Inhalt

•  
   Kurzfassung
•  
   Table of Contents
•  
   List of Figures
•  
   List of Tables
•  
   Introduction
•  
   Thesis Outline
•  
   Contributions
•  
   Previously Published and Co-authored Material
•  
  Spoken Dialogue and Spoken Dialogue Systems
•  
   Modelling Dialogue
•  
   The Shannon-Weaver Model of Communication
•  
   Layers of Communication
•  
   Emergence of Behaviour in Complex Systems
•  
   Establishing Common Ground
•  
   Taking Turns
•  
   Feedback and the Backward Channel
•  
  Components and Architecture for Spoken Dialogue Systems
•  
   Components of Spoken Dialogue Systems
•  
   Interconnection of Components
•  
   Discussion
•  
  State of the Art in Spoken Dialogue Systems
•  
   Commercial, Standards-based Systems
•  
   Related Research Systems
•  
   Advanced Commercial Systems
•  
   Summary and Discussion
•  
  Incremental Processing and its Evaluation
•  
   Timeliness and Incrementality
•  
   Aspects of Incrementality
•  
   Related Work on Evaluating Incremental Processing
•  
   Relation to Anytime Processing
•  
  Our Notion of Incremental Processing
•  
   Incremental Processors
•  
   Representing Incremental Data
•  
  Evaluation of Incremental Processors
•  
   Gold Standards for Evaluation
•  
   Evaluation with Incremental Gold Standards
•  
   Evaluation with Non-Incremental Gold Standards
•  
  Metrics for Evaluation of Incremental Processors
•  
   Similarity Metrics
•  
   Timing Metrics
•  
   Diachronic Metrics
•  
   Interrelations between Metrics
•  
   Summary
•  
   A Software Architecture for Incremental Spoken Dialogue Processing
•  
   The Data Model: Incremental Units
•  
   The IU Network
•  
   Triangular Data Models
•  
  The Processing Model
•  
   Incremental Modules and Inter-Module Communication
•  
   Alternative Processing Schemes
•  
   Infrastructure
•  
   Discussion
•  
   Incremental Speech Recognition
•  
   Automatic Speech Recognition in a Nutshell
•  
   Modelling Speech Data
•  
   The Language Model
•  
   The Pronunciation Model
•  
   Acoustic Modelling in the ASR Frontend
•  
   The Speech Recognizer
•  
   Hidden Markov Models
•  
   The Decoding Algorithm
•  
   Evaluating Speech Recognition
•  
   Refinements
•  
   The Sphinx-4 Speech Recognizer
•  
   Incrementalizing Speech Recognition
•  
   The InproTK Module for Incremental Rich Speech Recognition
•  
  INTELIDA: A Workbench for Evaluating iSR
•  
   The Library
•  
   Interactive Tools
•  
   Evaluation of Basic Incremental Speech Recognition
•  
   Variations of the Setup and Stability of Results
•  
   N-Best Processing
•  
   Optimization of Incremental Speech Recognition
•  
   Right Context
•  
   Hypothesis Smoothing
•  
   Advanced Smoothing Methods
•  
  Example Application: Incremental Command-and-Control
•  
   The Greifarm Domain
•  
   Cost-based Smoothing
•  
   Wizard-of-Oz Experiment
•  
   System Implementation
•  
   Evaluation
•  
   Summary and Discussion
•  
   Short-Term Estimation of Dialogue Flow
•  
   Floor Tracking in INPROTK
•  
   Architecture and Implementation
•  
   Example Application: Collaborating on Utterances
•  
   Domain and Setup
•  
   Experiment and Results
•  
   Discussion
•  
   Micro-Timing Prediction
•  
   Motivation for the Task
•  
   Related Work on Simultaneous Speech
•  
   System Architecture
•  
   Two Models for Micro-Timing
•  
  Evaluation
•  
   Corpus and Experiment Setup
•  
   End-of-Word Prediction: When to Start Speaking
•  
   Predicting the Micro-Timing of the Upcoming Word
•  
   Estimating the Reliability of the Predictions
•  
   Summary
•  
   Example Application: Speaking in Synchrony With the User
•  
   Discussion
•  
   Summary and Discussion
•  
   Incremental Speech Synthesis
•  
   Rationale for Incremental Speech Synthesis
•  
   Related Work
•  
   Requirements
•  
   Speech Synthesis in a Nutshell
•  
   Text-based Linguistic Processing
•  
   HMM-based Waveform Synthesis
•  
   Parameter Estimation with HMMs
•  
   Vocoding
•  
   Discussion of Alternative Synthesis Techniques
•  
   Evaluation of Speech Synthesis
•  
   MaryTTS
•  
   Incrementalizing Speech Synthesis
•  
  Incremental Speech Synthesis in INPROTK
•  
   Utterance Tree-based iSS
•  
   An Incremental Module for Speech Synthesis
•  
   Very Low-Latency Prosody Adaptation
•  
   Automatic Hesitation
•  
   Conformance to the Requirements
•  
   The Merit of iSS
•  
   Domain and System
•  
   Evaluation
•  
   Results
•  
   Discussion
•  
   Example Application: Integration with Incremental NLG
•  
   Use-Case: Adaptive Information Presentation
•  
   Implemented System
•  
  Evaluation
•  
   System Response Time
•  
   Subjective Evaluation
•  
   Discussion
•  
   Evaluating the Prosodic Quality of iSS
•  
   The Design Space for Incremental Prosody Production
•  
   Experiment
•  
   Evaluation
•  
   Qualitative Analysis
•  
   Quantitative Evaluation
•  
   Conclusion
•  
   Summary and Discussion
•  
  Conclusion and Outlook
•  
   Summary
•  
   Conclusion
•  
   Open Questions
•  
   Bibliography