报告新鑫外文翻译

1、西安邮电大学毕业设计（译文）学院：系部：微系专业：微班级：微1101学生导师倪德克：讲师起止时间：2015 年 3 月 16 日2015 年 6 月 21 日AMBE algorithm based on digital voice transmission systemCurrently, the voice processing system based on the analysis of the speech constituted to produce a variety of compression algorithms. Which the United States DVSI

2、the company's advanced multi-band excitation (ABME, Advanced Multi-Band Excitation) method has great advantages. Compared to other algorithms It not onlycode rate is low, while in the low baud rate can maintain a good degree ofeffectiveness of natural speech synthesis, also has goodto background

3、noise. If we add forward error correction FEC coding, channel errors can be a certain degree of error correction. AMBE speech coding because of these advantages, AMBE speech coding method in the field integrated communication network, satellite network, the mobile communication network and other pro

4、fessional networks havebeen widely used.AMBE-1000 is developed DVSI AMBE algorithm based on a high performance multi-rate speech coding / decoding chip, very suitable for digital voice communication and processing applications. This paper is designed based on a AMBE-1000 voice compression system, wh

5、ich with the design is simple, variable rate speech coding, good sound quality, power consumption, etc., can be used in a varietyof needs.1 .AMBE-1000'S Working Principle And Mode Of1.1 AMBE-1000'S Working Principle And CharacteristicsAMBE-1000 AMBE algorithm is based on speech coding / deco

6、ding chip, the speech coding / decoding speed can be 2400 9600bps interval between changes to 50bits. In a separate chip voice encoding and decoding channel, which can complete voice encoding and decoding; and all the encoding and decoding operations are done inside the chip, without external expans

7、ion memory. AMBE-1000 is an essential part of the encoder and a decoder, the two mutually independent. encoders Receive 8KHz sampled voice data stream (16bit linear, 8bit A law, 8bit u law) and to a certain rate ofoutput channel data. On the contrary, the decoder receives the channel data andsynthes

8、ized voice data stream. encoder and decoder interfaceis completelyasynchronous. AMBE-1000 with A / DD / A chip as the voice signals. sent to thedecoder for control data and voice data are different. input and output of the voice data stream must be the same format (16bit linear, 8bit A law, 8bit u l

9、aw), the channelinterface uses 8-bit or 16-bit microcontroller.Optional features include the chip echo cancellation, VAD (Voice ActivityDetection), power mode, data / FEC rate selection, these features or by the externalpin input to the decoder to determine thed frame, it is noteworthy that cancover

10、 hardware, softwared set.1.2 AMBE-1000 Operating ModeChannel Interface:Channel interface is used to describe the output from the encoder and the input compressed bit stream to the compressed bit stream decoder. The interfaces can also output status information, for example, can detect whether there

11、is dual tone multi-frequency voice signals (DTMF) input. In addition, The interface can codecs to perform more complex control operations (usually at initialization time). These control features include voice and the choice of error correction code rate, A / DD / Achip set.In most of the voice trans

12、mission system, the actual encoded bit stream format to be extracted out from the channel and system information and constitute a system of co-transmission in a data stream, sent through the transmission channel, the receiver has to be extracted out of and through the decoding AMBE-1000 device requi

13、red toconstitute a data stream format. channel interface flow diagram shown in Figure 2.AMBE-1000 has several modes: parallel and serial, with frame and withoutframe format, active and passive. Which, parallel passive frame mode is the mostflexible and practical in a working m, the study design is p

14、arallel with the passiveframe mode. We can pull and dip switches and the corresponding interface select pin connected, you can choose the appropriate mode, we can choose a similar approach to voice and error correction code rate. Thus, only design the corresponding serial and parallel interface, use

15、rs can format data based on pre-processing. AMBE-1000 data in a frame format, each frame consists of 17 16-bit words. encoder output 17 words per 20ms , the decoder will have received the same 17 words. When the encode / decode the data rate of less than 9600bps, the lack of bit complement 0. It sho

16、uld be notedthat, regardless of what AMBE-1000 rate of work, all 272 (17 words * 16 bits = 272bits) of frame data (including any unused at the end of zero) must be output or inputfrom the encoder decoder. no frame format used in serial mode only.When the code packagey (EPR) signal goes high, indicat

17、ing a coded dataframes waiting for output. When the CHP-RDN along the signal comes in, 8-bitparallel data is latched to the data bus, the microcontroller can beat this timeaccess the data on the data bus; and based on CHP-RDN 34 along the signal in ordertoa data byte. CHP-OBE output buffer is used t

18、o illustrate whether there is data.When the CHP-RDN signal pulse interval is greater than AMBE-1000 clock 350 times the work, you can ignore the CHP-OBE signal. When the AMBE-1000 work in parallel passive input mode, the microcontroller can DPE, CHP-WRN signal to writedata to the AMBE-1000. In the f

19、ollowing of 20ms, the microcontroller should beprepared to34 bytes of data in the frame.Clock And Reset:AMBE-1000's input clock frequency range is 26 30 MHz. It has three input methods: a, TTL clock input directly; Second, CMOS oscillator clock source or direct input; Third, using the crystal os

20、cillator circuit input. Effectively reset signal shouldbe low, and shall continue for at least 6 clock cycles. reposted elsewhere in the paperfordownload2 .AMBE-1000 The A / DD / A InterfaceAnalog voice signals between the chips with the AMBE-1000 through A / DD / A chip to connect to the .A / DD /

21、A chip select carefully, it is best to choose 16-bit linear device. Another of these devices should also consider signal to noise ratio andfiltering features.MC14LC5480 is MOTORORA company introduced a universal single-channel PCM codec filter. It has an input operational amplifier, the output of op

22、 amp input to the encoder. Decoders to receive the PCM data using a differential D / A converter to be expanded . D / A output is 3400Hz low-pass filter following the output signal is then filtered by the active RC filter to remove the band switched capacitor outputfilter.AMBE-1000 when used with an

23、y A / D or D / A interface, the voice sampling rate is 8KHz, which 8KHz voice data to input and output through the serial port.Serial control signal can be completely set by the user. The most flexible approach isto enter a decoder ID = 0x03H thed frame to set these control signals, theuser can perf

24、orm all of the configuration of SIOC. In order to simplify the A / DD / A chip configuration, a set of default configuration values You can pin C_SEL 2 0 to configuration. The default configuration values, indicating the direction of the interface and programmable device of the program word order. A

25、MBE-1000 the A /DD / A interface circuit shown in Figure 3 instructions.3 .System AnalysisThe system must take intothe speech signal gain is large enough, needto add the necessary high-performance amplifier chip, using the AD8542.System power, the micro-controller AT89C51 reset, MC14LC5480 voice of

26、the A/ DD / A conversion and PCM codec, AMBE-1000 is responsible for the process of compression codec. AMBE-1000 voice and data for each completed a coding and decoding operations, and AT89C51 a data exchange. And AT89C51 serial or parallelinterface to exchange data with other nodes, thus completing

27、 the process of digitalvoicecommunications.ThesoftwareprogrammainlyforAT89C51program. in the circuit design, with particular attention to the analog ground anddigital ground The separate analog and digital signal isolation in order to avoidbackground noise and interference.MSP430 Microcontroller-bas

28、edreal-time multi-tasking operating system1.In the MSP430 on the significance of the use of RTOSThe general view is that, MSP430 using the RTOS is no point in this is understandable. MSP430 hardware because of the limited resources (in MSP430F149 for example, only 2KB RAM), any commercial operating

29、systems can not be ported to the MSP430. At present the application MSP430 get RTOS, only the C / OS-II, but using C / OS-II must have an expensive C compiler, which severely restricted intheir use on the MSP430.It is based on the above, the author in the application MSP430 process, the preparation

30、of a MSP430F149 based on the RTOS, tentatively named M430/OS. It occupied less RAM, the code short, a little change can be applied to most otherMSP430 microcontroller.In the MSP430 MCU systems use M430/OS, the system has the followingmeanings: to achieve modular software design. The different functi

31、onal modules can be compiled into the corresponding task by the operating system according to level of calls, which do not have to implement the first function, which functions after theimplementation of the trouble. more rational and effective use of the limited CPU resources. The importance of the

32、 arrangement according to the task the task level, can guarantee the mostimportant task to be the most timely implementation. significantly reduce system failure rates. Low-priority task blocking occurswhen the high-priority implementation of the mandate are not affected.2M430/OS on the realization

33、of the MSP430F1492.1 M430/OS FeaturesM430/OS has the following characteristics: using preemptive kernel, that is, high-priority tasks can be from low-prioritytask 'grab' Back to the CPU control; each task is a task to open a separate stack; each tasked for more than a dozen to hundreds of by

34、tes of the taskstack, task stack size of the field data according to task, local variables and nested callof estimate; each task assigned a different priority, does not support the two tasks have thesame priority level; does not support semaphores, mailboxes function; task state only three: Running

35、(executing),y (y), hangs (suspended); system RAM occupied by the amount of = (task number 1) * 4) 6 bytes, notincluding the task stack; cess, the current version of a total of 86 lines of code in assembly code,256 bytes of object code; theoretically supports up to 126 tasks; task lock function: in s

36、ome low-priority code, operating system, the CPUpower do not want to switch to other tasks, then this code can be locked, when you run this code, it will not cause task switching;The task wake-up feature: In a task to generate a event to trigger other tasks torun (if it is triggered by task priority

37、 higher, it would immediately run).2.2 System Function Description OS_Init: multi-task initialization, the stack of tasks (task stack structureshown in Figure 1), the task delay count, the task initialization status. Initialization is complete, the system switches directly to the highest priority ta

38、sks, multi-taskingsystem startup. OS_Time_Dly: suspend the current task a specified period of time for othertasks to run. OS_Sched: task scheduling, it should first delay of each task by a number,and then to identify the highest priorityy task, andy to switch to this task. Ifthere is noy tasks, swit

39、ch to the idle task. OS_Task: idle task is a very important system tasks, when all tasks aresuspended when you run this task. It is mainly carried out on a counter_Counthas been accumulated, the user can calculate the basis of this counter CPU utilization. OS_Task_Lock: Lock task scheduling, task sc

40、heduling prohibited. Mainlyused to lock in a low priority in some of the re-entrant code or key code. OS_Task_Unlock: unlock task scheduling, and the opposite function of theabove subroutines. OS_Task_Wakeup: wake up the specified priority tasks, and generate a task scheduler, if the priority of the

41、 task to be awakened to run than the current high-priority tasks, the task will be to switch to the task to be awakened, or wait forthe next time a scheduling .基于 AMBE 算法的数字语音传输系统目前，语音处理系统中基于对语音的分析，产生了多种压缩编码算法。其国 DVSI 公司的先进多带激励(ABME，Advanced Multi-Band Excitation)算法具有较大优势。相比其他算法，它不仅码率低，同时在低波特率下能保持优良

42、的语音的自然度效果，还具有良抗背景噪声能力。如果加上 FEC 前向纠错编码，可对信道误码进行一定程度的纠错。正因为 AMBE 语音压缩编码的这些优点，AMBE 语音压缩编码方式在野战综合通信网、网等专业网中得到了广泛的使用。网、移动通信AMBE-1000 是DVSI 公司开发的基于AMBE 算法的一种高性能的多速率语音编码/解码，非常适合于数字语音通信及处理的场合。本文基于AMBE-1000设计了一个语音压，该系统具有设计简便、语音编码速率可变、音质良好、功耗小等优点，可应用于多种需要。1 AMBE-1000 的工作原理和模式1.1 AMBE-1000 的工作原理及特性AMBE-1000 是基

43、于 AMBE 算法的语音编码/解码率可以在 24009600bps 之间以 50bits 的间隔变化。在，其语音编码/解码速内部有相互的语音编码和解码通道，可同时完成语音的编码和解码任务;并且所有的编码和解码内部完成，不需要外扩的器。AMBE-1000 最基本的组成部分操作都在，两者相互。编码器接收 8KHz 采样的语音数就是一个编码器和一个据流(16bit 线性，8bit A 律，8bit u 律)并以一定的速率输出信道数据。相反，解码器接收信道数据并语音数据流。编码器和接口的时序是完全异步的。AMBE-1000 采用 A/D-D/A作为语音信号的接口。送往用于的数据和语音数据是不同的。输入

44、输出的语音数据流的格式必须是相同的(16bit线性的，8bit A 律，8bit u 律)，信道接口采用 8 位或 16 位的微器。可选择的功能回声抵消，VAD(语音激活检测)，电源模式，数据/前向纠错率的选择等，这些功能由管脚或输入到令帧来决定，值得注意的是，命令可覆盖硬件设置。1.2 AMBE-1000 的工作模式通道接口:通道接口用于描述从编码器输出的压缩比特流和输入到的压缩比特流。该接口也可输出状态，例如可以检测是否有双音多频的语音信号(DTMF)执行更复杂的操作(通常在初始化时)。这输入。此外，该接口可对编些功能语音和纠错码速率的选择、A/D-D/A的设置。在多数的语音传输系统中，实

45、际编码比特流以一定格式从通道中摘录出来，并和系统合在一块系统传送数据流，通过传输通道发送，在接收端又被AMBE-1000 所需格式的数据流。摘录出来，并通过AMBE-1000 有多种工作模式:并行和串行，有帧和无帧格式，主动和。其中，并行是并行帧模式是最灵活和实用的一种工作模式，本文设计方案采用的就帧模式。我们可以通过上拉电阻和拨位开关与相应的接口选择引脚相连，就可以选择相应的工作模式，我们还可通过类似的来选择语音和纠错码速率。这样，只需设计相应的串行和并行接口，用户就可以根据预定的格式对数据进行处理。AMBE-1000 的数据在有帧格式下，每帧由 17 个 16 位字组成。编码器每 20ms

46、 输出 17 个字，同样则要接收 17 个字。当编码/解码的数据率低于 9600bps 时，不足的位补 0。需要注意的是，无论 AMBE-1000 工作在什么速率，所有 272 位(17 字*16 bits=272 bits)的帧数据(任何未用的结尾零)编码器输出或输入。无帧格式只能用于串行模式。当编码包就绪(EPR)信号变，表明有一编码的数据帧等待输出。当CHP-RDN 下沿信号到来时，8 位并行数据就锁存到数据总线上，此时微器可以数据总线上的数据;并且根据CHP-RDN 下沿信号依次34 字节的一帧数据。CHP-OBE 则用来说明输出缓冲区是否有数据。当 CHP-RDN 信号的脉冲间隔大于

47、 AMBE-1000AMBE-1000 工作在并行工作时钟的 350倍时，可以忽略 CHP-OBE 信号。当器可以根据 DPE、CHP-WRN器应该准备该帧 34 字输入模式时，则微信号往 AMBE-1000 写数据。在随后的 20ms，微节数据。时钟和复位:AMBE-1000 的输入时钟频率范围为 2630 MHz。它有三种输入方式:一、TTL 时钟源直接输入;二、CMOS 时钟源或振荡器直接输入;三、采用晶体振荡电路输入。有效的复位信号应该是低电平，并且须持续最少 6 时钟周期。2 AMBE-1000 的 A/D-D/A 接口模拟语音信号与 AMBE-1000之间要通过 A/D-D/A来连

48、接。A/D-D/A的选择要慎重，最好要选 16 位线性的器件。另外也要考虑这些器件的信噪比和滤波特性。MC14LC5480 是MOTORORA 公司推出的一款通用单信道PCM 编码解码滤波器。它有一个输入运算放大器，运算放大器的输出输入到编码器。接收到 PCM 数据后使用一个差分 D/A 转换器加以扩展。D/A 的输出是 3400Hz 以下的低通滤波输出，信号再由有源 R-C 滤波器滤波以消除开关电容滤波器的能带输出。AMBE-1000 在使用时与任一 A/D 或 D/A 接口的语音采样率都是 8KHz，这8KHz 的语音数据通过串口来输入输出。串口的信号可以完全由使用者来设输入一帧 ID=0x03H定。最灵活的办法就是向令帧来设置这些信号，使用者可以对 SIOC 进行全部的配置。为了简化 A/D-D/A的配置，一组预设的配置值可以通过管脚 C_SEL20来配置。这些预设的配置值，指明了接口的方向以及可编程器件的的顺序。3 系统分析系统上电后，微器 AT89C51 复位，MC14LC5480 进行语音的 A/D-D/A转换和