鄭士康Jeng, Shyh-Kang臺灣大學:資訊網路與多媒體研究所陳彥廷Chen, Yen-TingYen-TingChen2010-05-052018-07-052010-05-052018-07-052008U0001-2606200911502800http://ntur.lib.ntu.edu.tw//handle/246246/180761iComper是一能理解演奏者即興時所傳達的意圖之互動爵士伴奏系統。在本論文中，我們提出一個基於隱馬可夫模型之音樂信號偵測機制，透過這個機制橋接伴奏者在聆聽音樂時，透過所學之音樂理論及聆聽感受得到的理解與即興者演奏時傳達出的意圖。我們將即興者演奏的旋律以音樂理論為基礎之特徵擷取方法取出資訊並對應至觀察現象之代號。偵測機制會使用從樂曲開始到現下小節的觀察現象序列給予最可能的狀態。透過本文提出的鼓節奏產生演算法，組合從資料庫中取出的打擊樂器樣本以及音量變化樣本得到打擊樂器之伴奏。iComper (interactive Comper) is an interactive Jazz accompaniment system which understands various soloist''s intention when he or she improvises. In this thesis, we propose an HMM-based musical sign detector to bridge accompanist''s realization when listening to music and intention given by the improviser. We apply music theory to our feature extractors on improviser''s melody and map features to observations. The detector gives the highest-probability state using the observation sequence from the start of a tune to the current measure. A re-drum algorithm is proposed to combine the chosen percussion patterns and the volume pattern to generate accompaniment.誌謝 ibstract ii要 iiiONTENTS ivISTS OF FIGURES viIST OF TABLES viihapter 1 Introduction 1 1.1 Motivation 1 1.2 Literature Survey 2 1.3 Goal of Thesis 3 1.4 Organization of Thesis 3hapter 2 Background 4 2.1 Hidden Markov Model 4 2.2 Playing a Jazz Tune 5 2.3 Tension 7 2.4 MIDI 8hapter 3 iComper System Overview 10 3.1 System Architecture 10 3.2 Components of iComper 12 3.2.1 Melody Transformer 12 3.2.2 Feature Extraction 12 3.2.3 HMM-based Musical Sign Detector 13 3.2.4 Re-Drum Algorithm 13 3.3 Modes of iComper 14 3.3.1 Learning Mode 14 3.3.2 Performing Mode 15hapter 4 HMM-based Musical Sign Detector 17 4.1 Design of Musical Sign Detector 17 4.2Pre-processing 18 4.2.1 Feature Extraction 19 4.2.2 Features/Symbol Map 21 4.3 Hidden Markov Model 22hapter 5 Re-Drum Algorithm 24 5.1 Design of Re-Drum Algorithm 24 5.2 Pre-processing 24 5.2.1 Onset Vector 25 5.2.2 Volume Ratio Vector 26 5.3 Re-Drum 26 5.3.1 Volume Pattern Selection 27 5.3.2 Percussion Pattern Selection 28 5.3.3 Percussion Generation 29hapter 6 Experiments and Discussions 31 6.1 Training the Probability Model 31 6.2 Testing Result of HMM-based Musical Sign Detector 33 6.3 Evaluation 37 6.4 Questionnaire 39hapter 7 Conclusions 43ppendix A Score of Autumn Leaves 44ppendix B MIDI Note Number Table 45eference 46application/pdf593457 bytesapplication/pdfen-US互動隱馬可夫鼓爵士伴奏interactiveHMMdrumjazzaccompanimentthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/180761/1/ntu-97-R96944009-1.pdf