Porting Of MP3Encoder Algorithm For X86and ARM Platfroms

With the rapid development of internet and multimedia technology, the researchand development of audio compression and coding have been thoroughly conducted.Among all the audio compression standards, MPEG-1layer3（MP3）audio codec is themost popular and used audio compression format because of its high compression ratio,wonderful performance. Although MP3decoder algorithm has been implemented andused widely in a lot of portable and handheld devices, the MP3encoder algorithm isstill a big challenge for many havdware platforms, bacause the complexity of theencoder is much higher than that of the decoder. The purpose of the thesis is to portingthe MP3encoder algorithm to the two mostly used platforms, X86and ARM.Firstly, the audio encoding algorithm of MP3standard is introduced, in whichseveral main modules in the algorithm, including psychoacoustic model, subband filterbanks, Modified Discrete Cosine Transform (MDCT), quantization and Huffmanencoding, are detailedly analyzed. Secondly, in order to improve the encoding quality ofMP3encoder algorithm, an advanced psychoacoustic model is utilized, in addition,optimized subband filter banks and quantization and encoding are also applied. Forporting of the optimized algorithm to the X86and ARM platforms, several othermethods have been used, such as encoding algorithm structure simplification, C codelevel optimization, compiler options optimization and so on, SSE and SSE2intrinsic arealso utilized for X86platform. For porting to ARM platform, the encoder algorthimwritten in floating-point C code have to be rewrite with fixed-point C code first. Testand evaluation results have shown that the improved MP3encoding algorithm is goodin quality and its performance has been greatly optimizated on both platforms, so it canbe generally used in many applications.