MPEG的聲音壓縮方法為一種擁有高品質以及高壓縮比的失真壓縮。MPEG中有關聲音壓縮／解壓縮的音訊部份，又可分為layer1、layer2和layer3。Layer1是3個layer中最簡單的，但是其壓縮品質不佳。Layer3則有最複雜的演算法但也有最佳的音質，而且有最高的壓縮比。在取樣率(Sampling rate)上，可以是32kHz、44.1kHz或是48kHz，並且支援單聲道(monophonic)、雙聲道(dual-monophonic)、立體聲(stereo)與聯合立體聲(joint-stereo)等模式可供選擇。由於MP3高品質及高壓縮比的特性，目前已被用在許多網路傳輸上，例如多媒體通訊、視訊會議…等。
在此篇論文中，我們設計了一個MP3音訊系統。根據複雜度分析的實驗結果，MP3的壓縮與解壓縮，主要可分為兩部分：大量的控制複雜部分(例如霍夫曼編碼/反量化以及stereo processing)，以及大量計算的部分(例如poly phase analysis/synthesis filter bank, MDCT/IMDCT)。我們將大量控制的部分用軟體來實現，而大量計算的部分則設計成硬體，利用軟硬體同步設計來完成整個MP3音訊系統。我們選擇ARM發展平台來實現與驗證整個音訊系統。ARM發展平台為一個完整的SoC設計環境，使用ARM匯流排來當作整個環境的傳輸協定。經過我們測試的結果，一首630kB的音訊檔(wave 格式)，在純軟體的情況下，壓縮成MP3，需要26分鐘43秒，但經過軟硬體同步設計之後，只需要一分鐘48秒，大大提升了系統的效能。

The MPEG audio compression is a lossy audio compression method, which provides high quality audio under high compression ratio. MPEG 1 audio standard provides three layers, layer I, layer II, and layer III. Layer I is the simplest but low audio quality, while layer III have the most complexity but offers the best audio quality. The MPEG audio standard offers four channel modes, which can be one of mono mode, dual channel mode, stereo mode, or joint stereo mode. The sampling frequency could be 32kHz, 44.1kHz, or 48kHz. Because of the high compression ratio and high sound quality, the MPEG1 layer III(MP3) has been used by network users as a sound compression method in many multimedia applications nowadays.
In this thesis, we design a MP3 audio system. From experiment results of the complexity analysis, MP3 encoder/decoder system could be divided into two parts, high control complexity part (such as Huffman coding, quantization/invert quantization, and stereo processing) and high computation complexity part (such as poly phase analysis/synthesis filter bank, MDCT/IMDCT). We implement the high control complexity part by software, and use hardware to implement the high computation complexity part. Finally, we apply software/hardware co-simulation to verify the entire MP3 codec system. We choose ARM integrator to implement and verify the MP3 system. ARM integrator is a complete environment for SoC design, which uses ARM bus as the transform protocol. In the software implementation, it requires 26 minutes and 43 seconds to compress an audio source(wave format) with 630kB to a MP3 format. However, it only requires 1 minute and 48 seconds after hardware/software co-design. The experiment results show it improves the system performance.

[ 1] “AMBATM specification rev2.0”.
[ 2] ARM INTERGRATOR/AP USER GUIDE
[ 3] ARM INTEGRATOR/CM920T-ETM INTEGRATOR/CM940T-ETM USER
GUIDE
[ 4] ARM INTEGRATOR/LM-XCV600E+ INTEGRATOR/LM-EP20K600E+
USER GUIDE
[ 5] ARM MULTI-ICE version 2.2 USER GUIDE
[ 6] ARM DEVELOPER SUITE version 1.1 COMPILER,LINKER, AND
UTILITIES GUIDE
[ 7] ARM FIRMWARE SUITE version 1.3 REFERENCE GUIDE
[ 8] ARM DEVELOPER SUITE version 1.1 CODEWARRIOR IDE GUIDE.
[ 9] ADS_AssemblerGuide
[10] ADS_DebugTargetGuide
[11] D. Pan. “A tutorial on MPEG/audio compression”, IEEE Multimedia, Vol. 2, No.2, Summer 1995.
[12] D. Y. Chan, J. F. Yamg and C.C. Fang. “Fast implementation of MPEG audio coder using recursive formula with fast discrete cosine transforms”, IEEE Trans. speech and aud io processing, Vol.4, No.2, pp144-148, March 1996.
[13] http://www.arm.com/
[14] ISO/IEC 11172-3. “Coding of moving picture and associated audio for digital storage media at up to about 1.5Mbit/s — Part 3 audio”, November 1991.
[15] ISO/IEC 13818-3. “Generic coding of moving pictures and associated audio:audio”, May 1994.
[16] J. M. Jou, K. M. Wu. “Research and development of a hierarchical interface design methodology and models for SoC IP integration”, department of electrical engineering NCKU, Tainan, Taiwan, R.O.C., June 2001.
[17] J. M. Jou, Jun-Sheng Zheng. “Hierarchical Interface Design Methodology: Using Real-Time MPEG1 Audio layer3 codec as a case”, department of electrical engineering NCKU, Tainan, Taiwan, R.O.C., June 2002.
[18] K. Brandenburg and H. Popp. “An introduction to MPEG Layer-3”, EBU technical review, June 2000.
[19] P. Singh, W. Moreno, N. Ranganathan and H. Neinhaus. “A flexible MPEG audio decoder layer III chip architecture”, IEEE 1998.
[20] R. Rangachar. “Analysis and improvement of the MPEG-1 audio layer III algorithm at low bit-rates”, Arizona state University, December, 2001.
[21] S. Shlien. “Guide to MPEG-1 audio standard”, IEEE Trans. on Broadcasting, Vol.40, No.4, December 1994.
[22] T. H. Tsai and Y. C. Yang. “A novel architecture design for MP3 audio decoder”, in the 11th VLSI Design/CAD Symposium, pp16-19, August 2000.
[23] T. Sakamoto, M. Taruki, and T. Hase. “A fast MPEG audio layer III algorithm for 32 bit MCU”, IEEE Trans. on Consumer Electronics, Vol. 45, No. 3, August 1999.
[24] T. Uzelac and M. Kovac. “A fast MPEG audio layer III software decoder”, IEEE 1998.
[25] Vladimir B. and K. R. Rao, “ An efficient implementation of the forward and inverse MDCT in MPEG audio coding”, IEEE signal processing letters, Vol.8,
No.2, February 2001.
[26] Y. H. Fan, V. K. Madisetti, and R. M. Mersereau. “On fast algorithm for
computing the inverse modified discrete cosine transform”, IEEE 1999.