Voice Processing System Based On Tms320dm642

Speech communication plays an important roll in modern life. With the rapid development of information society, people get much more information to communicate with each other, this makes the band-width resource become increasingly precious. So, reducing the bandwidth used for transporting speech signal and the transportation rate of telephone channel becomes more and more important. The speech coding technology can greatly compress the data while ensuring the same timbre. This can spare some bandwidth for other uses.The chips specifically used for speech compression are very expensive now, besides, the devices using these chips have low flexibility and great restricts on function extension, this makes it very difficult to add new functions and algorithms to it.TMS320DM642, brought up by TI(Texas Instrument), is considered as an excellent media processing unit. DM642 is based on C64x DSP core, it uses two-level buffer and offers plenty of peripheral interfaces, including multiple channel audio serial port, Ethernet port, etc. It can be used for speech communication and developing multi-media equipments conveniently. All these characteristics make it very simple to set up a system. We only need a DM642, external memory chip (SDRAM), audio codec and Ethernet PHY chip to build up a multi-media processing platform. Meanwhile, TI offers powerful software developing environment and abundant API functions, these tools offer favorable basement for exploiting speech communication system.This project focuses on developing a speech compression system. The algorithm used is the technology of coding of speech signals at 8 kbit/s using Conjugate-Structure Algebraic-Code-Excited Linear-Prediction (CS-ACELP) that is advised by G.729 and brought forward by ITU-T. My work is setting up a speech capturing system and realizing the pre-processing of the core algorithm, the hardware used for this is TI's TMS320DM642. The speech signal is captured by microphone, and then be converted into digital signal by an application specified A/D converter and wrote to memory. Once the memory is full of data, it will trigger a EDMA event. DSP responses this event and reads the sampling data. The data is then processed, which is called pre-processing and include high pass filtering, speech separation etc. After that, the data will be put the other buffer and converted into analog signal by an application specified D/A converter, and then be output. In addition, the user can real-time spectrum analysis,windowing,storing and display on the sampling data on PC environment.The pre-processing signal can be used as the input signal for the G.729 compression algorithm. The pre-processing is a good preparation for advanced work. This design realizes an original audio signal capturing and processing system, moreover, it has a simple scheme of hardware and software structure. The D/A and A/D converters can be adjusted according to practical demands, because of its strong flexibility, this system can also fulfill complex industrial circumstances. This design provides a platform for realizing of speech enhancement, speech recognition, and speech synthesis and speech communication system based on DSP. As performance of DSP improves, the system can realize more extensive functions, and consequently be used for complex audio signal processing.The research works of this paper are listed below:Chapter one introduces presentation bankground, research contents and goal of this subject.Chapter two summarizes the basic knowledge of speech processing, speech coding technology and development, expresses basic principle of G.729 speech coding.Chapter three introduces the CPU structure, the bus's features and pipeline technique of TMS320C6000 DSP, focus on the CPU structure and external devices of TMS320DM642.Chapter four introduces operating principle of the TMS320DM642 hardware development platform, TLV320AIC23B codec chip and multichannel audio serial port McASP. Chapter five discusses the software platform on system, including the eXpressDSP software developing technique, the code developing environment, the real-time operation system, DSP device driver model, EDMA module and chip-support library.Chapter six expresses the speech collection, including the design of soft module and hardware interface, parameter configuration and device driver configuration. This chapter introduces the realization of pre-progressing and gives this subject's meaning in general aspects.Chapter seven summarizes this system and indicates the research direction of this topic.