Design And Implementation Of Multi-channel Speech Recorder Based On DSP

Large capacity speech recorder is very useful in the field of traffic, electronic system, public security, aviation, military and communication, etc. Tape recorder, phone recorder and harddisk controlled by PC are often used now. The characteristics of the first two devices are low capacity, short data life, and analog is being replaced by digital; those of the last one are strong, complex, heavy, high cost, low compressing rate, high storage memory command, etc. With the fast development of digital signal processing, embedded sytem and mass storage, it's meaningful and practical to develop digital portable large capacity speech recorder.The multi-channel speech recorder is developed according to user's requirements. This recorder takes TMS320VC5509 as core processor and G.729A as codec. Designs and implements hardware system, transplants μC/OS-II RTOS, customizes FAT16 file system, develops device drivers, and optimizes G.729A algorithm at this specific platform, these are 3 mainly works in this paper. How to full use of resource and combine G.729A's characteristic to optimize G.729A effectively are the importance and difficulty in this paper. First, transplants G.729A to target, then, uses hardware macro instruction, dual MAC cells, dual A cells, zero cost loop and jump, combines G.729A algorithm, takes the level of algorithm, C language, compiler and specific function scheme to optimize G.729A. After optimization, the complexity decrease from 536MIPS to 18MIPS, the performace was enhanced about 30times. On this basis, uses XDAIS~? and DSP/BIOS? to implement multi-channel codec. It also adds speech recognition to this recorder.Firstly, this paper introduces project background and relevant technologies briefly. Secondly, it introduces the whole architecture of this recorder. Thirdly, it introduces design and implement of hardware system, which including SDRAM, Flash, power module, the interface of audio VO, MMC/SD card, USB, RS232, LCD monitor and keyboard, etc. Fourthly, it introduces how to transplant μC/OS-II, customize FAT16 file system, and develop device drivers. Finally, it introduces how to optimize G.729A, implement multi-channel G.729A codec and speech recognition in this paper.In the end, the paper summarizes all the work, including of application and development of multi-channel digital recorder in the future.