Speech in Mobile and Pervasive Environments -- Contents -- About the Series Editors -- List of Contributors -- Foreword -- Preface -- Acknowledgments -- 1 Introduction -- 1.1 Application design -- 1.2 Interaction modality -- 1.3 Speech processing -- 1.4 Evaluations -- 2 Mobile Speech Hardware: The Case for Custom Silicon -- 2.1 Introduction -- 2.2 Mobile hardware: Capabilities and limitations -- 2.2.1 Looking inside a mobile device: Smartphone example -- 2.2.2 Processing limitations -- 2.2.3 Memory limitations -- 2.2.4 Power limitations -- 2.2.5 Silicon technology and mobile hardware -- 2.3 Profiling existing software systems -- 2.3.1 Speech recognition overview -- 2.3.2 Profiling techniques summary -- 2.3.3 Processing time breakdown -- 2.3.4 Memory usage -- 2.3.5 Power and energy breakdown -- 2.3.6 Summary -- 2.4 Recognizers for mobile hardware: Conventional approaches -- 2.4.1 Reduced-resource embedded recognizers -- 2.4.2 Network recognizers -- 2.4.3 Distributed recognizers -- 2.4.4 An alternative approach: Custom hardware -- 2.5 Custom hardware for mobile speech recognition -- 2.5.1 Motivation -- 2.5.2 Hardware implementation: Feature extraction -- 2.5.3 Hardware implementation: Feature scoring -- 2.5.4 Hardware implementation: Search -- 2.5.5 Hardware implementation: Performance and power evaluation -- 2.5.6 Hardware implementation: Summary -- 2.6 Conclusion -- Bibliography -- 3 Embedded Automatic Speech Recognition and Text-to-Speech Synthesis -- 3.1 Automatic speech recognition -- 3.2 Mathematical formulation -- 3.3 Acoustic parameterization -- 3.3.1 Landmark-based approach -- 3.4 Acoustic modeling -- 3.4.1 Unit selection -- 3.4.2 Hidden Markov models -- 3.5 Language modeling -- 3.6 Modifications for embedded speech recognition -- 3.6.1 Feature computation -- 3.6.2 Likelihood computation -- 3.7 Applications -- 3.7.1 Car navigation systems.

3.7.2 Smart homes -- 3.7.3 Interactive toys -- 3.7.4 Smartphones -- 3.8 Text-to-speech synthesis -- 3.9 Text to speech in a nutshell -- 3.10 Front end -- 3.11 Back end -- 3.11.1 Rule-based synthesis -- 3.11.2 Data-driven synthesis -- 3.11.3 Statistical parameteric speech synthesis -- 3.12 Embedded text-to-speech -- 3.13 Evaluation -- 3.14 Summary -- Bibliography -- 4 Distributed Speech Recognition -- 4.1 Elements of distributed speech processing -- 4.2 Front-end processing -- 4.2.1 Device requirements -- 4.2.2 Transmission issues in DSR -- 4.2.3 Back-end processing -- 4.3 ETSI standards -- 4.3.1 Basic front-end standard ES 201 108 -- 4.3.2 Noise-robust front-end standard ES 202 050 -- 4.3.3 Tonal-language recognition standard ES 202 211 -- 4.4 Transfer protocol -- 4.4.1 Signaling -- 4.4.2 RTP payload format -- 4.5 Energy-aware distributed speech recognition -- 4.6 ESR, NSR, DSR -- Bibliography -- 5 Context in Conversation -- 5.1 Context modeling and aggregation -- 5.1.1 An example of composer specification -- 5.2 Context-based speech applications: Conspeakuous -- 5.2.1 Conspeakuous architecture -- 5.2.2 B-Conspeakuous -- 5.2.3 Learning as a source of context -- 5.2.4 Implementation -- 5.2.5 A tourist portal application -- 5.3 Context-based speech applications: Responsive information architect -- 5.4 Conclusion -- Bibliography -- 6 Software: Infrastructure, Standards, Technologies -- 6.1 Introduction -- 6.2 Mobile operating systems -- 6.3 Voice over internet protocol -- 6.3.1 Implications for mobile speech -- 6.3.2 Sample speech applications -- 6.3.3 Access channels -- 6.4 Standards -- 6.5 Standards: VXML -- 6.6 Standards: VoiceFleXML -- 6.6.1 Brief overview of speech-based systems -- 6.6.2 System architecture -- 6.6.3 System architecture: VoiceFleXML interpreter -- 6.6.4 VoiceFleXML: Voice browser -- 6.6.5 A prototype implementation -- 6.7 SAMVAAD.

6.7.1 Background and problem setting -- 6.7.2 Reorganization algorithms -- 6.7.3 Minimizing the number of dialogs -- 6.7.4 Hybrid call-flows -- 6.7.5 Minimally altered call-flows -- 6.7.6 Device-independent call-flow characterization -- 6.7.7 SAMVAAD: Architecture, implementation and experiments -- 6.7.8 Splitting dialog call-flows -- 6.8 Conclusion -- 6.9 Summary and future work -- Bibliography -- 7 Architecture of Mobile Speech-Based and Multimodal Dialog Systems -- 7.1 Introduction -- 7.2 Multimodal architectures -- 7.3 Multimodal frameworks -- 7.4 Multimodal mobile applications -- 7.4.1 Mobile companion -- 7.4.2 MUMS -- 7.4.3 TravelMan -- 7.4.4 Stopman -- 7.5 Architectural models -- 7.5.1 Client-server systems -- 7.5.2 Dialog description systems -- 7.5.3 Generic model for distributed mobile multimodal speech systems -- 7.6 Distribution in the Stopman system -- 7.7 Conclusions -- Bibliography -- 8 Evaluation of Mobile and Pervasive Speech Applications -- 8.1 Introduction -- 8.1.1 Spoken interaction -- 8.1.2 Mobile-use context -- 8.1.3 Speech and mobility -- 8.2 Evaluation of mobile speech-based systems -- 8.2.1 User interface evaluation methodology -- 8.2.2 Technical evaluation of speech-based systems -- 8.2.3 Usability evaluations -- 8.2.4 Subjective metrics and objective metrics -- 8.2.5 Laboratory and field studies -- 8.2.6 Simulating mobility in the laboratory -- 8.2.7 Studying social context -- 8.2.8 Long- and short-term studies -- 8.2.9 Validity -- 8.3 Case studies -- 8.3.1 STOPMAN evaluation -- 8.3.2 TravelMan evaluation -- 8.3.3 Discussion -- 8.4 Theoretical measures for dialog call-flows -- 8.4.1 Introduction -- 8.4.2 Dialog call-flow characterization -- 8.4.3 (m,q,a)-characterization -- 8.4.4 (m,q,a)-complexity -- 8.4.5 Call-flow analysis using (m,q,a)-complexity -- 8.5 Conclusions -- Bibliography -- 9 Developing Regions.

9.1 Introduction -- 9.2 Applications and studies -- 9.2.1 VoiKiosk -- 9.2.2 HealthLine -- 9.2.3 The spoken web -- 9.2.4 TapBack -- 9.3 Systems -- 9.4 Challenges -- Bibliography -- Index.