AAC音訊編碼器設計與實現

Design and Implementation of AAC audio coder

黃興洋(Hsing-Yang Huang)

指導教授 ： 顧孟愷

共同指導教授 ： 賴飛羆 ； 鄭士康(Shyh-Kang Jeng)

國立臺灣大學/電機資訊學院/資訊工程學研究所/碩士(2004年)

https://doi.org/10.6342/NTU.2004.00904

摘要

AAC在所有音訊壓縮標準中提供了最高的壓縮率與品質。然而AAC編碼器的複雜度也很高。這個高複雜度主要來自於在濾波器、心理聲學模型與位元分配模組中所執行的大量運算。在這篇論文中，我們探討了快速演算法並且嘗試化減濾波器與位元分配模組的複雜度。快速的MDCT演算法利用MDCT與O2DFT (Odd Time Odd Frequency DFT)的關係，將MDCT的計算化減為N/4點的FFT計算。在位元分配模組的部分，我們應用快速初始整體增益值搜尋與雜訊估測方法發展一個新的位元分配模組。新的位元分配模組將雙迴圈架構化減為單一迴圈架構。它的複雜度相較於AAC標準中所提的位元分配模組低釵h。這些演算法可以有效地化減濾波器與位元分配模組的複雜度。相較於知名的編碼器FAAC，我們的編碼器節省了50%的編碼時間而仍然維持好的編碼品質。

關鍵字

AAC ； FFT ； O2DFT ； MDCT ； 位元分配模組 ； 雜訊估測 ； 心理聲學模型 ； 濾波器

並列摘要

AAC provides the highest compression rate and quality among all audio coding standards. However, the complexity of AAC encoder is also very high. The high complexity mainly comes from the great amount of operations performed in filter bank, psychoacoustic model, and bit allocation module. In this thesis, we study fast algorithms and try to reduce the complexities of filter bank and bit allocation module. The fast MDCT algorithm takes advantage of the relationship between MDCT and O2DFT (Odd Time Odd Frequency DFT), reduces the MDCT computation to N/4 points FFT. As for bit allocation module, we apply fast initial gain search method and noise estimation method to develop a new bit allocation module. New bit allocation module reduces the two nested loop structure to one single loop. The complexity is much lower comparing to the bit allocation module proposed in AAC standard. These algorithms efficiently reduce the complexities of filter bank and bit allocation module. Comparing to the famous coder FAAC, our coder saves 50% encoding time and still maintains good quality.

並列關鍵字

O2DFT ； Bit Allocation Module ； AAC ； Psychoacoustic Model ； MDCT ； Filter Bank ； FFT ； Noise Estimation

參考文獻

[22] F. M. Chang, “A Study on the Fast Algorithm in MPEG-2/4 Advanced Audio Coding,” Master’s thesis, National Taipei University of Technology, Taipei, Taiwan, ROC, 2004.

[4] T. Painter, “Perceptual Coding of Digital Audio,” Proceedings of IEEE, Vol. 88, No. 4, April 2000.

[7] M. Schroeder, B. S. Atal, and J. L. Hall, “Optimizing Digital Speech Coders by Exploiting Masking Properties of The Human Ear,” J. Acoustic. Soc. Amer., pp 1647-1652, Dec. 1979.

[8] J. Johnston, “Estimation of Perceptual Entropy Using Noise Masking Criteria,” in Proc. ICASSP-88, May 1988, pp. 2524-2527

[10] T. Y. Chang, “Research and Implementation of MP3 Encoding Algorithm,” Master’s thesis, National Chiao Tung University, Hsinchu, Taiwan, ROC, 2002.