MPEG解碼,音高調整及次頻編碼之音訊處理演算法研究

Abstract

本論文重點有二：音訊處理器設計以及次頻濾波之研究。在音訊處理 器設計方面，我們希望設計出一多功能音效處理器其前級為MPEG 音訊解 碼器，後級則為音調升降之處理單元。此整合設計的動機乃是有鑑在高品 質數位音視訊需求上，MPEG(Moving PicturExpert Group)標準的普遍採 用，以及音高調整功能在未來隨取視訊(Video On Demand)系統上的實用 性(如：Karaoke On Demand)。結合上述兩者功能的音訊處理器必可在未 來的多媒体音訊應用上，提供使用者較多的音訊服務選擇。論文前半將探 討MPEG 音訊標準及音調升降原理並提出我們的質疑和改進的作法，以期 在隨後提出的處理器架構上能有更好的效能及較低的複雜度。 在次頻 濾波處理的研究成果上，除了已找到將MPEG音訊中次頻合成濾波運算簡化 的演算法外，在次頻濾波器組設計上亦有新的發現。在與傳統濾波器組的 比較後我們發現此一新次頻濾波器組可大幅減少運算上的複雜度。故在處 理一般訊號時此設計亦不失為一種新的選擇。 In this thesis, new algorithms and architectures for a multifunction audio processor and design of subband filter banks are proposed. In the former, a new processor which combines functions of MPEG-audio decoding and pitch- shifting processing is aimed. The motivation of this integrated design is because of the great popularity of MPEG standard in many applications of high quality audio-video systems, and practicality of pitch-shifting in the future video- on-demand system. This new multifunction audio processor can provide users a wider choice of audio services. In order to efficiently improve the complexity and performance of processor, we concentrate our attentions on the algorithms of MPEG decoding and pitch-shifting at first. After these efforts, several new efficient algorithms for audio processing are successfully obtained . In the research of subband coding, we not only find a fast algorithm of MPEG synthesis subband filtering, but also propose a new design of 2-channel filter bank which has less complexity compared with conventional designs. This filter bank can provide a new option for general audio signal processing.