From content-based music emotion recognition to emotion maps of musical pieces

목차 (Table of Contents)

• CONTENTS
• 1 Introduction = 1
• 1.1 Motivation = 1
• 1.2 Organization of This Book = 2
• Part Ⅰ Emotion in Music
• 2 Representations of Emotions = 7
• 2.1 Perception of Emotions = 7
• 2.2 Categorical Approach = 7
• 2.3 Dimensional Approach = 10
• 2.4 Summary = 11
• 3 Human Annotation = 13
• 3.1 Introduction = 13
• 3.2 Length of a Musical Segment = 13
• 3.3 Audio Music Data = 15
• 3.3.1 Data Set = 15
• 3.3.2 Music Experts = 16
• 3.3.3 Annotation Process = 16
• 3.3.4 Results = 17
• 3.4 MIDI Music Data = 20
• 3.4.1 Data Set = 20
• 3.4.2 Hierarchical Emotion Model = 21
• 3.4.3 Annotation Process = 22
• 3.4.4 Results = 22
• 3.5 Summary = 23
• Part Ⅱ Emotion Detection in MIDI Files
• 4 MIDI Features = 27
• 4.1 Introduction = 27
• 4.2 MIDI Features Tools = 28
• 4.3 Description of MIDI Features = 28
• 4.3.1 Rhythm Features = 29
• 4.3.2 Harmony Features = 33
• 4.3.3 Harmony-Rhythm Features = 38
• 4.3.4 Dynamic Features = 40
• 4.4 Conclusions = 40
• 5 Hierarchical Emotion Detection in MIDI Files = 43
• 5.1 Introduction = 43
• 5.2 Related Work = 43
• 5.3 MIDI Music Data = 44
• 5.4 Feature Extraction = 45
• 5.5 Construction of Classifiers = 46
• 5.5.1 First Level Classifiers = 46
• 5.5.2 Second Level Classifiers = 49
• 5.6 Hierarchical Classification = 50
• 5.7 Emotion Tracking in MIDI Files = 51
• 5.7.1 System Construction = 51
• 5.7.2 Musical Composition Segmentation = 52
• 5.8 Results of Emotion Tracking in MIDI Files = 53
• 5.8.1 Emotion Histograms of Musical Compositions = 53
• 5.8.2 Emotion Maps = 54
• 5.8.3 Quantity of Changes of Emotion = 56
• 5.9 Conclusions = 57
• Part Ⅲ Emotion Detection in Audio Files
• 6 Audio Features = 61
• 6.1 Introduction = 61
• 6.2 Features from Audio Analysis Tools = 61
• 6.2.1 Essentia = 62
• 6.2.2 Marsyas = 63
• 6.3 Selected Audio Features and Labeled Emotions = 63
• 6.3.1 Timbre Features = 64
• 6.3.2 Rhythm Features = 70
• 6.3.3 Tonal Features = 71
• 6.4 Conclusions = 74
• 7 Detection of Four Basic Emotions = 75
• 7.1 Introduction = 75
• 7.2 Related Work = 75
• 7.3 Music Data = 77
• 7.4 Feature Extraction = 78
• 7.5 Results = 78
• 7.5.1 Construction of One Classifier Recognizing Four Emotions = 78
• 7.5.2 Construction of Binary Classifiers = 79
• 7.5.3 Evaluation of Different Combinations of Feature Sets = 81
• 7.5.4 Emotion Maps = 81
• 7.6 Conclusions = 83
• 8 Emotion Tracking of Radio Station Broadcasts = 85
• 8.1 Introduction = 85
• 8.2 System Construction = 86
• 8.3 Music Data = 87
• 8.3.1 Training Data = 87
• 8.3.2 Recorded Radio Broadcasts = 87
• 8.4 Feature Extraction = 88
• 8.5 Construction of Classifiers = 88
• 8.5.1 Music／Speech Classifier = 88
• 8.5.2 Classifier for Emotion Detection = 89
• 8.6 Results of Emotion Tracking of Radio Stations = 89
• 8.6.1 Comparison of Radio Stations = 90
• 8.6.2 Emotion Maps of Radio Station Broadcasts = 90
• 8.7 Conclusions = 93
• 9 Music Emotion Maps in the Arousal-Valence Space = 95
• 9.1 Introduction = 95
• 9.2 Related Work = 95
• 9.3 Music Data = 96
• 9.4 Feature Extraction = 97
• 9.5 Regressor Training = 97
• 9.6 Evaluation of Different Combinations of Feature Sets = 98
• 9.7 Selected Features Dedicated to the Detection of Arousal and Valence = 99
• 9.8 Emotion Maps = 101
• 9.8.1 Emotion Maps of Two Compositions = 101
• 9.8.2 Features Describing Emotion Maps = 103
• 9.8.3 Comparison of Musical Compositions = 103
• 9.9 Conclusions = 106
• 10 Comparative Analysis of Musical Performances by Using Emotion Tracking on the Arousal-Valence Plane = 107
• 10.1 Introduction = 107
• 10.2 Related Work = 107
• 10.3 System Construction = 108
• 10.4 Music Data for Regressor Training = 108
• 10.5 Regressor Training = 110
• 10.6 Aligning of Audio Recordings = 110
• 10.7 Analyzed Performances = 112
• 10.8 Results = 113
• 10.8.1 Influence of Recording Alignment on Emotion Values of the Compared Segments = 113
• 10.8.2 Performances and Arousal over Time = 115
• 10.8.3 Performances and Valence over Time = 116
• 10.8.4 Performances and Arousal-Valence over Time = 117
• 10.8.5 Arousal-Valence Trajectory = 119
• 10.8.6 Scape Plot = 119
• 10.8.7 Arousalscape = 121
• 10.8.8 Valencescape = 122
• 10.8.9 AVscape = 123
• 10.9 Evaluation = 124
• 10.9.1 Parameters Describing the Most Similar Performances = 124
• 10.9.2 Ground Truth for Performing Similarity Assessments Between Performances = 125
• 10.9.3 Evaluation Parameters = 126
• 10.10 Conclusions = 129
• References = 131
• Index = 137