標題: HE-AAC編碼器上之有效率的Time/Frequency Grid 設計
Efficient Design of Time/Frequency Grid in HE-AAC Encoder
作者: 唐守宏
Shou-Hung Tang
劉啟民
李文傑
Chi-Min Liu
Wen-Chieh Lee
資訊科學與工程研究所
關鍵字: 頻寬擴展;時頻格子;重建高頻訊號;頻帶摺疊;HE-AAC;SBR;Spectral Band Replication;Time/Frequency Grid;spectral envelope coding
公開日期: 2005
摘要: SBR (Spectral Band Replication) 編碼技術主要是一種頻寬擴展工具,並且和AAC結合,這種新的壓縮技術稱為MPEG-4 High Efficiency (HE) AAC。在HE-AAC中,經由SBR負責高頻的訊號壓縮,AAC就可以利用大部分的位元數來壓縮低頻的訊號。從訊號的相似性來看,SBR的基本精神就是根據高低頻訊號之間的相似性,利用複製低頻訊號來重建高頻訊號。時頻格子 (Time/Frequency grid) 主要在決定重建訊號的單位,並且是HE-AAC中的核心模組。時頻格子主要包含了頻率表格的決定、時間區塊的決定,每個時間區塊的頻率解析度,以及每一個訊號框的格式。其中,頻率表格的選取以及時間區塊的大小決定了重建訊號的品質以及所花費的位元數;訊號框格式則是會影響訊號框中時間區塊的數目以及分佈型態。因此,時頻格子同時影響了品質以及所花費位元個數。 要成功的設計出一個有效率的時頻格子機制,必須要同時考慮到品質以及位元的影響。因此,在本篇論文中提出了一個衡量品質的標準,然後將時頻格子的決定轉化成格子狀的搜尋問題,最後,提出一個有效率的搜尋法來來找出最佳的解答。除此之外,也在搜尋法中加入所消耗位元數目的考量。在品質的量測上,主觀跟客觀的測試都會拿來驗證本篇論文所提出的設計,客觀的測試是採用ITU所發展的PEAQ測試系統來評估音訊壓縮後的誤差程度。
Spectral Band Replication (SBR) has been combined with MPEG AAC as bandwidth extension tool. The resulting scheme is referred to as the MPEG-4 High Efficient (HE) AAC or AACplus. With SBR module taking care of the high frequency contents, the conventional AAC encoder can compress the low frequency part using most of the available bits. From the similarity between the low and high bands, SBR reconstructs the high bands by replicating the low bands. In the SBR, the time-frequency (T/F) grids deciding the replication unit in high frequency bands are the kernel module in SBR. Frequency table decision, frame class decision, time segments and associated frequency resolution decision are the main design issues in T/F grid. The chosen frequency table and number of time segments determine the quality and consumed bits of reconstructed audio. Frame class restricts the number of time segments and the distribution of time borders. Therefore, the design of T/F grid should be greatly involved with quality and consumed bits. This thesis proposes an approach that formulates the decision of the T/F grid into a trellis-lattice search problem and presents an efficient search algorithm to find the optimum path. Both subjective and objective tests are conducted to check the quality improvement over existing methods. The objective test measures used is the recommendation system by ITU-R Task Group 10/4.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT009317582
http://hdl.handle.net/11536/78793
Appears in Collections:Thesis


Files in This Item:

  1. 758201.pdf