進階音訊編碼器之快速位元分配演算法

進階音訊編碼器（Advanced Audio Coding，AAC）是MPEG（Moving Pictures Expert Group）於1997年所提出的，其標準格式定義在國際標準組織（ISO/IEC 13818-7）中，高效能進階音訊編碼器（High Efficiency AAC，HE AAC）是將AAC與頻帶重現（Spectral Band Replication，SBR）兩種技術作結合，第二版高效能進階音訊編碼器（High Efficiency AAC v2，HE AAC v2）則是又加入了第三種技術：參量立體聲（Parametric Stereo，PS）。本文以HE AAC v2為基礎提出兩種快速位元分配方法，單階段快速位元分配演算法（One-Stage Fast Bit Allocation Algorithm for Advanced Audio Coding，簡稱為AAC 1-stage FBA）與雙階段快速位元分配演算法（Two-Stage Fast Bit Allocation Algorithm for Advanced Audio Coding，簡稱為AAC 2-stage FBA），這兩種方法是利用生理聽覺遮罩（Psychoacoustic Masking）快速算出比例係數（Scalefactor）以完成位元分配。實驗中的音訊資料庫共分成12類，其中包含九種樂器獨奏、交響樂、及男女聲演唱。實驗結果顯示，AAC 1-stage FBA演算法在計算比例係數時平均減少了48 ％的時間。此兩種快速位元分配方法也與其他音訊編碼器如MP3，AC-3，Ogg Vorbis等所產生的音訊品質做比較，由客觀及主觀評估的結果顯示，AAC 1-stage FBA演算法在位元率高於64 kbps時品質比MP3差，低於64 kbps時品質比MP3好，比較特殊的是男聲及長笛，不管位元率是多少皆優於MP3。AAC 2-stage FBA演算法則呈現出如同HE AAC v2平滑下降後的品質，因此在任何位元率下皆優於MP3、AC-3編碼器。

關鍵字

進階音訊編碼；頻帶重現；參量立體聲

並列摘要

The MPEG Advanced Audio Coding (AAC) was first proposed in 1994. Its standard form is defined by the International Standards Organization. The High Efficiency AAC (HE AAC) combines two kinds of technologies, AAC and Spectral Band Replication (SBR). The version two High Efficiency AAC (HE AAC v2) further includes the third kind of technology, Parametric Stereo (PS). Two fast bit allocation algorithms based on psychoacoustic masking for determining scalefactors of the HE AAC v2 are proposed in this thesis. One is the One-Stage Fast Bit Allocation Algorithm called AAC 1-stage FBA. The other is the Two-Stage Fast Bit Allocation Algorithm referred to as AAC 2-stage FBA. The audio database used in the experiments is divided into 12 categories, including 9 kinds of musical instruments, symphony, and males and females singing. The experimental result shows that the AAC 1-stage FBA algorithm can reduce 48 % of the time consuming for scalefascors calculation. Both fast algorithms are also compared with other audio codecs, such as MP3, AC-3, and Ogg Vorbis. The objective and subjective tests show that the performance of the AAC 1-stage FBA algorithm is worse than that of MP3 when bit rates above 64 kbps. But it performs better than MP3 if bit rates below 64 kbps. Male singing and flute music are special cases for the AAC 1-stage FBA algorithm since it always performs better than MP3. The AAC 2-stage FBA algorithm performs graceful degradation of the HE AAC v2 for it is clearly superior to MP3 and AC-3 codecs at any bit rates in the experiments.

並列關鍵字

Advanced Audio Coding ； Spectral Band Replication ； Parametric Stereo

參考文獻

[1] K. Brandenburg, “MP3 and AAC Explained,” presented at the 17” AkS Conference on High Quality Audio Coding, Florence 1999.

[6] G. A. Soulodre, M. Lavoie, “Subjective Evaluation of MPEG Layer II with Spectral Band Replication,” 117th AES Convention, San Francisco, CA, USA, Oct. 2004.

[11] J. Princen, A. Johnson, A. Bradley, “Subband/Transform Coding Using Filter Bank Designs Based on Time Domain Aliasing Cancellation,” IEEE ICASSP pp. 2161 – 2164, University of Surrey, Guildford, Surrey U.K, 1987.

[13] T. Painter and A. Spanias, “Perceptual coding of digital audio,” Proc. IEEE, vol. 88, pp. 451–513, Apr. 2000.

[20] N. S. Jayant-Peter Noll, Digital Coding of Wave forms, Prentice Hall Inc., Englewood Cliffs, New Jersey, 1984.

國際替代計量

進階音訊編碼器之快速位元分配演算法

未授權

主題瀏覽