| MP3 is the third audio proto compression mode of MPEG-I. The characteristics of high compression ratio and low sound quality distortion make MP3 to be the most prevalent audio format on PC,network and PDA currently. Owing to its popularity of application, the research on the topic of MP3 is quite comprehensive .There are two common techniques in realizing MP3 decoding in the embedded system.Hardware Decoding is to decode through the special MP3 decoder hardware, which has an advantage of high performance and low power consumption. But it has a disadvantage of low extensibility and lack of the general handle ability. Currently most of the commercial MP3 players use such technique. The other one is Software Decoding, which completely relys on the microprocessor or DSP processing ability to implement MP3 decoding, by means of MP3 fixed-point or floating-point decode program. It requires quite high performance for microprocessor or DSP, especially for embedded system of multi-task processing.The essay stands at the view of Software decoding to implement MP3 decoding in multi-task platform. The multi-task platform means to implement single task while occupying less processor resources, which requires the high efficiency of MP3 decoder. The operation amount of floating-point code is quite high and the single task is hard to meet the requirement of real-time decoding in most embedded systems. So it is quite essential to transit MP3 floating-point to fixed-point code and optimize the fix-point code. The essay rigorously analyzes the transission of floating-point to fixed-point, which will minimize the error in the process of modification, and thus guarantee the excellent quality of decoder. Optimization includes not only the algorithm level and high language level, but also the assembly level aiming at the processor of MCF5329. The operation amount after optimization of MP3 is just one sixteen of the original one, which greatly reduces the difficulty of the algorithm.The MP3 decoding algorithm has passed many tests and been implemented in target board MCF5329EVB finally. The test results indicate that the decoding performance upgrades dramatically. |