The Research And Implementation Of A Speech Sampling And Processing System Based On DSP

The purpose of this project is to complete the design of a real time voice signalmonitoring and processing system in embedded hardware platform to meet the needs ofsecurity equipment needs. Based on the G.729A voice compression protocols, this articlemake a thorough research on the multimedia remote monitoring communication systemfor voice signal acquisition and compression subsystem.First, according to the demand characteristics of the subject, select the appropriatehardware platform.This paper uses a downhole monitoring equipment DM642corechip.Focuses on the handling of multimedia audio and video capture and playback voicesubsystem design and implementation of hardware-related.In-depth study and DM642audio chip peripherals MACSP TLV23interface applications. In addition, outside thesystem as needed for the expansion of the data memory and program memory，these areall the basis for achieving high performance of DM642.Second, this article introduced the related theory of speech coding, and thedecoding algorithm principle and realization process were analyzed in detail around thekey technology of the G.729A agreement, which researched the software algorithmrealization of process deeply combined with G.729A decoding theory. Transplante it intoDSP development environment CCS platform, and then kept the decoding algorithmprocedures had a multi-level optimization, made it suitable and efficient operation of theDM642platform.And then, according to the specific needs of mine-equipment, the RF5framework isused to develop DSP software. Which is recommended by TI company.RF5frameworkincluding real-time operating system DSP/BIOS and driver development kit DDK, youcan do to maximize code reuse, shorten the development cycle, especially formulti-threaded, multi-channel and multi-algorithm structure of high-density applications,easy system expansion and upgrades. First, voice driver is designed based on DDK, andG.729A codec modules for the standard algorithm is packaged, then using SCOMcommunication mechanism to complete voice codec system’s multi-task programming.The system closed-loop test well.Optimized G.729A algorithm has a good evaluation ofsubjective and objective.The reconstructed speech has high intelligibility, each module is operating normally, the overall system is stable. The optimized G.729A algorithmcode not only achieve real-time operation based on DM642platform, but also takes upless memory and CPU resources.The system is fully able to achieve multi-channel audiocodec.Finally, the article summed up the above work done, and according to the subjectapplication characteristics, putted forward further need to perfect the system of theplace.