为了进一步研究各种音频编码、方便算法的测试与评估，建立一个音频编码试验平台是十分必要的。作为平台构建的一部分，本文工作是用Matlab实现MPEG-1 Layer3（MP3）音频压缩标准。MPEG-1音频标准在降低比特率的前提下实现了人耳对声音的听觉上的无损压缩。MPEG-1音频部分三层：Layer1、Layer2及Layer3，提供了多种音频质量、压缩率和算法复杂度选择。Layer3在压缩率、时间复杂度上都高于Layer1、Layer2，在相同数据率的情况能获得更好的质量。本文针对Layer3做了一些研究，在此基础上，编写了包含滤波器组（PQMF），MDCT，心理声学模型，非均匀量化，分段多表霍夫曼编码等算法模块的MP3编码器的全部Matlab程序代码，并完成了部分程序调试。

In order to further study of various audio coding schemes and facilitate coding algorithm testing and evaluating, the establishment of an experimental platform for audio coding is necessary. As part of the platform construction work, this paper aimed to realize a MPEG-1 Layer3 (MP3) audio compression standard encoder by MATLAB programing. MPEG-1 Audio standard achieves lossless compression quality to human ear with quite low bit rates. MPEG-1 Audio includes three layers: Layer1, Layer2 and Layer3. It provides wide range of audio quality, compression algorithm and complexity choices. Compared to Layer1 and Layer2, Layer3 is higher in both compression ratio and time complexity. With the same data rate, Layer3 can get better quality. Research on Layer3 has been done in this paper, including the filterbanks (PQMF), MDCT, psychoacoustic model, non-uniform to quantify, sub-table Huffman algorithm. Coding work of MP3 encoder has been finished by MATLAB, and program debugging is partly completed.