The Design And Implementation Of Real-time Voice Coding System Based On ARM

Speech coding is a vital part in the communication system, and how to get high quality voice at a lower coding rate is a hot research in speech coding field. Traditional communication system uses a dedicated digital signal processor(DSP) to be responsible for the operation of voice codecs, but with the rapid development of integrated circuits, ARM microprocessor computing power is growing, it can perform per second millions of instructions, which can use it to handle digital signals. Compared with the DSP, it has low power consumption, low cost, thus being increasingly used in digital signal processing.The main contents of this dissertation is to study how to achieve real-time processing of the speech signal coding on the ARM platform. The operation of encoding the speech signal is accomplished by G.729 A algorithm, it has the advantages of a lower coding rate,a higher speech quality, a shorter delay. At the same time due to its high computational complexity, a large amount of code, making the standard G.729 A source code directly applied to the ARM platform is not feasible. Therefore, it is necessary to optimize the G.729 A standard source code and algorithms. How to optimize is related with G.729 A can run successfully on ARM platforms. To complete the system design, hardware development platform must be built. Where the main processor using Samsung S3C2440 A chip, which is a high-performance, low-power, low-cost microprocessors. Audio processor using a Philips UDA1341 TS chip, in order to simplify the design of the system. They transmit data using DMA mode by the IIS interface bus, This dissertation describes some of the key technologies used during speech coding, such as linear predictive analysis, vector quantization. Followed by a detailed analysis of the principles of G.729 A codec to provide a theoretical basis for the read G.729 A standard source. Next, the S3C2440 A and UDA1341 TS chip used by the hardware platform made structure function analysis.On this basis, to complet the hardware circuit designation. After the hardware platform has been built,to complete the software designation including in-depth analysis of the main program design, audio module driver design, G.729 A standard source code. Finally, according to the characteristic of the ARM platform, G.729 Amade a series of optimizations, including C-level optimization, assembly-level optimization algorithm-level and algorithm-level optimization. After optimization of the input speech tests to compare the coding efficiency and voice quality before and after optimization. Test results show that these optimization measures taken G.729 A present system can meet real-time encoding of the speech signal decoding operation.