The Design And Optimization Of Multimedia Audio System Based On Garfield

Usually, two methods to realize multimedia audio prcdess in embedded system are used. First, we can rely on embedded CPU to realize coding of audio data by softeare. Second, coding computation can be implemented throuth application specific hardware. In the first situation, it depends on the high performance of CPU. In the second situation, though specific hareware can get higher performance and lower power, but poorer function extending abbility.In this paper, speech and MP3 subsystem based on Garfield use UI-SERVER program mode of ASIXOS to realize user control functions. This paper focus on speech process performance analysis and optimization.This optimization is based on algrithem, code and the on-chip-memory, which greatly enchanced the performance of the speech system. Additonal, father optimization on code level for MP3 subsystem is done. Although it has reached real-time through many methods before this paper, it can be reference for the last research. Meanwhile, we deeply introduce the design of audio system on ASIXOS platform from application, server and driver programme level and application in multi function phone system. At last, test results of system are given. When system clock is 70MHZ, speech system can work at 29.92KBps and 33.98KBps for encoding and decoding and MP3 system can decode a fram in 23.82ms. Audio system has been real-time.By now, Garfiled has been taped out and produced through MPW, and the audio system has been designed. Now the system can fluently record speech or play MP3 music on the prototype of multi function phone, which satisfies the requirements of the project.