MPEG Layer 3 (MP3)是目前最為普遍的數位音訊壓縮格式，不管是在一些硬體的裝置或是軟體上的應用。最近，MP3已經成為個人數位音樂娛樂的代名詞。許多SOC (system-on-chip)系統都加入MP3當作它的附加功能，例如一些手機或隨身的裝置。市面上也有許多MP3 player的產品在販賣。由於考慮到市場的快速需求，目前一般的做法會採用DSP或是RISC processor的方式去實現。但是採用這種做法並不是一個最佳化的方式，因為它們並不是針對MP3所做的硬體設計，所以會有些多餘的電路，造成較大的功率消耗和面積。另外過多的記憶體使用量也是一個問題。 因為一些手攜式裝置上的應用，功率消耗和產品的成本變得格外重要。記憶體的使用就必須很小心規劃，因為記憶體的使用相當的耗成本以及功率消耗。我們採用了完全ASIC的方式去實現MP3解碼器，可以改善processor所產生的問題，使我們的設計達到低功率和低成本以適用於一般消費性的裝置中。 除了以ASIC做設計所產生的優點，透過仔細觀察MP3的解碼流程，我們可以進一步做改善。在本論文中，針對我們的MP3解碼器架構提出了一些降低功率的技術以及採用最精簡的記憶體使用量。 MPEG Layer 3 (MP3) is the most popular audio compression format in the world for both hardware-based devices and software-based applications. Presently, MP3 has turned into a synonym for personalized music entertainment for millions of people. Based on the consideration of fast time-to-market, a general-purpose DSP or RISC processor is the common implementation approach for MP3 decoder. Since the hardware is not dedicated for MP3 application, some architecture parts in the processor are not utilized completely. The cost of each product is relatively high and the power dissipation is also a problem. Since the MP3 decoder is targeted to fit into a small portable, it is necessary to minimize the power consumption and cost. Identically, it is always reasonable to reduce the memory requirements since memory is expensive and consumes power. By use of the pure-ASIC approach, we can provide a consumer-economical solution for MP3 audio decoder with the advantages of low-cost and low-power design. Instead of the benefits obtained from dedicated hardware design, it still exists some improvements by well analysis on the individual features of MP3 decoding. This thesis is targeted on the architecture implementation with some proposed techniques to achieve a low power and memory-optimized design.
