郭斯彥臺灣大學：電子工程學研究所謝均毅Hsieh, Chun-YiChun-YiHsieh2007-11-272018-07-102007-11-272018-07-102006http://ntur.lib.ntu.edu.tw//handle/246246/57395Digital audio coding technology has played an important role in our daily life for entertainment and communication .Focusing on audio signal compression rate, we ignore the problems of calculation complexity, and then AAC standard was set. AAC provides the highest compression rate and quality among all audio coding standard. However, the complexity of AAC encoder is very high and it comes from the psychoacoustic model and the filter bank. In this thesis, we provide a new view to simplify the filter bank. AAC uses MDCT module for filter bank. To simplify the complexity of MDCT, we try to deal with the audio signals in time domain, and then send them to the filter bank. Since the properties of MDCT, the operation of the signal in time domain will get the certain response in frequency domain. So we can also simplify the complexity of MDCT by this method.Chapter 1 Introduction 1 1.1. Background and Motivation ___________________________1 1.2. Thesis Outline_______________________________________2 Chapter 2 Perceptual Coding 3 2.1. Psychoacoustics______________________________________3 2.2. Absolute Threshold of Hearing________________________3 2.3. Critical Bands_______________________________________4 2.4. Masking Effect_______________________________________5 2.4.1. Simultaneous masking______________________6 2.4.2. Temporal masking__________________________8 2.5. Psychoacoustic Model 2_______________________________9 Chapter 3 AAC Algorithms 15 3.1. AAC overview________________________________________15 3.2. Filter Bank_________________________________________16 3.2.1 MDCT and IMDCT_____________________________________17 3.2.2 Window Shape Adaptation____________________________19 3.2.3 Block Switching____________________________________20 3.3. Temporal Noise Shaping______________________________23 3.4. Prediction__________________________________________25 3.5. M/S Stereo and Intensity Encoding___________________25 3.6. Quantization________________________________________26 3.6.1 Nonuniform Quantization____________________________27 3.6.2 Iteration Process_________________________27 3.7. Noiseless Encoding__________________________________31 3.7.1 Spectrum Clipping__________________________________32 3.7.2 Sectioning_________________________________________32 3.7.3 Grouping and Interleaving__________________________33 3.7.4 Scalefactors_______________________________________33 3.7.5 Huffman Coding_____________________________________34 Chapter 4 Implementation and Discussion 36 4.1. Another Form of MDCT________________________________36 4.1.1. Introduction_____________________________36 4.1.2. Discussion________________________________________40 4.2. Linear Combination of MDCT _________________________42 4.2.1. Introduction_____________________________42 4.2.2. Discussion________________________________________45 Chapter 5 Conclusion 48 5.1. Test and Result ____________________________________48 5.1.1. New Form of MDCT__________________________________49 5.1.2. Linear Combination of MDCT________________________50 5.2. Conclusion__________________________________________51 Appendix A 54 Appendix B 601316126 bytesapplication/pdfen-US音訊壓縮濾波器aacfilter bankmdctthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/57395/1/ntu-95-R93943083-1.pdf