Research On MPEG-4 AAC Encoding Technology And Implement Of Fixed-point DSP Program

The coding standard of MPEG-4 AAC (Advanced Audio Coding) represents the world's most advanced audio coding technology. It will be widely used in many different areas such as digital audio broadcast, digital voice storage, home theater and network transmission. DSP (Digital signal processor) is used in real-time processing areas. The fixed-point DSP is widely used, running at high speed and with high cost-effectiveness considerations. According to their advantages, the implement of the AAC real-time encoder on fixed-point DSP has practicality.The AAC algorithm is very complicated. To improve the encode effect is a key issue to implement the AAC real-time encoder. We research on MPEG-4 AAC encoding standards and the algorithm principium. The key point is optimizing the psychoacoustic model module, filter banks module and quantizer module. These modules'demand for computation and memory capacity is high, and their algorithm is very complicated. The fast algorithm is used to replace existing algorithm in the psychoacoustic model, and the quantification module and the window function use the method of combining lookup in table and a small amount of calculation. In theory, it can effectively reduce the computational complexity.Only the fixed point calculation can be do on Fixed point DSP. The other key issue to implement the encoder on fixed-point DSP is converting the floating-point data from floating-point notation to fixed-point notation. Different operations have their own characteristics, so different solutions are used in changing the float-point operation to fixed-point operation. Then we use the method of replacing the modules one by one to achieve the whole program's fixed-point calculation.The compare, which is between the program before the optimization procedure and the one after optimization, shows that the optimization is effective. It decreases the encode time obviously. The performances of the fixed-point program and the floating-point program are tested. In theory and in practice, it demonstrates that the fixed-point program can provide high compression ratio and high reconstruction audio signal quality at the same time.