- 學位論文
MP3轉為AAC的快速音訊格式轉編碼
Efficient Audio Format Transcoding from MP3 to AAC
摘要
數位音訊壓縮技術在過去20年來已經大大改善，新的和更有效的計劃預計在未來融入生活中。因此，轉換舊格式的音訊資料到新的或目前的格式，對於維護現有的音訊檔案有其必要性。這項研究的主要目標是把流行的MP3格式的音訊檔案，轉換成更有效地AAC格式。我們首先研究了這兩種格式的編碼方法。接著，我們尋找更有效的手段，以更快速及準確的方式進行格式轉換。 MP3和AAC編碼格式是失真式資料壓縮。一個簡單的轉換方法是將MP3檔案解碼回到原始波形資料，然後使用AAC的編碼器編碼波形資料。在第二次編碼時，由於人類心理聲學模型參數使用了經過MP3壓縮後的數據，會造成更大的誤差。因此，直接轉換的方法可能會導致品質下降。這一個方法的另一個缺點是昂貴的計算時間，因為它需要完全的解碼和壓縮過程。本篇論文提出了一個方法，提取MP3解碼過程的資訊並重用於編碼上，因此我們可以略過一些AAC編碼過程中較龐大的計算步驟。我們執行這個概念並實作和測試，於實驗結果中表明，該方法可以降低計算複雜度，並維持編碼後的音訊品質。
並列摘要
The digital audio compression technique has been significantly improved in the past 20 years and new and more efficient schemes are expected to emerge in the future. Thus, the conversion of an old format audio to a new and current format becomes essential in preserving the existing compressed audio files. The main objective of this study is to convert the popular MP3 format audio to the more efficient AAC format. We first study the encoding methods of these two formats. Then, we look for more efficient means to do the format conversion in a fast and accurate manner. Both MP3 and AAC encoding formats are lossy data compression. A straightforward conversion scheme is decoding the MP3 file into the waveforms and then encoding the waveforms using an AAC coder. During the second encoding, the human psychoacoustic model parameters are extracted from the MP3 data, which are distorted. Therefore, this direct conversion scheme may lead to a decline in quality. Another drawback of this scheme is computationally expensive because it requires a full decoding process and a full compression process. In this thesis, we propose a structure that reuses the information extracted in the MP3 coding process, and thus, we can skip a few heavy-computational steps in the AAC coding process. We implement this concept and our experimental results show that this method can reduce the computational complexity and, in the mean time, increase the final coded audio quality.