基于G.729A的嵌入式IP电话终端的设计与实现
|
摘要
针对第三代IP电话的技术现状和发展趋势，本论文提出了一种新型的嵌入式IP电话终端解决方案，并完成了该终端的核心部分，语音编解码器的设计与实现。
该语音编解码器的硬件基于TMS320VC5410，编解码算法遵循ITU-T G.729A协议，能够实现语音信号的采集/回放、编码/解码以及同嵌入式CPU通信等功能，在8kbit/s的码率下能够提供获得良好的语音质量。
现有的G.729A定点算法大都是针对通用计算机编写的，没有考虑算法的实时性问题。因此需要根据DSP的结构特点，对算法进行优化。在对ITU-T提供的G.729A标准C源代码进行测试时发现，一些单次调用消耗运算量不大的基本运算，往往被调用几千次，消耗了整个算法运算量的90％以上。采用了内联函数和宏语言相结合的方法对这些基本运算进行了优化，优化后G.729A编码器的计算复杂度由530MIPS降低到94MIPS。这一步是整个算法优化的关键。接着，采用DSPLIB库函数对算法中的数字信号处理函数进行了优化，根据DSP的结构特点，改写了部分循环和判断转移语句，最终整个算法的计算复杂度降低到30MIPS，完全满足实时要求。整个优化过程始终着眼于底层函数的优化，没有把主要精力放在固定码本搜索、自适应搜索等单次调用耗时较大的算法上。实践表明，本文提出的自底向上的优化方法，效果良好，优化迅速，对其他算法的优化也有借鉴意义。
提出了“DSP+嵌入式CPU”的IP语音终端解决方案，并给出了IP电话编解码器的系统构架和芯片选型。在电路模块分析中，重点介绍了电源环境的构建、语音信号采集/回放(A/D、D/A)模块的设计、DSP与嵌入式CPU通信接口设计以及DSP核心部分设计，并给出了在原理图设计、制板与调试过程中总结的经验。最后，详细介绍了IP电话编解码器软件的设计，主要有系统自举启动设计、DSP芯片初始化、TLC320AD50C驱动程序设计以及系统软件流程设计等。
Based on the analysis of current situation and the development trend of IP phone, this paper has put forward the solution of a kind of new-type embedded IP phone terminal and has finished designing the IP phone codec which is the key part of IP phone terminal.
The hardware of the IP phone codec to be designed is based on the fixed point digital signal processor (TI'S TMS320VC5410) while the compression and decompression core in the software of DSP is based on the ITU-T vG.729A .IP phone codec carryout the task of collecting /playing-back .coding /decoding of speech signal and communication with embedded CPU.etc.
The existing fixed-point implementation of G.729A is generally written on all-purpose computer, and is no considered real-time quality problems . So the algorithm is needed to be optimized . when testing the G.729A ANSI C source codes offered by ITU-T, we found that some basic operations are called several thousand times and consist of more than 90% of the operation quantity of the whole algorithm.After optimizing these basic operations with the method of combining intrinsic with macro language,the complexity of calculation of G.729A encoders is reduced from 530 MIPSs to 94 MIPSs. This is a key for the optimizing of the whole algorithm . Then,the digital signal processing functions in G.729A are optimized by DSPLIB.According to the structure characteristic of DSP,some sentences of circulation and shifting are rewritten.Finally,the complexity of calculation of the whole algorithm is reduced to 30 MIPSs,which satisfies the real-time request very well.
This thesis has put forward IP phone terminal solution of DSP+ embedded CPU ". On the basis of this scheme, this text introuduces the IP telephone systematic framework and the selection of chips for the codec. The circuit design ,focuses on the following hardware modules: power structure environments, A/D,D/A inverter, DSP and embedded CPU's communication interface,DSP modules,and so on. The problems and the corresponds solutions which found in the design and debug are discussed,too. At last, the software design is presented in detail,including system booting,initialization of DSP registers,drivers for AD converter and system software.
该语音编解码器的硬件基于TMS320VC5410，编解码算法遵循ITU-T G.729A协议，能够实现语音信号的采集/回放、编码/解码以及同嵌入式CPU通信等功能，在8kbit/s的码率下能够提供获得良好的语音质量。
现有的G.729A定点算法大都是针对通用计算机编写的，没有考虑算法的实时性问题。因此需要根据DSP的结构特点，对算法进行优化。在对ITU-T提供的G.729A标准C源代码进行测试时发现，一些单次调用消耗运算量不大的基本运算，往往被调用几千次，消耗了整个算法运算量的90％以上。采用了内联函数和宏语言相结合的方法对这些基本运算进行了优化，优化后G.729A编码器的计算复杂度由530MIPS降低到94MIPS。这一步是整个算法优化的关键。接着，采用DSPLIB库函数对算法中的数字信号处理函数进行了优化，根据DSP的结构特点，改写了部分循环和判断转移语句，最终整个算法的计算复杂度降低到30MIPS，完全满足实时要求。整个优化过程始终着眼于底层函数的优化，没有把主要精力放在固定码本搜索、自适应搜索等单次调用耗时较大的算法上。实践表明，本文提出的自底向上的优化方法，效果良好，优化迅速，对其他算法的优化也有借鉴意义。
提出了“DSP+嵌入式CPU”的IP语音终端解决方案，并给出了IP电话编解码器的系统构架和芯片选型。在电路模块分析中，重点介绍了电源环境的构建、语音信号采集/回放(A/D、D/A)模块的设计、DSP与嵌入式CPU通信接口设计以及DSP核心部分设计，并给出了在原理图设计、制板与调试过程中总结的经验。最后，详细介绍了IP电话编解码器软件的设计，主要有系统自举启动设计、DSP芯片初始化、TLC320AD50C驱动程序设计以及系统软件流程设计等。
Based on the analysis of current situation and the development trend of IP phone, this paper has put forward the solution of a kind of new-type embedded IP phone terminal and has finished designing the IP phone codec which is the key part of IP phone terminal.
The hardware of the IP phone codec to be designed is based on the fixed point digital signal processor (TI'S TMS320VC5410) while the compression and decompression core in the software of DSP is based on the ITU-T vG.729A .IP phone codec carryout the task of collecting /playing-back .coding /decoding of speech signal and communication with embedded CPU.etc.
The existing fixed-point implementation of G.729A is generally written on all-purpose computer, and is no considered real-time quality problems . So the algorithm is needed to be optimized . when testing the G.729A ANSI C source codes offered by ITU-T, we found that some basic operations are called several thousand times and consist of more than 90% of the operation quantity of the whole algorithm.After optimizing these basic operations with the method of combining intrinsic with macro language,the complexity of calculation of G.729A encoders is reduced from 530 MIPSs to 94 MIPSs. This is a key for the optimizing of the whole algorithm . Then,the digital signal processing functions in G.729A are optimized by DSPLIB.According to the structure characteristic of DSP,some sentences of circulation and shifting are rewritten.Finally,the complexity of calculation of the whole algorithm is reduced to 30 MIPSs,which satisfies the real-time request very well.
This thesis has put forward IP phone terminal solution of DSP+ embedded CPU ". On the basis of this scheme, this text introuduces the IP telephone systematic framework and the selection of chips for the codec. The circuit design ,focuses on the following hardware modules: power structure environments, A/D,D/A inverter, DSP and embedded CPU's communication interface,DSP modules,and so on. The problems and the corresponds solutions which found in the design and debug are discussed,too. At last, the software design is presented in detail,including system booting,initialization of DSP registers,drivers for AD converter and system software.
引文
[1] 杨行峻，迟惠生．语音信号数字处理．北京：电子工业出版社，1995
[2] 鲍长春．低比特率数字语音编码基础．北京：北京工业大学出版社，2001
[3] 王炳锡．语音编码．北京：科学出版社，1997
[4] ITU-T. Recommendation G. 729, Coding of speech at 8 kbit/s using conjugate structure algebraic-code-excited linear-prediction(CS-ACELP). 1996
[5] ITU-T. Recommendation G. 729-Annex A, Coding of speech at 8 kbit/s using conjugate structure algebraic-code-excited linear-prediction(CS-ACELP) Annex A: Reduced complexity 8 kbit/s CS-ACELP speech codec. 1996
[6] ITU-T. Recommendation H. 323, Packed based multimeda communications system .1999
[7] Texas Instruments Incorporated. TMS320VC5410 Fixed-Point Digital Signal Processor. 1998
[8] Texas Instruments Incorporated. Optimized DSP Library for C Programmers on the TMS320C54x. 2000
[9] Texas Instruments Incorporated. TMS320C54x DSP Reference Set, Volume 5: Enhanced Peripherals. 1999
[10] Texas Instruments Incorporated. TMS320C54x Optimizing C/C++ Compiler User' s Guide. 2001
[11] Texas Instruments Incorporated. TMS320C54x DSP Reference Set Volume 1: CPU and Peripherals. 1999
[12] Texas Instruments Incorporated. TLC320AD50C/TLC320AD52C Sigma-Delta Analog Interface Circuits With Master-Slave Function. 2000
[13] Texas Instruments Incorporated. TPS767D301, TPS767D318, TPS767D325 DUAL-OUTPUT LOW-DROPOUT VOLTAGE REGULATORS, 1999
[14] SAMSUNG. S3C44BOX RISC MICROPROCESSOR, 2000
[15] 李红涛．IP语音电话及其安全性．信息安全与通信保密，2002，3：50-52
[16] 胡宏新，黄永峰．浅谈VOIP关键技术及其发展趋势．电信快报，2002，11：21-24
[17] 马云鹏．VOIP用户网关和终端的实现．电子质量，2003，4：156-159
[18] 宋道元，王永灵，王进军等．基于DSP的VOIP路由器用户板的设计与实现．微计算机信息， 2003，3：55-57
[19] Tanyer, S. Gokhun, Ozer. Voice activity detection in nonstationary noise .IEEE Trans Speech Audio Process,2000,8(2):478-472
[20] Fuchs, H.V., Zha, X.Creating low-noise environments in communication rooms.Applied Acoustics,2001,62(12): 1375-1396
[21] Itoh Y, Tajima K, Kuwabara N. Measurement of subjective communication quality for optical mobile communication systems by using mean opinion score,IEEE Int Symp Person Indoor Mobile Radio Commun PIMRC,2000,2:1330-1334
[22] Itoh, Yoshihiro.Evaluation of subjective communication quality of optical mobile communication systems by mean opinion score.IEICE Transactions on Communieations,2000,83(8):1775-1782
[23] 欧阳缮， 陈云宇．一种改进的语音信号非线性自适应预测编码方案．电路与系统学报．1999,4(2)：7-9
[24] Barthold, Jim .SBC gives satellite a try.Telephony,2002,242(16):14-14
[25] Vittore, Vince .SBC targets small business with new bundled packages. Telephony , 2003,244(37): 10
[26] 李文元，张毓忠．数字音频压缩自适应变换编码算法的研究．信号处理．1998,14(3)：245-248
[27] Gleason J M, Sherman J W. ATC algorithm evaluation,Proc SPIE Int Soc Opt Eng, Proc SPIE Int Soc Opt Eng,1991,1406:169-170
[28] 赵鹤芹．LPC语音合成器的设计与实现．自动化学报，1990，16(5)：473-474
[29] 王田，崔慧娟．改进型LPC声码器语音编码算法．通信学报，1996，17(5)：1-8
[30] 王承发，孙立新．基于LPC全极点模型增强算法在高噪声背景下命令语音识别系统．哈尔滨工业大学学报，1998，30(1)：68-72
[31] Zhang Shude, Lockhart, Gordon.Embedded scheme for regular pulse excited (RPE) linear predictive coding.ICASSP IEEE Int ConfAcoust Speech Signal Process Proe, 1995,1:37-40
[32] Parsa V, Jamieson D G. Interactions between speech coders and disordered speech.Speech Communication,2003,40(3):365-385
[33] 唐昆，崔慧娟．高质量4-8kb/s变速率有限状态ACELP语音编码算法研究．电子学报．2000， 28(1):21-25
[34] 潘苏．用CELP实现GSM话音半速率编码．电子学报，1997，25(1)：73-76
[35] Shijun Yang, Xixian Chen.Improved adaptive CELP coder at 5.2 kbit/s.Proceedings of the 1993 IEEE Region 10 Conference on Computer, Communication, Control and Power Engineering.1993,3 (5):299-302
[36] 李晋声，倪维桢．16kbit/s LD-CELP算法的再研究．北京邮电大学学报，1995，]8(4)：43-50
[37] Kumar Arun, Gersho AIIen.LD-CELP speech coding with nonlinear prediction.IEEE Signal Processing Letters, 1997,4(4):89-91
[38] Chen Juin-Hwey, Cox Richard V. LD-CELP: A high quality 16 kb/s speech coder with low delay, IEEE Global Telecommun Conf Exhib, 1990,1:528-532
[39] Chen Jian, Cao Jianjie. Study on LD-CELP coder at the bit rate of 8Kb/s.Shanghai Jiaotong Daxue Xuebao, 1996,30(11 ):75-81
[40] 薛质，顾尚杰．G．729 CS-ACELP语音译码器丢失帧的重建机制。上海交通大学学报，1999， 33(1)：119-121
[41] 刘志勇，唐昆．一种高质量的8kb／sACELP语音编码算法及其实时实现．电子学报，1997，25(7)：72-74
[42] Kataoka, Akitoshi,Improved CS-CELP speech coding in a noisy environment using a trained sparse conjugate codebook.ICASSP IEEE Int Conf Acoust Speech Signal Process Proc, 1996,1:29-32
[43] Kataoka Akitoshi, Moriya Takehiro, Hayashi Shinji. 8-kb/s conjugate structure CELP (CS-CELP)
speech coder.IEEE. Trans Speech Audio Process 1996,4(6):_401-411
[44] 胡金波，陈慧剑．DSP环境下C语言编程的优化实现．现代电子技术，2002，]2：82-84
[45] 李小红．TMS320C62x DSP的软件开发与优化编程．计算机工程，2002，1：244，246
[46] 田小永，古志民．基于TMS320C55xDSP的代码优化．计算机应用，2003，1：56-59
[47] 周洁，杨心怀．RISC CPUARM芯片的应用和选型．电子技术应用，2003，8：6-1
[48] 王进军，宋道元．黄本雄．HPI方式自举在TMS320VC5402 DSP芯片上的实现．电子技术应用，2002，11：61-63
[49] 邓思豪，曾春年．C54x DSP的HPI与PC机并口技术研究．电子技术应用．2002．7：66-68
[50] 刘东华，尹军，梁光明．基于TMS320VC5402 MCBSP串口的DMA方式数据传输的实现．微处理机，2001，2：43-46
[2] 鲍长春．低比特率数字语音编码基础．北京：北京工业大学出版社，2001
[3] 王炳锡．语音编码．北京：科学出版社，1997
[4] ITU-T. Recommendation G. 729, Coding of speech at 8 kbit/s using conjugate structure algebraic-code-excited linear-prediction(CS-ACELP). 1996
[5] ITU-T. Recommendation G. 729-Annex A, Coding of speech at 8 kbit/s using conjugate structure algebraic-code-excited linear-prediction(CS-ACELP) Annex A: Reduced complexity 8 kbit/s CS-ACELP speech codec. 1996
[6] ITU-T. Recommendation H. 323, Packed based multimeda communications system .1999
[7] Texas Instruments Incorporated. TMS320VC5410 Fixed-Point Digital Signal Processor. 1998
[8] Texas Instruments Incorporated. Optimized DSP Library for C Programmers on the TMS320C54x. 2000
[9] Texas Instruments Incorporated. TMS320C54x DSP Reference Set, Volume 5: Enhanced Peripherals. 1999
[10] Texas Instruments Incorporated. TMS320C54x Optimizing C/C++ Compiler User' s Guide. 2001
[11] Texas Instruments Incorporated. TMS320C54x DSP Reference Set Volume 1: CPU and Peripherals. 1999
[12] Texas Instruments Incorporated. TLC320AD50C/TLC320AD52C Sigma-Delta Analog Interface Circuits With Master-Slave Function. 2000
[13] Texas Instruments Incorporated. TPS767D301, TPS767D318, TPS767D325 DUAL-OUTPUT LOW-DROPOUT VOLTAGE REGULATORS, 1999
[14] SAMSUNG. S3C44BOX RISC MICROPROCESSOR, 2000
[15] 李红涛．IP语音电话及其安全性．信息安全与通信保密，2002，3：50-52
[16] 胡宏新，黄永峰．浅谈VOIP关键技术及其发展趋势．电信快报，2002，11：21-24
[17] 马云鹏．VOIP用户网关和终端的实现．电子质量，2003，4：156-159
[18] 宋道元，王永灵，王进军等．基于DSP的VOIP路由器用户板的设计与实现．微计算机信息， 2003，3：55-57
[19] Tanyer, S. Gokhun, Ozer. Voice activity detection in nonstationary noise .IEEE Trans Speech Audio Process,2000,8(2):478-472
[20] Fuchs, H.V., Zha, X.Creating low-noise environments in communication rooms.Applied Acoustics,2001,62(12): 1375-1396
[21] Itoh Y, Tajima K, Kuwabara N. Measurement of subjective communication quality for optical mobile communication systems by using mean opinion score,IEEE Int Symp Person Indoor Mobile Radio Commun PIMRC,2000,2:1330-1334
[22] Itoh, Yoshihiro.Evaluation of subjective communication quality of optical mobile communication systems by mean opinion score.IEICE Transactions on Communieations,2000,83(8):1775-1782
[23] 欧阳缮， 陈云宇．一种改进的语音信号非线性自适应预测编码方案．电路与系统学报．1999,4(2)：7-9
[24] Barthold, Jim .SBC gives satellite a try.Telephony,2002,242(16):14-14
[25] Vittore, Vince .SBC targets small business with new bundled packages. Telephony , 2003,244(37): 10
[26] 李文元，张毓忠．数字音频压缩自适应变换编码算法的研究．信号处理．1998,14(3)：245-248
[27] Gleason J M, Sherman J W. ATC algorithm evaluation,Proc SPIE Int Soc Opt Eng, Proc SPIE Int Soc Opt Eng,1991,1406:169-170
[28] 赵鹤芹．LPC语音合成器的设计与实现．自动化学报，1990，16(5)：473-474
[29] 王田，崔慧娟．改进型LPC声码器语音编码算法．通信学报，1996，17(5)：1-8
[30] 王承发，孙立新．基于LPC全极点模型增强算法在高噪声背景下命令语音识别系统．哈尔滨工业大学学报，1998，30(1)：68-72
[31] Zhang Shude, Lockhart, Gordon.Embedded scheme for regular pulse excited (RPE) linear predictive coding.ICASSP IEEE Int ConfAcoust Speech Signal Process Proe, 1995,1:37-40
[32] Parsa V, Jamieson D G. Interactions between speech coders and disordered speech.Speech Communication,2003,40(3):365-385
[33] 唐昆，崔慧娟．高质量4-8kb/s变速率有限状态ACELP语音编码算法研究．电子学报．2000， 28(1):21-25
[34] 潘苏．用CELP实现GSM话音半速率编码．电子学报，1997，25(1)：73-76
[35] Shijun Yang, Xixian Chen.Improved adaptive CELP coder at 5.2 kbit/s.Proceedings of the 1993 IEEE Region 10 Conference on Computer, Communication, Control and Power Engineering.1993,3 (5):299-302
[36] 李晋声，倪维桢．16kbit/s LD-CELP算法的再研究．北京邮电大学学报，1995，]8(4)：43-50
[37] Kumar Arun, Gersho AIIen.LD-CELP speech coding with nonlinear prediction.IEEE Signal Processing Letters, 1997,4(4):89-91
[38] Chen Juin-Hwey, Cox Richard V. LD-CELP: A high quality 16 kb/s speech coder with low delay, IEEE Global Telecommun Conf Exhib, 1990,1:528-532
[39] Chen Jian, Cao Jianjie. Study on LD-CELP coder at the bit rate of 8Kb/s.Shanghai Jiaotong Daxue Xuebao, 1996,30(11 ):75-81
[40] 薛质，顾尚杰．G．729 CS-ACELP语音译码器丢失帧的重建机制。上海交通大学学报，1999， 33(1)：119-121
[41] 刘志勇，唐昆．一种高质量的8kb／sACELP语音编码算法及其实时实现．电子学报，1997，25(7)：72-74
[42] Kataoka, Akitoshi,Improved CS-CELP speech coding in a noisy environment using a trained sparse conjugate codebook.ICASSP IEEE Int Conf Acoust Speech Signal Process Proc, 1996,1:29-32
[43] Kataoka Akitoshi, Moriya Takehiro, Hayashi Shinji. 8-kb/s conjugate structure CELP (CS-CELP)
speech coder.IEEE. Trans Speech Audio Process 1996,4(6):_401-411
[44] 胡金波，陈慧剑．DSP环境下C语言编程的优化实现．现代电子技术，2002，]2：82-84
[45] 李小红．TMS320C62x DSP的软件开发与优化编程．计算机工程，2002，1：244，246
[46] 田小永，古志民．基于TMS320C55xDSP的代码优化．计算机应用，2003，1：56-59
[47] 周洁，杨心怀．RISC CPUARM芯片的应用和选型．电子技术应用，2003，8：6-1
[48] 王进军，宋道元．黄本雄．HPI方式自举在TMS320VC5402 DSP芯片上的实现．电子技术应用，2002，11：61-63
[49] 邓思豪，曾春年．C54x DSP的HPI与PC机并口技术研究．电子技术应用．2002．7：66-68
[50] 刘东华，尹军，梁光明．基于TMS320VC5402 MCBSP串口的DMA方式数据传输的实现．微处理机，2001，2：43-46