002.4.p.45-47
[47]张勇.C/C++语言硬件程序设计-基于TMS320C5000系列DSP.西安电子科技大学出版社,2003.5.p.127-156
[48]彭启琮,李玉柏,管庆DSP技术的发展与应用,高等教育出版社,2002.p.221-256
[49]张雄伟,陈亮,徐光辉.DSP集成开发与应用实例.电子工业出版社,2002.
[50]Thomas.F.quatieri.离散时间语音信号处理.北京电子工业出版社.2004.8:1-30
[51]杨行峻,迟惠生.语音信号数字处理.电子工业出版社,1995:60-72
[52]张雄伟,曹铁勇.DSP芯片的原理与开发应用(第2版).电子工业出版社,2000..p.178-201