Design And Implementation Of A Multi-channel Stereo Audio Encoding System

Digital audio coding is one of the most important techniques used in the field ofmultimedia processing. It compresses the original audio data by removing variouskinds of redundant information, and makes it easier to store or transmit audio data.Digital audio coding is widely used in daily applications like DAB (Digital AudioBroadcasting), digital TV systems, remote TV and telephone conferences and so on.In these applications, the encoders often come across the task of processing multipleinputs at the same time. This could be quite a challenge if there are numbers of inputsand each input consists of multiple stereo channels.In recent years, embedded processors like DSP (Digital Signal Processor) havebeen widely used in all kinds of signal processing equipment due to their highperformance in digital signal processing and low power consumption. The researchand implementation of real-time audio encoding systems based on DSP platform canmake audio capturing and storage more efficient. The audio encoders can also be usedin other coding systems working together with video encoders to produce digital TVprograms.The thesis implements a real-time multi-channel stereo audio encoding systembased on TMS320C6727from TI. The system is capable of encoding PCM samplesfrom multiple inputs and packetizing the bitstream to PES format as the output. Thethesis first analyses the coding framework of state-of-the-art audio coding standardand basic audio encoding algorithms, then describes the structures and functions ofmain modules in TMS320C6727, and introduces the optimization methods used inporting MP2encoding algorithms to DSP platform. The procedures of receiving PCMdata and sending PES packets through McASP and SPI interfaces with the help ofdMAX module are fully described. In the last section of the thesis, a method ofresponding multiple interrupts is proposed and the way of booting from external flashdevices is also described.The proposed audio encoding system is capable of processing4-way inputs oftotally8channels in real time. The encoding parameters used are a sampling rate of48KHz, a MP2bitstream format and an encoded bit rate of128Kbps.