顧孟愷臺灣大學：資訊工程學研究所吳明禧Wu, Ming-HsiMing-HsiWu2007-11-262018-07-052007-11-262018-07-052004http://ntur.lib.ntu.edu.tw//handle/246246/53825要實作波形表合成法是很簡易的，但是要做到最佳化並不容易。其中主要用到的參數是由預錄好的聲音訊號所取出來的。如何選擇一些能夠完全表示出音樂訊號特性的樣本是此系統一個很重要的議題。 分析與合成系統一般被用在心理聲學研究與音樂訊號的合成上,它也被用於研究各種樂器的音色，將聲音分解成少數幾個基本成分再由這些成分合成聲音。我們可以將合成出來的聲音與原來的聲音做比較，來判斷這些基本的部分能否完全的表示原來的聲音。 在本論文中，我們提出了一個聲音分析合成系統。此系統將輸入訊號切割成一個一個的區塊，而此種兼具時間與頻率特性的表示法可以完全的表現樂器音色的演變。此外，觀察音樂訊號中頻率的變化量可以指引我們了解欲選取樣本的分佈情形。此分析合成系統是個強而有力的聲音訊號處理工具，它可以處理各種不同的音樂訊號，並找出其中關鍵的區塊。並在合成訊號時，使用Phase Vocoder來重建原本的訊號。Wavetable synthesis is straightforward in implementation but subtle in optimization. The main parameters of wavetable synthesis are extracted from a recorded tone. Selecting samples which can completely exhibit the varying characteristics of musical signal is the most important subject of this system. Analysis/synthesis systems are often used for psychoacoustic research and music composition. They have been used in studies on instrument timbre to reduce sounds to basic elements and then synthesize sounds from these elements. We can then compare the synthesized sound to the original to determine if the basic elements completely describe the sound. In this thesis we present a sound analysis and synthesis system that incorporates knowledge of the auditory system. It divides the input signal into frames and exhibits the evolution of timbre by a time-frequency representation. The spectrum variation can give guidance to understand the distribution of selected samples. The analysis/synthesis system is a power sound processing tool: it can deal with various music signals and find some “key frames”. When synthesizing, Phase Vocoder is applied to reconstructs the original signal.Abstract i Contents ii List of Figures v List of Tables vii Chapter 1 Introduction 1 1.1 Motivation and Objective ……………………………3 1.2 Thesis Organizations …………………………………4 Chapter 2 Digital Sound Syntheses 5 2.1 Some Digital Sound Syntheses…………………………5 2.1.1 Additive Synthesis …………………………………6 2.1.2 Subtractive Synthesis ………………………………7 2.1.3 FM Synthesis …………………………………………7 2.1.4 Physical Modeling Synthesis ………………………8 2.2 Wavetable Synthesis …………………………………9 2.2.1 Looping………………………………………………….9 2.2.2 Envelope Generation…………………………………12 2.2.3 Pitch Shifting……………………………………….13 2.3 Sample Selection of Wavetable Synthesis…………15 Chapter 3 Short Time Fourier Analysis 17 3.1 Sound and Ear ………………………………………….17 3.1.1 Pitch ………………………………………………….18 3.1.2 Harmonic Spectrum……………………………………19 3.1.3 Timbre …………………………………………………20 3.2 Discrete Fourier Transform …………………………23 3.3 Short Time Fourier Transform ………………………24 3.3.1 Selection of Analysis Window…………………….27 3.3.2 Selection of FFT Length ………………………….30 3.3.3 Selection of Hop…………………………………….31 3.4 Phase Vocoder……………………………………………32 Chapter 4 Analysis and Synthesis Model 37 4.1 Analysis Procedure ……………………………………37 4.1.1 Digital Filtering……………………………………38 4.1.2 End Point Detection…………………………………38 4.1.3 Period Detection ……………………………………40 4.1.4 Envelope Extraction ………………………………42 4.1.5 Spectrum Analysis……………………………………44 4.2 Synthesis Procedure……………………………………46 4.2.1 Synthesis Using Phase Vocoder……………………47 4.2.2 Envelope Normalization and Generation…………48 Chapter 5 Results and Discussions 49 5.1 Implementation …………………………………………49 5.2 Criteria …………………………………………………50 5.3 Experiment Results…………………………………….51 5.3.1 Piano……………………………………………………51 5.3.2 Violin …………………………………………………56 5.3.3 Trumpet…………………………………………………59 5.4 Reduction Ratio…………………………………………62 Chapter 6 Conclusions 64 References 65419947 bytesapplication/pdfen-US分析合成系統波形表合成法wavetable synthesisanalysis/synthesis systemthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/53825/1/ntu-93-R91922077-1.pdf