標題: | MPEG-2 AAC中 TNS 及預測模組之探討 On the TNS and Prediction Modules in MPEG-2 Advanced Audio Coding |
作者: | 楊存中 Tswen-chung Yang 劉啟民 Dr. Chi-Min Liu 資訊科學與工程研究所 |
關鍵字: | 噪音分布;動態區塊切換;預測模組;音訊編碼;音訊壓縮;編碼增益;MPEG-2 AAC;Temporal Noise Shaping;TNS;Preecho;Prediction;Window switching |
公開日期: | 1998 |
摘要: | 延續了MPEG-1 layer 3的設計,在MPEG-2 AAC音訊編碼標準中,另外
加入了一些新的模組,來達到更有效率的編碼。其中主要的兩個,包
括Temporal Noise Shaping (TNS)以及預測模組。
為了消除由於時間上噪音的錯誤分布所產生的preecho現象,AAC提出
了TNS的模組來解決這個問題。過去最常用來減輕此一問題的方法是動
態區塊切換,而這個方法同樣也使用在AAC中。在本篇論文中,我們將
探討這兩種方法的特性及優缺點,以找出較有效率的消除preecho的方
式。
另外我們也發現到在AAC的原始設計中,對於TNS編碼的音框,在部分
範圍的頻線上使用預測並不會帶來編碼增益。在這篇論文的第二部分
中,將針對這個問題作深入的探討,並提出兩種新的預測模組設計方
法,以減少運算量並保持預測增益。 For higher coding efficiency, compared to MPEG-1 layer III, MPEG-2 AAC has incorporated two new modules: the "Temporal Noise Shaping (TNS)" and the "Prediction." TNS module is adopted in AAC to reduce the preecho effect, which is caused by the improper temporal noise spreading. The adaptive window switching is the usual method to ease the preecho effect. This thesis compares the TNS method and the window switching method to demonstrate the feasibility. Furthermore, in the design of AAC, we find that the prediction is not necessary for the designated frequency lines in TNS coded frames. At the second part of this thesis, this problem is discussed. We propose two new algorithms for the prediction module, which can greatly reduce the amount of calculations without loosing prediction gain. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#NT870392019 http://hdl.handle.net/11536/64039 |
Appears in Collections: | Thesis |