This paper presents a general software optimization technique which enables an embedded system to playWindows Media Audio (WMA) fluently without the support of floating point unit (FPU). We employ fixed-point arithmetic operations, instead of floating-point, to optimize the computational overhead during the audio decoding process. Thus the proposed performance improvements by programming in C language are useful for the implementation of the real-timeWMAcompatible decoder on ARM920T based embedded system. This work achieved performance increase by reducing the CPU usage rate from 100% to 45% with great precision on the average around 1-bit error.
淡江理工學刊=Tamkang Journal of Science and Engineering 11(4), pp.375-386