Spoken Dialogue Systems Technology and Design

2. Advantages of Systems ProvidingMulti-Modal Interaction......Page 2
 3. Quality of Experience......Page 4
   4.2.5 Action Discussion Layer.......Page 9
  4.2 DIHANA Corpus......Page 13
Cover......Page 1
Spoken Dialogue SystemsTechnology and Design......Page 3
 Preface......Page 5
  4.2 IP-Television......Page 7
   4.2.2 Response Layer.......Page 8
   4.3.2 Inter-Annotator Agreement.......Page 11
 5. Semantic Content Tags......Page 15
 References......Page 20
Chapter 1 MULTILINGUAL SPEECH INTERFACES FOR RESOURCE-CONSTRAINED DIALOGUE SYSTEMS......Page 23
 1. Introduction......Page 24
  2.1 Review of Multilingual Speech Recognition......Page 25
 3. The NGT Annotation Model......Page 6
 Contents......Page 10
 Contributing Authors......Page 14
  5.1 Tag Specifications......Page 16
  5.3 Adaptation of the Language Model......Page 18
  6.3 Speech Synthesis......Page 21
  4.1 SwitchBoard Corpus......Page 12
  5.2 Annotation of Semantic Content Tags......Page 17
 6. Conclusions and Future Work......Page 19
 References......Page 22
  2.2 Review of Non-Native Speech Recognition......Page 26
 1. Introduction......Page 155
  4.1 Training and Test Data......Page 27
  4.2 Benchmark System......Page 29
  5.2 Multilingual MWC System......Page 30
  5.3 Model Merging......Page 33
  5.4 Adaptation with Non-Native Speech......Page 35
  6.1 Projections between GMMs......Page 37
  6.2 Scalable Architecture......Page 42
  6.3 Footprint......Page 45
 7. Summary......Page 46
 Notes......Page 47
Chapter 2 ONLINE LEARNING OF BAYES RISK-BASED OPTIMIZATION OF DIALOGUE MANAGEMENT FOR DOCUMENT RETRIEVAL SYSTEMS WITH SPEECH INTERFACE......Page 51
 1. Introduction......Page 52
  2.1 System Overview......Page 54
  2.2 Knowledge Base (KB)......Page 55
  2.3 Backend Retrieval System......Page 56
  2.4 Backend Question-Answering System......Page 57
  2.6 Field Test of Trial System......Page 58
  3.1 Choices in Generating Responses......Page 59
  3.3 Generation of Response Candidates......Page 60
  3.4 Definition of Bayes Risk for Candidate Response......Page 61
  4.1 Parameter Optimization by Online Learning......Page 64
  4.2 Optimization using Maximum LikelihoodEstimation......Page 65
  4.4 Online Learning Method using Reinforcement Learning......Page 66
 5. Evaluation of Online Learning Methods......Page 68
 6. Conclusions......Page 70
 References......Page 71
Chapter 3 TOWARDS FINE-GRAIN USER-SIMULATION FOR SPOKEN DIALOGUE SYSTEMS......Page 75
 1. Introduction......Page 76
 2. Related Work......Page 78
  2.1 Rule-based User Simulators......Page 79
  2.2 Corpus-based User Simulators......Page 81
  2.3 Hybrid User Simulators......Page 85
   2.4.1 Direct Methods.......Page 86
   2.4.2 Indirect Methods.......Page 87
  3.1 The Initial User Simulator......Page 88
  3.2 The Enhanced User Simulator......Page 90
 4. Experiments......Page 91
  4.1 Speech Database and Scenario Corpus......Page 92
  4.2 Language Models for Speech Recognition......Page 93
  5.1 Detection of Problems in the Performanceof the Dialogue System......Page 94
   5.1.2 Findings for the Medium Cooperativeness Level.......Page 95
   5.1.3 Findings for the Low Cooperativeness Level.......Page 97
  5.2 Future Work......Page 98
 References......Page 99
Chapter 4 SALIENT FEATURES FOR ANGER RECOGNITION IN GERMAN AND ENGLISH IVR PORTALS......Page 104
 1. Introduction......Page 105
 3. Overview of Database Conditions......Page 106
 4. Selected Corpora......Page 108
 5. Prosodic and Acoustic Modeling......Page 109
   5.1.4 Spectrals.......Page 110
   5.1.5 Formants.......Page 111
 6. Feature Ranking......Page 113
 7. Normalization......Page 115
  8.1 Cross Validation......Page 116
  8.2 Evaluation Measurement......Page 117
  9.1 Analyzing Feature Distributions......Page 118
  9.2 Optimal Feature Sets......Page 120
  9.3 Optimal Classification......Page 121
  10.0.1 Signal Quality.......Page 122
  10.0.3 Speech Transcription.......Page 123
 Acknowledgments......Page 124
Chapter 5 PSYCHOMIME CLASSIFICATION AND VISUALIZATION USING A SELF-ORGANIZING MAP FOR IMPLEMENTING EMOTIONAL SPOKEN DIALOGUE SYSTEM......Page 127
 1. Introduction......Page 128
  2.1 Onomatopoeias and Psychomimes......Page 129
  2.2 Emotional Spoken Dialogue Systems......Page 130
  3.1 What is SOM?......Page 131
  3.2 Natural Language Processing Studies usingSOM......Page 133
  4.1 Psychomimes and their Groupings......Page 135
  4.2 Corpus......Page 136
  4.4 SOM Parameters......Page 138
   4.5.1 Determination of Group Areas.......Page 139
   4.5.2 Recall and Precision.......Page 141
   4.5.3 Effects of Selecting Frames and Combinationsof Frames.......Page 145
   4.5.4 Effects of Narrowing Area of Groups.......Page 148
   4.5.5 How to Take Advantage of Knowledge.......Page 151
 References......Page 152
  2.1 The Nature of Dialogue Research......Page 156
  2.2 Academic and Commercial Research......Page 157
  2.3 Three Decades of Research in SpokenDialogue Systems......Page 158
  2.4 Application Areas for Dialogue Research......Page 163
  3.1 Conducting Research in Spoken Dialogue Systems......Page 165
  3.2 The Availability of Resources for the Designand Development of Spoken Dialogue Systems......Page 166
  4.1 Incorporating Dialogue into Voice Search......Page 168
  4.2 Using Dialogue Systems in AmbientIntelligence Environments......Page 170
  4.3 CHAT......Page 171
  4.4 SmartKom and SmartWeb......Page 172
  4.5 TALK......Page 173
  4.6 COMPANIONS......Page 174
  4.7 Atraco......Page 175
 5. Concluding Remarks......Page 176
 Web Pages......Page 177
 Notes......Page 178
Chapter 7 DIALOGUE CONTROL BY POMDPUSING DIALOGUE DATA STATISTICS......Page 182
 1. Introduction......Page 183
  2.1 POMDP Structure......Page 185
  2.2 Running Cycle and Value Iteration......Page 187
  3.1 Purpose of Dialogue Control......Page 188
  3.2 Automatically Acquiring POMDPParameters and Obtaining a Policy forTarget Dialogues......Page 189
  3.3 Reflecting Action Predictive Probabilities in Action Control......Page 192
 4. Evaluation and Results......Page 196
 5. Discussion......Page 198
 6. Future Work......Page 200
 7. Conclusions......Page 202
 Acknowledgments......Page 203
 1. Introduction......Page 206
 2. Overview of the Data-Management CenteredPrototyping Method......Page 208
  3.1 Data Model Definition......Page 210
  3.2 Controller Script......Page 211
  3.4 Adding a Multi-Modal Interface to a GUIWeb Application......Page 212
  3.5 Generating Speech Interaction......Page 213
  3.7 Generation of Dialogue Flow......Page 215
  3.8 The Result of the Prototyping......Page 216
 4. Prototyping of a DB-Search Dialogue System......Page 217
 5. Prototyping of a Multi-Modal InteractivePresentation System......Page 219
  5.1 Dialogue Pattern Generation from Metadata......Page 220
  5.2 Generation of QA Database......Page 222
  5.3 Adaptation of the Language Model......Page 223
  5.4 Implementation and Evaluation......Page 224
  6.1 User Model in Multi-Modal InteractionSystems......Page 226
  6.2 User Model Component of MIML......Page 227
 7. Conclusions......Page 228
 Notes......Page 229
 1. Introduction......Page 231
 2. Advantages of Systems ProvidingMulti-Modal Interaction......Page 232
  2.1 Modality Relations......Page 233
 3. Quality of Experience......Page 234
 4. Audio-Video Quality Integration inAV-Transmission Services......Page 235
  4.1 Videotelephony......Page 236
  4.2 IP-Television......Page 237
 5. Quality of Embodied Conversational Agents......Page 239
 6. Quality of Systems with Multiple InputModalities......Page 241
  6.1 Smart Office......Page 242
  6.2 Mobile......Page 243
  6.3 Summary......Page 244
 References......Page 246
 1. Introduction......Page 249
 2. Kyoto Tour Guide Dialogue Corpus......Page 251
  4.1 Annotation Unit......Page 254
   4.2.1 General Layer.......Page 256
   4.2.4 Constrain Layer.......Page 257
   4.2.6 Others Layer.......Page 258
   4.3.2 Inter-Annotator Agreement.......Page 259
   4.3.3 Analysis of the Occurrence Tendency during the Progress of the Episode.......Page 260
  4.4 Preliminary Experiment to Estimate SA Tags via SVM......Page 262
 5. Semantic Content Tags......Page 263
  5.1 Tag Specifications......Page 264
  5.2 Annotation of Semantic Content Tags......Page 265
  6.2 Dialogue Management......Page 267
  6.3 Speech Synthesis......Page 269
 7. Conclusions......Page 270
 1. Introduction......Page 273
 2. The HMM-based Annotation Model......Page 275
 3. The NGT Annotation Model......Page 278
 4. Corpora......Page 282
  4.1 SwitchBoard Corpus......Page 284
  4.2 DIHANA Corpus......Page 285
 5. Experimental Results......Page 286
 6. Conclusions and Future Work......Page 291
 Notes......Page 292
Index......Page 295