Home

Keynote Speech 1: Is Syntactic Structure Prosodically Recoverable? KN-1

Speaker: Mario Rossi, Institut de Phonetique d'Aix-en-Provence, Laboratoire Parole et Langage, France

Chair: Louis Pols, ESCA, Univ. of Amsterdam, The Netherlands

Keynote Speech 2: Conversational Interfaces: Advances and Challenges KN-9

Speaker: Victor Zue, MIT, USA

Chair: Paul Daalsgard, Aalborg University, Denmark

Keynote Speech 3: Prosodic Modelling in Text-to-Speech Synthesis KN-19

Speaker: Jan P. H. van Santen, Bell Labs-Lucent Technologies, USA

Chair: Hiroya Fujisaki, Science Univ. of Tokyo, Japan

Keynote Speech 4: Robust Speech Recognition: Review and Perspectives

Chair: Joseph Mariani, LIMSI-CNRS, France

Speech 4A (8:35-9:00): Impact of the Unknown Communication Channel on Automatic Speech Recognition: A Review KN-29

Speaker: Jean-Claude Junqua, Panasonic Technologies Inc., USA

Speech 4B (9:00-9:25): Statistical Techniques for Robust ASR: Review and Perspectives KN-33

Speaker: Jerome Bellegarda, Apple Computer, USA

Speech 4C (9:25-9:50): Using Missing Feature Theory to Actively Select Features for Robust Speech Recognition with Interruptions, Filtering and Noise KN-37

Speaker: Richard Lippmann, Lincoln Laboratory MIT, USA

Keynote Speech 5: Perspectives of Speech Technology Research Highlighted in Eurospeech '97

Speaker: Wolfgang Hess, Univ. of Bonn, Germany

Chair: George Kokkinakis, WCL, Univ. of Patras, Greece

SESSION: M4A

Acoustic Modelling I

Chair: Roger Moore, DRA, UK

M4A.1 Using Multiple Time Scales in a Multi-Stream Speech Recognition System 3

Stephane Dupont, *Herv? Bourlard

FPMs-TCTS, Belgium

*IDIAP, Switzerland

M4A.2 Speech Recognition Using HMM-State Confusion Characteristics 7

Wakita Yumi, Harald Singer, Yoshinori Sagisaka

ATR Interpreting Telecommunications Res. Labs., Japan

M4A.3 Bottom-up and Top-down State Clustering for Robust Acoustic Modeling 11

Cristina Chesta, Pietro Laface, *Franco Ravera

Politecnico di Torino, Italy

*Centro Studi e Laboratori Telecomunicazioni (CSELT), Italy

M4A.4 Comparison of Optimization Methods for Discriminative Trainining Criteria 15

Ralf Schluter, W. Macherey, S. Kanthak, Hermann Ney, Lutz Welling

RWTH Aachen, Germany

M4A.5 Clustering Beyond Phoneme Contexts for Speech Recognition 19

Clark Z. Lee, Douglas O'Shaughnessy

INRS Telecommunications, Canada

M4A.6 Influence of Outliers in Training the Parametric Trajectory Models for Speech Recognition 23

Rathinavelu Chengalvarayan

Bell Labs-Lucent Technologies, USA

SESSION: M4B

Dynamic Articulatory Measurements

Chair: Renι Carre, ENST, France

M4B.1 Adaptation of Natural Articulatory Movements

to the Control of the Command Parameters of a Production Model 27

Laurence Candille, Henri Meloni

CERI, France

M4B.2 Three-Dimensional Coarticulatory Strategies of Tongue Movement 31

Maureen Stone, *Andrew Lundberg, *Edward Davis, Rao Gullapalli, Moriel NessAiver

Univ. of Maryland, USA

*John Hopkins Univ., USA

M4B.3 From Laryngographic and Acoustic Signals to Voicing Gestures 35

Nathalie Parlangeau, Regine Andrι-Obrecht

IRIT - Equipe IHMPT, France

M4B.4 Ultrasonographic Measurement of Cricothyroid Space in Speech 39

Erkki Vilkman, Raija Takalo, Maatta Taisto, *Anne-Maria Laukkanen, Jaana Nummenranta, Lipponen Tero

Univ. of Oulu, Finland

*Univ. of Tampere, Finland

M4B.5 Coarticulation and Articulatory Compensations Studied by Dynamic MRI 43

Didier Demolin, M. George, V. Lecuit, *T. Metens, /‡A. Soquet, †H. Raeymaekers

Univ. Libre de Bruxelles, Belgium

*Magnetic Resonance Unit,Hopital Erasme

†Philips Medical Systems, Belgium

‡Collaborateur Scientifique-FNRS

M4B.6 Determining Tongue Articulation: From Discrete Fleshpoints to Continuous Shadow 47

Pierre Badin, Enrico Baricchi, Anne Vilain

Institut de la Communication Parleι, France

SESSION: M4C

Language Identification

Chair: Marc Zissman, MIT Lincoln Laboratory, USA

M4C.1 Predicting, Diagnosing and Improving Automatic Language Identification Performance 51

Marc A. Zissman

MIT Lincoln Laboratory, USA

M4C.2 Language Identification with Language-Independent Acoustic Models 55

Cristobal Corredor-Ardoy, Jean Luc Gauvain, Martine Adda-Decker, Lori Lamel

LIMSI, France

M4C.3 Bayesian Methods for Language Verification 59

Eluned S. Parris, Harvey Lloyd-Thomas, Michael Carey, *Jerry H. Wright

Ensigma Ltd, UK

*Bristol Univ., UK.

M4C.4 Use of Recurrent Network for Unknown Language Rejection in Language Identification System 63

Hingkeung Kwan, Keikichi Hirose

Univ. of Tokyo, Japan

M4C.5 Language-Identification Based on Cross-Language Acoustic Models and Optimized Information Combination

Ove Andersen, Paul Dalsgaard 67

Aalborg Univ., Denmark

M4C.6 Phonetic-Context Mapping in Language Identification 71

Jiri Navratil, Werner Zuehlke

Technische Univ. Ilmenau, Germany

SESSION: M4D

Neural Networks for Speech and Language Processing

Chair: Wolfgang Hess, Univ. of Bonn, Germany

M4D.1 Discriminative Feature and Model Design for Automatic Speech Recognition 75

Mazin Rahim, Yoshua Bengio, Yann LeCun

AT&T Labs-Research, USA

M4D.2 Large Vocabulary Speech Recognition with Context Dependent MMI-Connectionist/HMM Systems Using the WSJ Database 79

Jφrg Rottland, Christoph Neukirchen, Daniel Willett, Gerhard Rigoll

Gerhard-Mercator-Univ. Duisburg, Germany

M4D.3 Automatic Selection of Segmental Acoustic Parameters by Means of Neural-Fuzzy Networks for Reordering in N-best HMM Hypotheses 83

Thierry Moudenc, Guy Mercier

France Telecom, France

M4D.4 Comparison Results for Segmental Training Algorithms for Mixture Density HMMS 87

Mikko Kurimo

Helsinki Univ. of Technology, Finland

M4D.5 A Connectionist Approach to Machine Translation

Asuncion Castano, *Francisco Casacuberta 91

Universitδt Jaume I, Spain

*Univ. Politecnica de Valencia, Spain

M4D.6 Continuous Speech Recognition Using a Context Sensitive ANN and HMM2s 95

Nicolas Pican, Jean-Francois Mari, Dominique Fohr

CRIN-CNRS & INRIA Lorraine, France

SESSION: MAA

Training Techniques, Efficient Decoding in ASR

Chair: Jerome Bellegarda, Apple Computer, USA

MAA.1 Acoustic Modeling Based on the MDL Principle for Speech Recognition 99

Koichi Shinoda, Takao Watanabe

NEC Corporation, Japan

MAA.2 Discriminative Utterance Verification Using Multiple Confidence Measures 103

Piyush Modi, Mazin Rahim

AT&T, USA

MAA.3 Subspace Distribution Clustering for Continuous Observation Density Hidden Markov Models 107

Enrico Bocchieri, Brian Mak

AT&T Labs-Research, USA

MAA.4 A Comparative Study of Methods for Phonetic Decision-Tree State Clustering 111

H.J. Nock, M.J.F. Gales, Steve Young

Cambridge Univ., UK

MAA.5 Comparing Gaussian and Polynomial Classifica-tion in SCHMM-Based Recognition Systems 115

Alfred Kaltenmeier, Jurgen Franke

Daimler Benz AG, Germany

MAA.6 Maximum Likelihood Successive State Splitting Algorithm for Tied-Mixture HMNET 119

AlexAndrι Girardi, Harald Singer, Kiyohiro Shikano, Satoshi Nakamura

Nara Institute of Science and Technology, Japan

MAA.7 String-Level MCE for Continuous Phoneme Recognition 123

Erik McDermott, Shigeru Katagiri

ATR Interpreting Telecommunications Res. Labs., Japan

MAA.8 HMM State Clustering Across Allophone Class Boundaries 127

Ze'ev Rivlin, Ananth Sankar, Harry Bratt

SRI International, USA

MAA.9 Weighted Determinization and Minimization for Large Vocabulary Speech Recognition 131

Mehryar Mohri, Michael Riley

AT&T Labs-Research, USA

MAA.10 Parallel Speech Recognition 135

Steven Phillips, Anne Rogers

AT&T Labs-Research, USA

MAA.11 Fast Likelihood Computation Methods for Continuous Mixture Densities in Large Vocabulary Speech Recognition 139

Stefan Ortmanns, Thorsten Firzlaff, Hermann Ney

RWTH Aachen Univ. of Technology, Germany

MAA.12 A Static Lexicon Network Representation for Cross-Word Context Dependent Phones 143

Kris Demuynck, Jacques Duchateau, Dirk Van Compernolle KU Leuven-ESAT, Belgium

MAA.13 Decision-Tree Based Quantization of the Feature Space of a Speech Recognizer 147

Mukund Padmanabhan, L.R. Bahl, D. Nahamoo, De Souza Peter

IBM, USA

MAA.14 Sub-Vector Clustering to Improve the Memory

and Speed Performance of the Acoustic Likelihood Computation 151

Mosur Ravishankar, *R. Bisiani, E. Thayer

Carnegie Mellon Univ., USA

*Univ. of Milan, Italy

MAA.15 The Incorporation of Path Merging in a Dynamic Network Recogniser 155

Simon Hovell

BT Laboratories, UK

MAA.16 Improvement on Connected Digits Recognition Using Duration Constraints in the Asynchronous Decoding Scheme 159

Miroslav Novak

IBM, USA

MAA.17 Explicit Word Error Minimization in N-Best List Rescoring 163

Andreas Stolcke, Yochai Konig, Mitchel Weintraub

SRI International, USA

MAA.18 Efficient 2-Pass N-Best Decoder 167

Long Nguyen, Richard Schwartz

BBN Systems and Technologies, USA

MAA.19 A Memory Management Method for a Large Word Network 171

Tomohiro Iwasaki, Yoshiharu Abe

Mitsubishi Electric Corporation, Japan

SESSION: MAB

Prosody

Chair: Nick Campbell, ATR, Japan

MAB.1 Persistence of Prosodic Features Between Dialectal and Standard Italian Utterances in Six Sub-Varieties of a Region of Southern Italy (Salento): First Assessments of

the Results of a Recognition Test and an Instrumental Analysis 175

Antonio Romano

Univ. Stendhal, France

MAB.2 Improving the Phonetic Annotation by Means of Prosodic Phrasing 179

Halewijn Vereecken, *Annemie Vorstermans, Jean-Pierre Martens, *Bert Van Coile

Univ. of Gent, Belgium

*L&H, Belgium

MAB.3 A Descriptive Study of Prosodic Phenomena in MPUR (West Papuan Phylum) 183

Cecilia Ode

Leiden Univ., The Netherlands

MAB.4 Automated Quantitative Analysis of Fo Contours of Utterances From a German ToBI-Labeled Speech

Database 187

HansJφrg Mixdorff, *Hiroya Fujisaki

TU Dresden, Germany

*Science Univ. of Tokyo, Japan

MAB.5 Identification and Automatic Generation of Prosodic Contours for Text-to-Speech Synthesis System in

French 191

De Tournemire Stephanie

France Telecom, France

MAB.6 Quantitative Analysis and Formulation of Tone Concatenation in Chinese FO Contours 195

Jin-Fu Ni, Ren-Hua Wang, *Keikichi Hirose

Univ. of Science and Technology of China, ROChina

*Univ. of Tokyo, Japan

MAB.7 An Environment for the Labelling and Testing of Melodic Aspects of Speech 199

Christel Brindopke, Arno Pahde, Franz Kummert, Gerhard Sagerer

Univ. of Bielefeld, Germany

MAB.8 PROPAUSE: A Syntactico-Prosodic System Designed to Assign Pauses 203

David Casacuberta, Lourdes Aguilar, Rafael Marin

Universitδt Autonoma de Barcelona, Spain

MAB.9 Integrated Dialog Act Segmentation and Classification Using Prosodic Features and Language Models 207

Volker Warnke, *Ralf Kompe, Heinrich Niemann, Elmar Noeth

Univ. of Erlangen-Nόremburg, Germany

*Sony International (Europe), Germany

MAB.10 Evaluation of Prosodic Characteristics in Retold Stories in Dutch by Means of Semantic Scales 211

Monique E. van Donzel, Florien J. Koopmans-van Beinum

Univ. of Amsterdam, The Netherlands

MAB.11 Text-to-Intonation in Spontaneous Swedish 215

Gosta Bruce, Marcus Filipsson, Johan Frid, *Bjφrn Granstrφm, *Kjell Gustafson, Merle Horne, David House

Lund Univ., Sweden

*KTH, Sweden

MAB.12 Synthesing Attitudes with Global Rhythmic and Intonation Contours 219

Yann Morlec, Gerard Bailly, Veronique Auberge

Institut de la Communication Parleι, France

MAB.13 Prosody-Particle Pairs as Discourse Control

Signs 223

Dafydd Gibbon, Claudia Sassen

Univ. of Bielefeld, Germany

MAB.14 Focus Detection with Additional Information of Phrase Boundaries and Sentence Mode 227

Anja Elsner

Univ. of Bonn,Germany

MAB.15 The Role of Prosody in Infants' Native-Language Discrimination Abilities: The Case of Two Phonologically Close Languages 231

Laura Bosch, Nuria Sebastian-Galles

Universitδt de Barcelona, Spain

MAB.16 Prosodic Cycles and Interpersonal Synchrony in American English and Swedish 235

Eugene H. Buder, *Anders Eriksson

Univ. of Memphis, USA

*Umeε Univ., Sweden

MAB.17 Relating Prosody to Syntax: Boundary Signalling in Swedish 239

Eva Strangert

Umeε Univ., Sweden

MAB.18 On Representation of Fundamental Frequency of Speech for Prosody Analysis Using Reliability Function

Mitsuru Nakai, Hiroshi Shimodaira 243

Japan Advanced Institute of Science and Technology, Japan

MAB.19 Efficient Method of Establishing Words Tone Dictionary for Korean TTS System 247

Seong-hwan Kim, Jin-young Kim

Chonnam National Univ., South Korea

MAB.20 Perception of Questions and Statements in Neapolitan Italian 251

D’Imperio Mariapaola *David House

Ohio State University, USA

*Lund Univ., Sweden

SESSION: T1A

Keyword and Topic Spotting

Chair: Joseph Mariani, LIMSI-CNRS, France

T1A.1 Key-Phrase Spotting Using An Integrated Lan-guage Model of N-Grams and Finite-State Grammar 255

Lin Qiguang, Dave Lubensky, Michael Picheny, P. Srinivasa Rao

IBM, USA

T1A.2 Efficient Methods for Detecting Keywords in Continuous Speech 259

Jochen Junkawitsch, *Gόnther Ruske, Harald Hφge

Siemens AG, Germany

*Munich Univ.of Technology, Germany

T1A.3 Providing Sublexical Constraints for Word Spotting Within the ANGIE Framework 263

Raymond Lau, Stephanie Seneff

MIT Laboratory for Computer Science, USA

T1A.4 Usefulness of Phonetic Parameters in a Rejection Procedure of an HMM Based Speech Recognition System

Katarina Bartkova, Denis Jouvet 267

France Telecom, France

T1A.5 Keyword Spotting Using FO Contour Matching

Yoichi Yamashita, *Riichiro Mizoguchi 271

Ritsumeikan Univ., Japan

*Osaka Univ., Japan

T1A.6 A Frame and Segment Based Approach for Topic Spotting 275

Elmar Noeth, Stefan Harbeck, Heinrich Niemann, Volker Warnke

Univ. of Erlangen-Nόremburg, Germany

SESSION: T1B

Robustness in Recognition and Signal Processing I

Chair: Helmut Mangold, Daimler Benz, Germany

T1B.1 Cyclic Autocorrelation-Based Linear Prediction Analysis of Speech 279

K.K. Paliwal, Yoshinori Sagisaka

ATR Interpreting Telecommunications Res. Labs., Japan

T1B.2 Novel Filler Acoustic Models for Connected Digit Recognition 283

Ilija Zeljkovic, Shrikanth Narayanan

AT&T Labs-Research, USA

T1B.3 A Non-Iterative Model-Adaptive E-CMN/PMC Approach for Speech Recognition in Car Environments

Makoto Shozakai, Satoshi Nakamura, Kiyohiro Shikano

Nara Institute of Science and Technology, Japan 287

T1B.4 Discriminative Feature Extraction for Speech Recognition in Noise 291

la Torre Angel de, Antonio M. Peinado, Antonio J. Rubio, Pedro Garcia

Universidad de Granada, Spain

T1B.5 Noise Robust Recognition Using Feature Selective Modeling 295

Michael K. Brendborg, Borge Lindberg

Aalborg Univ., Denmark

T1B.6 Mixture Input Transformations for Adaptation of Hybrid Connectionist Speech Recognizers 299

Victor Abrash

SRI International, USA

SESSION: T1C

Modelling of Prosody

Chair: Hiroya Fujisaki, Science Univ. of Tokyo, Japan

T1C.1 Metrical Representation of Demarcation and Constituency in Long Noun Phrases 303

Christos Malliopoulos, *George Mikros

National Technical Univ. of Athens, Greece

*ILSP, Greece

T1C.2 A System of Stylized Intonation Contours for

German 307

Hannes Pirker, *Kai Alter, Erhard Rank, John Matiasek, Harald Trost, †Germnot Kubin

Austrian Research Institute for Artificial Intelligence (OFAI), Austria

*Max-Planck-Institute of Cognitive Neuroscience, Germany

†Vienna Univ. of Technology, Austria

T1C.3 A Method of Representing Fundamental Frequency Contours of Japanese Using Statistical Models of Moraic Transition 311

Keikichi Hirose, Kouji Iwano

Univ. of Tokyo, Japan

T1C.4 Modeling Arbitrarily Long Sentence-Spanning F0 Contours by Parametric Concatenation of Word-Spanning Patterns 315

Stavroula-Evita Fotinea, *Michael Vlahakis, *George Carayannis

National Technical Univ. of Athens, Greece

*ILSP, Greece

T1C.5 Strong Interaction Between Factors Influencing Consonant Duration 319

R. J. J. H. van Son, *Jan P.H. van Santen

Univ. of Amsterdam, The Netherlands

*Bell Labs-Lucent Technologies, USA

T1C.6 Speech Timing in Slovenian TTS 323

Jerneja Gros, Nikola Pavesic, France Mihelic

Univ. of Ljubljana, Slovenia

SESSION: T1D

Microphone Arrays for Speech Enhancement

Chair: Alan Bradley, RMIT, Australia

T1D.1 Small Microphone Arrays with Optimized Directivity for Speech Enhancement 327

Matthias Dorbecker

Aachen Univ. of Technology, Germany

T1D.2 Microphone Array Design Measures for Hands-Free Speech Recognition 331

Masaaki Inoue, Satoshi Nakamura, Takeshi Yamada, Kiyohiro Shikano

Nara Institute of Science and Technology, Japan

T1D.3 Noise Reduction by Paired Microphones 335

Masato Akagi, Mitsunori Mizumachi

Japan Advanced Institute of Science and Technology, Japan

T1D.4 A Microphone Array for Speech Enhancement Using Multiresolution Wavelet Transform 339

Djamila Mahmoudi

LTS-DE, EPFL, Switzerland

T1D.5 A Two-Channel Adaptive Microphone Array with Target Tracking 343

Yoshifumi Nagata, Hiroyuki Tsuboi

Toshiba Corporation, Japan

T1D.6 Use of Different Microphone Array Configurations for Hands-Free Speech Recognition in Noisy and Reverberant Environment 347

Diego Giuliani, Marco Matassoni, Maurizio Omologo, Piergiorgio Svaizer

Istituto per la Ricerca Scientifica e Tecnologica (IRST), Italy

SESSION: T2A

Multilingual Recognition

Chair: Richard Lippman, MIT Lincoln Lab., USA

T2A.1 YINHE: A Mandarin Chinese Version of the Galaxy System 351

Chao Wang, James R. Glass, Helen Meng, Joe Polifroni, Stephanie Seneff, Victor Zue

MIT Laboratory for Computer Science, USA

T2A.2 Multilingual Speech Recognition for Flexible Vocabularies 355

Patrizia Bonaventura, *Filippo Gallocchio, †Giorgio Micca

Centro Studi e Laboratori Telecomunicazioni (CSELT), Consultant, Italy

†Centro Studi e Laboratori Telecomunicazioni (CSELT), Turin, Italy

*Univ.di Padova, Italy

T2A.3 A Study of Multilingual Speech Recognition 359

Fuliang Weng, Harry Bratt, Leonardo Neumeyer, Andreas Stolcke

SRI International, USA

T2A.4 Multilingual Speech Recognition: The 1996 Byblos Callhome System 363

Jayadev Billa, Kristine Ma, John W. McDonough, George Zavaliagkos, David R. Miller, Kenneth N. Ross, Amro El-Jaroudi

BBN Systems and Technologies, USA

T2A.5 Japanese LVCSR on the Spontaneous Scheduling Task with JANUS-3 367

Tanja Schultz, Detlef Koll, Alex Waibel

Univ. Karlsruhe, Germany

T2A.6 Fast Bootstrapping of LVCSR Systems with Multilingual Phoneme Sets 371

Tanja Schultz, Alex Waibel

Univ. Karlsruhe, Germany

SESSION: T2B

Language Specific Speech Analysis

Chair: John J. Ohala, Univ. of California, USA

T2B.1 Factors of Variation in the Production of the German Dorsal Fricative 375

Bernd Pompino-Marschall, Christine Mooshammer

Center for General Linguistics, Berlin, Germany

T2B.2 EPG and Aerodynamic Evidence for the Copro-duction and Coarticulation of Clicks in ISIZULU 379

Kimberly Thomas

Univ. of California, USA

T2B.3 Formant Trajectory Dynamics in Swabian

Diphthongs 383

Anja Geumann

Muenchen Univ., Germany

T2B.4 The Gestural Organization of Vowels and Consonants: A Cinefluorographic Study of Articulator Gestures in Greenlandic 387

Sidney A.J. Wood

Univ. of Lund, Sweden

T2B.5 The Perception of Coronals in Western Arrernte

Victoria B. Anderson 389

Univ. of California, USA

T2B.6 Acoustic Modelling of American English /r/ 393

Carol Y. Espy-Wilson, *Shrikanth Narayanan, Suzanne E. Boyce, †Abeer Alwan

Boston Univ., USA

*AT&T Labs-Research, USA

†Univ. of California, USA

SESSION: T2C

Feature Estimation I

Chair: Paul Daalsgard, Aalborg University, Denmark

T2C.1 Acoustic Parameters Optimised for Recognition of Phonetic Features 397

Anya Varnich Hansen

Aalborg Univ., Denmark

T2C.2 Heterogeneous Acoustic Measurements for Phonetic Classification 401

Andrew K. Halberstadt, James R. Glass

MIT, USA

T2C.3 Cepstral-Time Matrices and LDA for Improved Connected Digit and Sub-Word Recognition Accuracy

Ben Milner 405

BT Laboratories, UK

T2C.4 Data-Driven Design of Rasta-Like Filters 409

Van Vuuren Sarel , / *Hynek Hermansky

Oregon Graduate Institute of Science and Technology, USA

*International Computer Science Institute, USA

T2C.5 Evaluating Feature Set Performance Using the F-Ratio and J-Measures 413

Simon Nicholson, *Ben Milner, Stephen Cox

UEA, UK

*BT Laboratories, UK

T2C.6 Robust Speech Parameters Located in the Frequency Domain 417

Javier Hernando, Climent Nadeu

Universitδt Politecnica de Catalunya, Spain

SESSION: T2D

Speech Coding I

Chair: Isabel Trancoso, INESC, Portugal

T2D.1 A Simple and Efficient Algorithm for the Compression of MBROLA Segment Databases 421

van der Vrecken Olivier, Nicolas Pierret, Thierry Dutoit, Vincent Pagel, Fabrice Malfrere

TCTS Lab, Belgium

T2D.2 A Segmental Formant Vocoder Based On Linearly Varying Mixtures of Gaussians 425

Parham Zolfaghari, Tony A. Robinson

Cambridge Univ., UK

T2D.3 Voice Mimic System Using Articulatory Codebook for Estimation of Vocal Tract Shape 429

Samir Chennoukh, Daniel Sinder, Gael Richard, James Flanagan

CAIP Center, Rutgers Univ., USA

T2D.4 Adaptive Transform Coding for Linear Predictive Residual 433

Damith J. Mudugamuwa, Alan B. Bradley

RMIT, Australia

T2D.5 Performance Evaluation of Objective Quality Measures for Coded Speech 437

Akira Takahashi, Nobuhiko Kitawaki, *Paolino Usai, †David Atkinson

NTT, Human Interface Labs, Japan

*Centro Studi e Laboratori Telecomunicazioni (CSELT), Italy

†NTIA, USA

T2D.6 Between Recognition and Synthesis - 300 Bits/ Second Speech Coding 441

Mohamed Ismail, Keith Ponting

Speech Research Unit, UK

SESSION: TMA

Feature Estimation II, Pitch and Prosody

Chair: Egidio Giachin, Centro Studi e Laboratori Telecomunicazioni (CSELT), Italy

TMA.1 A Modified Zero-Crossing Method for Pitch Detection in Presence of Interfering Sources 445

Francois Gaillard, Frederic Berthommier, *Gang Feng, *Jean-Luc Schwartz

ICP/INPG, France

*Institut de la Communication Parleι, France

TMA.2 Using Simulated Annealing Expectation Maximization Algorithm for Hidden Markov Model Parameters Estimation 449

Jacques Simonin, Chafic Mokbel

France Telecom, France

TMA.3 Covariation of Subglottal Pressure, F0 and Glottal Parameters 453

Gunnar Fant, *Stellan Hertegard, Anita Kruckenberg, Johan Liljencrants

KTH, Sweden

*Karolinska Hospital and Huddinge Univ. Hospital, Sweden

TMA.4 The Fractal Behaviour of Unvoiced Plosives: A Means for Classification 457

Anastasios Delopoulos, Maria Rangoussi

NTUA, Greece

TMA.5 A Method for Analysis of the Local Speech Rate Using an Inventory of Reference Units 461

Sumio Ohno, Hiroya Fujisaki, Hideyuki Taguchi

Science Univ. of Tokyo, Japan

TMA.6 Analysis and Modeling of Fundamental Frequency Contours of Greek Utterances 465

Hiroya Fujisaki, Sumio Ohno, Takashi Yagi

Science Univ. of Tokyo, Japan

TMA.7 Characteristics of Slow, Average and Fast Speech and their Effects in Large Vocabulary Continuous Speech Recognition 469

Fernando Martinez, Daniel Tapias, Jφrge Alvarez, Paloma Leon

Telefonica I+D, Spain

TMA.8 Analysis of Children's Speech: Duration, Pitch, and Formants 473

Sungbok Lee, Alexandros Potamianos, Shrikanth Narayanan

AT&T Labs-Research, USA

TMA.9 A Method of Measuring Formant Frequencies at High Fundamental Frequencies 477

Hartmut Traunmuller, *Anders Eriksson

Stockholm Univ., Sweden

*Univ. of Umeε, Sweden

TMA.10 Analysis of Speaking Rate Variation in Stress-Timed Languages 481

Tom Brondsted, Jens Printz Madsen

Aalborg Univ., Denmark

TMA.11 Automatic Identification of the Phoneme Boundaries Using a Mixed Parameter Model 485

Paul Micallef, *Ted Chilton

Univ. of Malta, Malta

*Univ. of Surrey, UK

TMA.12 Pitch Detection Reliability Assessment for Forensic Applications 489

Sergey Koval, Veronika Bekasova, Michael Khitrov, Andrey Raev

St Petersburg State Univ., Russia

TMA.13 Efficient Estimation of Perceptual Features for Speech Recognition 493

Zhihong Hu, Etienne Barnard

Oregon Graduate Institute, USA

TMA.14 Towards Decomposing the Sources of Variability in Speech 497

Narendranath Malayath, Hynek Hermansky, Alexander Kain

Oregon Graduate Institute of Science and Technology, USA

TMA.15 Use of Vector-Valued Dynamic Weighting Coefficients for Speech Recognition: Maximum Likelihood Approach 501

Rathinavelu Chengalvarayan

Bell Labs-Lucent Technologies, USA

TMA.16 Automatic Segmentation: Data-Driven Units of Speech 505

S.W. Beet, L. Baghai-Ravary

Aculab plc, UK

TMA.17 On Robust Time-Varying AR Speech Analysis Based on T-Distribution 509

Dejan Bajic

Institute of Applied Mathematics and Electronics, Yugoslavia

TMA.18 A Simple Phoneme Energy Model for the Greek Language and its Application to Speech Recognition 513

Dimitris Tambakas, Iliana Tzima, Nikos Fakotakis, George Kokkinakis

Univ. of Patras, Greece

TMA.19 A Macroscopic Analysis of an Emotional Speech Corpus 517

James E.H. Noad, Sandra P. Whiteside, Phil Green

Univ. of Sheffield, UK

TMA.20 Restoration of Pitch Pattern of Speech Based on a Pitch Generation Model 521

Hiroshi Shimodaira, Mitsuru Nakai, Akihiro Kumata

Japan Advanced Institute of Science and Technology, Japan

TMA.21 The Research of Correlation Between Pitch and Skin Galvanic Reaction at Change of Human Emotional State 525

A.V. Agranovski, O.Y. Berg, D.A. Lednov

Spetsvuzavtomatika Design Bureau, Russia

TMA.22 K-NN Versus Gaussian in HMM-Based Recognition System 529

Claude Montacie, Marie-Jose Caraty, Fabrice Lefθvre

Univ. Pierre et Marie Curie - CNRS, France

TMA.23 Spectral Methods for Voice Source Parameters Estimation 533

Boris Doval, Christophe D'Alessandro, Benoit Diard

LIMSI-CNRS, France

SESSION: TMB

Speech Synthesis Techniques

Chair: Rolf Carlson, KTH, Sweden

TMB.1 Optimising Unit Selection with Voice Source and Formants in the CHATR Speech Synthesis System

Wen Ding, Nick Campbell 537

ATR Interpreting Telecommunications Res. Labs., Japan

TMB.2 A New Framework to Provide High-Controlla-bility Speech Signal and the Development of a Workbench

for it 541

Masanobu Abe, Hideyuki Mizuno, Satoshi Takahashi, Shin'ya Nakajima

NTT, Japan

TMB.3 Shape-Invariant Prosodic Modification Algorithm for Concatenative Text-to-Speech Synthesis 545

Eduardo R. Banga, Carmen Garcia-Mateo, Xavier Fernandez-Salgado

Univ. of Vigo, Spain

TMB.4 An RNN-Based Spectral Information Generation for Mandarin Text-to-Speech 549

Hwang Shaw-Hwa, *Sin-Horng Chen, Chang Saga

Industrial Technology Research Institute (ITRI), Taiwan

*NCTU, Taiwan

TMB.5 Methods for Optimal Text Selection 553

Jan P.H. van Santen, *Adam L. Buchsbaum

Bell Labs-Lucent Technologies, USA

*AT&T Labs, USA

TMB.6 High Resolution Prosody Modification for Speech Synthesis 557

Francisco M. Gimenez de los Galanes, David Talkin

Entropic Research Lab, USA

TMB.7 Text-To-Speech Conversion with Neural Networks: A Recurrent TDNN Approach 561

Orhan Karaali, Gerald Corrigan, Ira Gerson, Noel Massey

Motorola, USA

TMB.8 Data Driven Formant Synthesis 565

Jesper Hogberg

KTH, Sweden

TMB.9 Speech Synthesis Using Non-Uniform Units in the Verbmobil Project 569

Simon King, *Thomas Portele, *Florian Hofer

Univ. of Edinburgh, UK

*Univ. of Bonn, Germany

TMB.10 On the Pronunciation Mode of Acronyms in Several European Languages 573

Isabel Trancoso, *M. Ceu Vianna

INESC, Portugal

*CLUL, Portugal

TMB.11 Evaluation of Speech Synthesis Systems for Dutch in Telecommunication Applications in GSM and PSTN Networks 577

Toni Rietveld, Joop Kerkhoff, *M.J.W.M. Emons, *E.J. Meijer, *Angelien A. Sanderman, *A.M.C. Sluijter

Univ. of Nijmegen, The Netherlands

*KPN Research, The Netherlands

TMB.12 Automatic Diphone Extraction for an Italian Text-to-Speech Synthesis System 581

Bianca Angelini, *Claudia Barolo, Daniele Falavigna, Maurizio Omologo, †Stefano Sandri

Istituto per la Ricerca Scientifica e Tecnologica, Italy

*Eikon Informatica, Italy

†Centro Studi e Laboratori Telecommunicazioni S.p.a, Italy

TMB.13 Simplification of TTS Architecture vs. Operational Quality 585

Eric Keller

Univ. of Lausanne, Switzerland

TMB.14 Felix - A TTS System with Improved Pre-Processing and Source Signal Generation 589

Georg Fries, Antje Wirth

Deutsche Telekom Berkom GmbH, Germany

TMB.15 Investigating the Limitations of Concatenative Synthesis 593

Mike Edgington

BT Laboratories, UK

TMB.16 Speech Coding and Synthesis Using Parametric Curves 597

Luis Miguel Teixeira de Jesus, Gavin C. Cawley

Univ. of East Anglia, UK

TMB.17 Automatically Clustering Similar Units for Unit Selection in Speech Synthesis 601

Alan W Black, Paul Taylor

Univ. of Edinburgh, UK

TMB.18 Improvements on a Trainable Letter-to-Sound Converter 605

Li Jiang, Hsiao-Wuen Hon, Xuedong Huang

Microsoft Corporation, USA

TMB.19 On a Cepstral Pitch Alteration Technique for Prosody Control in the Speech Synthesis System with High Quality 609

Bae MyungJin, KyuHong Kim, WonCheol Lee

Soongsil Univ., Korea

TMB.20 Diphone Concatenation Using a Harmonic Plus Noise Model of Speech 613

Yannis Stylianou, Thierry Dutoit, Juergen Schroeter

AT&T Labs-Research, USA

SESSION: TMC

Technology for S&L Acquisition, Speech Processing Tools

Chair: Petros Maragos, ILSP, Greece

TMC.1 The "Sketchboard": A Dynamic Interpretative Memory and its Use for Spoken Language

Understanding 617

Gerard Sabah

LIMSI-CNRS, France

TMC.2 Speech Technology Integration and Research Platform: A System Study 621

Qiru Zhou, Chin-Hui Lee, Wu Chou, Andrew Pargellis

Bell Labs, USA

TMC.3 Speech Recognition on SPHERIC - An IC for Command & Control Applications 625

Dieter Geller, Markus Lieb, Wolfgang Budde, Oliver Muelhens, Manfred Zinke

Philips GmbH, Germany

TMC.4 Muse: A Scripting Language for the Development of Interactive Speech Analysis and Recognition Tools 629

Michael K. McCandless, James R. Glass

SLS/LCS/MIT, USA

TMC.5 Language Learning Based on Non-Native Speech Recognition 633

Silke Witt, Steve Young

Cambridge Univ., UK

TMC.6 Task Modelling by Sentence Templates 637

Ute Kilian, Klaus Bader

Daimler Benz AG, Germany

TMC.7 Extraction And Representation Rhythmic Components of Spontaneous Speech 641

Shigeyoshi Kitazawa, Hideya Ichikawa, Satoshi Kobayashi, *Nishinuma Yukihiro

Shizuoka Univ., Japan

*Univ. de Provence, France

TMC.8 Automatic Pronunciation Scoring of Specific Phone Segments for Language Instruction 645

Yoon Kim, Horacio Franco, Leonardo Neumeyer

SRI International, USA

CCRMA,Stanford University,USA

TMC.9 Automatic Detection of Mispronunciation for Language Instruction 649

Orith Ronen, Leonardo Neumeyer, Horacio Franco

SRI International, USA

TMC.10 Continuous Formant-Tracking Applied to Visual Representations of the Speech and Speech Recognition

Agustin Alvarez, Rafael Martinez, Victor Nieto, Victoria Rodellar, Pedro Gomez

Universidad Politecnica de Madrid, Spain 653

TMC.11 A Call System Using Speech Recognition to Train the Pronunciation of Japanese Long Vowels, the Mora Nasal and Mora Obstruents 657

Goh Kawai, Keikichi Hirose

Univ. of Tokyo, Japan

TMC.12 An Educational and Experimental Workbench for Visual Processing of Speech Data 661

Jan Nouza, Miroslav Holada, Daniel Hajek

TU of Liberec, Czech Republic

TMC.13 A 3 Channel Digital CVSD Bit-Rate Conversion System Using a General Purpose DSP 665

Yong-Soo Choi, *Hong-Goo Kang, †Kim Sung-Youn, ‡Young-Cheol Park, †Dae-Hee Youn

Yonsei Univ., Korea

*AT&T Labs-Research, USA

†ASSP Lab.,Dept of Elec.Engin., Korea

‡Samsung Biomedical Research Institute, Korea

TMC.14 SLIM Prosodic Module for Learning Activities in a Foreigh Language 669

Rodolfo Delmonte, Mirela Petrea, Ciprian Bacalu

Universita Cα Garzoni-Moro Foscari, Italy

TMC.15 Barge-in Revised 673

Bernhard Kaspar, Karlheinz Schuhmacher, Stefan Feldes

Deutsche Telecom Berkom, GmbH, Germany

TMC.16 WaveEdit, An Interactive Speech Processing Environment for Microsoft Windows Platform 677

Mohammad Akbar

Univ. Joseph Fourier, France

TMC.17 Subarashii: Japanese Interactive Spoken Language Education 681

Farzad Ehsani, Jared Bernstein, Amir Najmi, Ognjen Todic

Entropic Research Labs, USA

TMC.18 Deploying Speech Applications Over the Web

David Goddeau, *William Goldenthal, *Chris Weikat

Digital Equipment Corp., USA 685

*Cambridge Research Laboratory,

TMC.19 CSLUsh: An Extendible Research Environment

Johan Schalkwyk, Jacques de Viller, Sarel van Vuuren , Pieter Vermuelen 689

Oregon Graduate Institute of Science and Technology, USA

TMC.20 A Flexible Client-Server Model for Multilingual CTS/TTS Development 693

Tibor Ferenczi, Ge'za Ne'meth, *Ga'bor Olaszy, Zoltan Gaspar

Technical Univ. of Budapest, Hungary

*Linguistics Institute of Hungarian Academy of Sciences, Hungary

TMC.21 Critically Sampled PR Filterbanks of Nonuniform Resolution Based on Block Recursive Famlet

Transform 697

Unto K. Laine

Helsinki Univ. of Technology, Finland

TMC.22 Automatic Detection of Accent in English Words Spoken by Japenese Students 701

Nobuaki Minematsu, Nariaki Ohashi, Seiichi Nakagawa

Toyohashi Univ. of Technology, Japan

TMC.23 An English Conversation and Pronunciation CAI System Using Speech Recognition Technology 705

Yasuhiro Taniguchi, Allan A. Reyes, Hideyuki Suzuki, Seiichi Nakagawa

Toyohashi Univ. of Technology, Japan

TMC.24 Bringing Spoken Language Systems to the Classroom 709

Stephen Sutton, Ed Kaiser, A. Cronk, Ron Cole

Oregon Graduate Institute, USA

TMC.25 Automatic Assessment of Foreign Speakers' Pronunciation of Dutch 713

Catia Cucchiarini, Lou Boves

Univ. of Nijmegen, The Netherlands

TMC.26 Use of Low Power EM Radar Sensors for Speech Articulator Measurements 717

John F. Holzrichter, Greg C Burnett

Lawrence Livermore National Laboratory, USA

TMC.27 Real Time Measurements of the Vocal Tract Resonances During Speech 721

Julien Epps, Annette Dowd, John Smith, Joe Wolfe

Univ. of New South Wales, Australia

SESSION: TMD

Phonetics and Phonology

Chair: Thomas Portele, Univ. of Bonn, Germany

TMD.1 Linguistic Criteria for Building and Recording Units for Concatenative Speech Synthesis in Brazilian

Portoguese 725

Eleonora Cavalcante Albano, Patricia Aparecida Aquino

UNICAMP, Brazil

TMD.2 "Four-and-Twenty,Twenty-Four".What's in a Number? 729

Knut Kvale, *Arne Kjell Foldvik

Telenor R&D, Norway

*Norwegian Univ. of Science and Technology, Norway

TMD.3 Vowel Nasalization in Brazilian Portuguese: An Articulatory Investigation 733

Antonio de Moraes Joao

Universidade Federal do Rio de Janeiro, Brazil

TMD.4 Rhythmic Organization Pecularities of the Spoken Text 737

Elena Steriopolo

Kiev State Linguistic Univ., Ukraine

TMD.5 Obtaining Confidence Measures from Sentence Probabilities 739

Bernhard Rueber

Philips GmbH, Germany

TMD.6 Sentence Design for Speech Synthesis and Speech Recognition Database by Phonetic Rules 743

Yiqing Zu

Chinese Academy of Social Sciences, China

TMD.7 Identification of Regional Variants of High German from Digit Sequences in German Telephone Speech 747

Christoph Draxler, Susanne Burger

Ludwig-Maximilians-Univ. Muenchen, Germany

TMD.8 Aerodynamic Constraints on the Production of Palatalized Trills: The Case of the Slavic Trilled [r] 751

Darya Kavitskaya

UC Berkeley, USA

TMD.9 An Experimental Phonetic Study of the Interrelationship Between Prosodic Phrase and Syntactic Structure 755

Cheol-jae Seong, *Sanghun Kim

Chungnam National Univ., Korea

*Electronics and Telecommunications Research Institute, Korea

TMD.10 Individual Differences Between Vowel Systems of German Speakers 759

Sebastian J.G.G. Heid

Institute fuer Phonetik und Sprachliche Kommunikation, Germany

TMD.11 Tempo and Its Change in Spontaneous Speech

Anton Batliner, *Andreas Kiebling, †Ralf Kompe, Heinrich Niemann, Elmar Noeth 763

Univ. of Erlangen-Nόremburg, Germany

*Ericsson Eurolab,

†Sony Stuttgart Technology Center,

TMD.12 A Corpus-Based Approach to Diphthong Analysis of Standard Slovenian 767

Bojan Petek, Rastislav Sustarsic

Univ. of Ljubljana, Slovenia

TMD.13 Catalan Vowel Duration 771

Loudres Aguilar, Julia A. Gimenez, Maria Machuca, Rafael Marin, Montse Riera

Universitδt Autonoma de Barcelona, Spain

TMD.14 The Intonation of Vocatives in Spoken Neapolitan Italian 775

Maria Rosaria Caputo

Napoli, Italy

TMD.15 A Comparative Acoustic Study of Spontaneous and Read Italian Speech 779

Caldognetto Emanuela Magno, Claudio Zmarich, Franco Ferrero

CNR, Italy

TMD.16 A Contribution to the Estimation of Naturalness in the Intonation of Italian Spontaneous Speech 783

Mario Refice, *Michelina Savino, *Martine Grice

DEE,Politecnico di Bari, Italy

*Univ. of the Saarland,Germany

TMD.17 Diphthongs and the Process of Monophthongization in Austria German: A First

Approach 787

Sylvia Moosmόller

Austrian Academy of Sciences, Austria

TMD.18 The Prosody of Broad and Narrow Focus in English: Two Experiments 791

Steve Hoskins

duPont Hospital for Children/Univ. of Delaware, USA

TMD.19 The Domain of Accentual Lengthening in Scottish English 795

Alice Turk, Laurence White

Edinburgh Univ., UK

TMD.20 Spontaneous Dialogue: Some Results About the F0 Predictions of a Pragmatic Model of Information

Processing 799

Mariette Bessac, Geneviθve Caelen-Haumont

Univ. Joseph Fourier, France

TMD.21 Phonetic Characteristics of Double Articulations in Some Mangbutu-Efe Languages 803

Didier Demolin, Bernard Teston

Univ. Libre de Bruxelles, Belgium,

Univ. de Lyon 2, France and

Univ. d'Aix en Provence, France

TMD.22 Intonation Modeling for the Southern Dialects of the Basque Language 807

Inmaculada Hernaez, Inaki Gaminde, Borja Etxebarria, Pilartxo Etxebarria

Univ. of the Basque Country, Spain

SESSION: T3A

Confidence Measures in ASR

Chair: Jose M. Pardo, UPM, Spain

T3A.1 A Low-Cost Phonetic Transcription Method 811

Pablo Fetter, Udo Haiber, Peter Regel-Brietzmann

Daimler Benz AG, Germany

T3A.2 Word and Acoustic Confidence Annotation for Large Vocabulary Speech Recognition 815

Lin Chase

Carnegie Mellon Univ., USA

T3A.3 A Senone Based Confidence Measure for Speech Recognition 819

Zachary Bergen, *Wayne Ward

Berdy Medical Systems, USA

*Carnegie Mellon Univ., USA

T3A.4 OOV Utterance Detection Based on the Recognizer Response Function 823

Erica Bernstein, Ward R. Evans

The MITRE Corporation, USA

T3A.5 Estimating Confidence Using Word Lattices 827

Thomas Kemp, Thomas Schaaf

Univ. of Karlsruhe, Germany

T3A.6 Improved Estimation,Evaluation and Applications of Confidence Measures for Speech Recognition 831

Man-Hung Siu, Herbert Gish, Fred Richardson

BBN Inc, USA

SESSION: T3B

Speaker and Language Identification

Chair: Sadaoki Furui, NTT, Japan

T3B.1 Improved Speaker Verification System With Limited Training Data On Telephone Quality Speech 835

Salleh Hussain, Fergus R. McInnes, Mervyn A. Jack

Univ. of Edinburgh, UK

T3B.2 Verbal Information Verification 839

Qi Li, Biing-Hwang Juang, Qiru Zhou, Chin-Hui Lee

Bell Labs, USA

T3B.3 A Segment-Based Speaker Verification System Using SUMMIT 843

Sridevi V. Sarma, Victor Zue

MIT, USA

T3B.4 Speaker Verification on the Word Wide Web 847

Michael Sokolov

Digital Equipment Corp., USA

T3B.5 Text-Prompted Versus Sound-Prompted Passwords in Speaker Verification Systems 851

Johan Lindberg, Hakan Melin

KTH, Sweden

T3B.6 GMM Sample Statistic Log-Likelihoods for Text-Independent Speaker Recognition 855

Michael Schmidt, John Golden, Herbert Gish

BBN Systems and Technologies, USA

SESSION: T3C

Perception of Prosody

Chair: Joseph Olive, Bell Labs, USA

T3C.1 The Influence of Phrase Boundaries on Perceived Prominence in Two-Peak Intonation Contours 859

Toni Rietveld, Carlos Gussenhoven

Univ. of Nijmegen, The Netherlands

T3C.2 Testing the Meaning of Four Dutch Pitch Accent Types 863

Johanneke Caspers

Univ. of Leiden, The Netherlands

T3C.3 A Perceptual Study for Modelling Speaker-Depen-ent Intonation in TTS and Dialog Systems 867

Joachim J. Mersdorf, Thomas Domhover

Ruhr Univ., Germany

T3C.4 Can we Perceive Attitudes Before the End of Sentences? The Gating Paradigm for Prosodic Contours

Veronique Auberge, Tuulikki Grepillat, A. Rilliard 871

Univ. Stendhal, France

T3C.5 To What Extent is Perceived Focus Determined by F0-Cues? 875

Mattias Heldner, Eva Strangert

Umeε Univ., Sweden

T3C.6 Temporal-Alignment Categories of Accent-Lending Rises and Falls 879

David House, *Dik Hermes, †Frederic Beaugendre

Lund Univ., Sweden

*IPO, The Netherlands

†Lernout & Hauspie Speech Products, The Netherlands

SESSION: T3D

Applications of Speech Technology I

Chair: Klaus Fellbaum, Univ. of Cottbus, Germany

T3D.1 WebGalaxy -- Integrating Spoken Language And Hypertext Navigation 883

Raymond Lau, Giovanni Flammia, Christine Pao, Victor Zue

MIT, USA

T3D.2 Pitch Estimation of Singing for Re-Synthesis and Musical Transcription 887

Micheal Carey, Eluned S. Parris, *Graham D. Tattersall

Ensigma Ltd, UK

*Snape Signals Research, UK

T3D.3 Automated Lip Synchronisation for Human-Computer Interaction and Special Effect Animation

Christian Martyn Jones, Satnam Singh Dlay 891

Univ. of Newcastle, UK

T3D.4 Developing Web-Based Speech Applications 895

Charles T. Hemphill, Yeshwant Muthusamy

Texas Instruments, USA

T3D.5 Automatic Post-Synchronization of Speech Utterances 899

Werner Verhelst

Vrije Univ. of Brussel, Belgium

T3D.6 Automatic Generation of Hyperlinks Between Audio and Transcript 903

Jordi Robert-Ribes, *Rami G. Mukhtar

Advanced Computational Systems CRC, Australia

*CSIRO, Australia

SESSION: T4A

Spontaneous Speech Recognition

Chair: Roberto Billi, Centro Studi e Laboratori Telecomunicazioni (CSELT), Italy

T4A.1 Transcription of Broadcast News 907

Jean-Luc Gauvain, Lori Lamel, Gilles Adda, Martine Adda-Decker

LIMSI, France

T4A.2 Can Continuous Speech Recognizers Handle Isolated Speech? 911

Fil Alleva, Xuedong Huang, Mei-Yuh Hwang, Li Jiang

Microsoft Corporation, USA

T4A.3 Toward Automatic Transcription of Japanese Broadcast News 915

Tatsuo Matsuoka, *Yuichi Taguchi, Katsutoshi Ohtsuki, †Sadaoki Furui, *Katsuhiko Shirai

NTT, Japan

*Waseda Univ., Japan

†Tokyo Institute of Technology, Japan

T4A.4 Automatic Detection of Semantic Boundaries 919

Mauro Cettolo, Anna Corazza

Istituto per la Ricerca Scientifica e Tecnologica (IRST), Italy

T4A.5 Connected Digit Recognition in Spontaneous

Speech 923

Etienne Bauche, Bojana Gajic, Yasuhiro Minami, Tatsuo Matsuoka, *Sadaoki Furui

NTT, Japan

*Tokyo Institute of Technology, Japan

T4A.6 Advances in Transcription of Broadcast News 927

Francis Kubala, Hubert Jin, *Spyros Matsoukas, Long Nguyen, Richard Schwartz, John Makhoul

BBN Systems and Technologies, USA

*Northearstern Univ., USA

SESSION: T4B

Language Specific Segmental Features

Chair: Gunnar Fant, KTH, Sweden

T4B.1 The Domain of Final Lengthening in Production and Perception in Dutch 931

Tina Cambier-Langeveld, Marina Nespor, *Vincent J. van Heuven

Univ. of Amsterdam/HIL, The Netherlands

*Leiden Univ./HIL, The Netherlands

T4B.2 Voicing Assimilation as a Cue for Cluster Identification 935

Christine Meunier

Univ. de Genθve, Switzerland

T4B.3 On the Perceptual Relevance of Degemination in Dutch 939

Saskia M.M. te Riele, Manon Loef, van O. Herwijen

Utrecht Univ., The Netherlands

T4B.4 Does Deletion of French SCHWA Lead to Neutralization of Lexical Distinctions? 943

Cecile Fougeron, *Donca Steriade

UCLA & Paris III, USA & France

*UCLA, USA

T4B.5 An Approach of the Catalan Palatals Discrimination Based on Durational Patterns of Spectral Evolution 947

Marielle Bruyninckx, Bernard Harmegnies

Univ. de Mons-Hainaut, Belgium

T4B.6 Syllable and Segment Duration at Different Speaking Rates in the Slovenian Language 951

Jerneja Gros, Nikola Pavesic, France Mihelic

Univ. of Ljubljana, Slovenia

SESSION: T4C

Speaker Recognition I

Chair: George Doddigton, SRI, USA

T4C.1 Hybrid Networks Based on RBFN and GMM for Speaker Recognition 955

Wei-Ying Li, Douglas O'Shaughnessy

Univ. du Quebec, Canada

T4C.2 A Discriminative Training Algorithm for Gaussian Mixture Speaker Models 959

Jialong He, Li Liu, Gόnther Palm

Univ. of Ulm, Germany

T4C.3 Comparison of Background Normalization Methods for Text-Independent Speaker Verification 963

Douglas A. Reynolds

MIT Lincoln Laboratory, USA

T4C.4 Speaker Verification with Limited Enrollment Data

Owen Kimball, Michael Schmidt, Herbert Gish, Jason Waterman 967

BBN Systems and Technologies, USA

T4C.5 Speaker Verification in the Telephone Network: Research Activities in the Cave Project 971

Frederic Bimbot, *Hans-Peter Hutter, †Cedric Jaboulet, ‡Johan W. Koolwaaij, ¤Johan Lindberg, Jean Benoit Pierrot

ENST/CNRS, France

*Ubilab-UBS, Switzerland

†IDIAP, Switzerland

‡KUN, The Netherlands

¤KTH, Sweden

T4C.6 Speaker Verification with GSM Coded Telephone Speech 975

Mark Kuitert, Lou Boves

Univ. of Nijmegen, The Netherlands

SESSION: T4D

Speech Synthesis:Linguistic Analysis

Chair: Bjφrn Granstrφm, KTH, Sweden

T4D.1 Parsers, Prominence, and Pauses 979

Nick Campbell, Tony Hebert, Ezra Black

ATR Interpreting Telecommunications Res. Labs., Japan

T4D.2 Automatic Assignment of Part-of-Speech to Out-of-Vocabulary Words for Text-To-Speech Processing 983

Frederic Bechet, Marc El-Beze

Univ. of Avignon, France

T4D.3 Text-to-Prosody Parsing in an Italian Speech Synthesizer. Recent Improvements 987

Barbara Gili Fivela, *Silvia Quazza

Scuola Normale Superiore, Italy

*Centro Studi e Laboratori Telecomunicazioni (CSELT), Italy

T4D.4 Tagging Syllables 991

Brigitte Krenn

Univ. of the Saarland, Germany

T4D.5 Assigning Phrase Breaks from Part-of-Speech Sequences 995

Alan W. Black, Paul Taylor

Univ. of Edinburgh, UK

T4D.6 Prediction of Word Prominence 999

Christina Widera, Thomas Portele, Maria Wolters

Univ. of Bonn, Germany

SESSION: TAA

Speech Analysis & Modelling

Chair: Pierre Badin, ICP, INPG, France

TAA.1 Acoustic and Perceptual Properties of Phonemes in Continuous Speech as a Function of Speaking Rate 1003

Hisao Kuwabara

Tokyo Univ. of Science & Technology, Japan

TAA.2 New Results in Vowel Production: MRI, EPG, and Acoustic Data 1007

Shrikanth Narayanan, *Abeer Alwan, *Yong Song

AT&T Labs-Research, USA

*Univ. of California, USA

TAA.3 The Temporal Properties of Spoken Japanese are Similar to those of English 1011

Takayuki Arai, Steven Greenberg

International and Computer Science Institute and

Univ. of California at Berkeley, USA

TAA.4 The Amplitudes of the Peaks in the Spectrum: Data from /a/ Context 1015

Anna Esposito

IIASS, Italy

TAA.5 Acoustical Characteristics of Speech and Voice in Speech Pathology 1019

Natalija Bolfan-Stosic, Mladen Hedjever

Univ. of Zagreb, Croatia

TAA.6 Pronuncation Modeling Applied to Automatic Segmentation of Spontaneous Speech 1023

Andreas Kipp, *Maria-Barbara Wesenick, *Florian Schiel

IPSK Univ. of Munich, Germany

*Ludwig-Maximilians-Univ. Muenchen, Germany

TAA.7 Dynamic and Static Improvements to Lexical Baseforms 1027

Simon Downey, Richard Wiseman

BT Laboratories, UK

TAA.8 Signal Driven Generation of Word Baseforms from Few Examples 1031

Andreas Hauenstein

pc-plus GmbH, Germany

TAA.9 Modeling the Acoustic Differences Between L1 and L2 Speech: The Short Vowels of Africaans and South-African English 1035

Elizabeth Botha, *Louis C.W. Pols

Univ. of Pretoria, South Africa

*Univ. of Amsterdam, The Netherlands

TAA.10 Laryngeal Movements and Speech Rate: An X-ray investigation 1039

Beatrice Vaxelaire, Rudolph Sock

Institut de Phonetique de Strasbourg, France

TAA.11 How Flexible is the Human Voice?-A Case Study of Mimicry 1043

Anders Eriksson, Par Wretling

Umea Univ., Sweden

TAA.12 The Effect of Low-Pass Filtering on Estimated Voice Source Parameters 1047

Helmer Strik

Univ. of Nijmegen, The Netherlands

TAA.13 Vowel Development of /i/ and /u/ in 15-36 Month Old Children at Risk and Not at Risk to Stutter 1051

Susan M. Fosnot

Organization UCLA, USA

TAA.14 Optopalatograph: Development of a Device for Measuring Tongue Movement in 3D 1055

Alan Wrench, Alan McIntosh, William Hardcastle

Queen Margaret College, UK

TAA.15 Speech Synthesis and Prosody Modification Using Segmentation and Modelling of the Excitation Signal

Juana M. Gutierrez-Arriola, Francisco M. Gimenez de los Galanes, Mohammed H. Savoji, Jose M. Pardo

Universidad Politecnica de Madrid, Spain 1059

TAA.16 How Can the Control of the Vocal Tract Limit the Speaker's Capability to Produce the Ultimate Perceptive Objectives of Speech? 1063

Christophe Savariaux, Louis-Jean Boe, Pascal Perrier

ICP, INPG, France

TAA.17 A Step Toward General Model for Symbolic Description of the Speech Signal 1067

Goran S. Jovanovic

Institute for Applied Mathematics and Electronics, Yugoslavia

TAA.18 Referring in Long Term Speech by Using Orientation Patterns Obtained from Vector Field of Spectrum Pattern 1071

Kiyoshi Furukawa, Masayuki Nakazawa, Takashi Endo, Oka Ryuichi

Real World Computing Partnership, Japan

SESSION: TAB

Robustness in Recognition and Signal Processing II

Chair: Alex Waibel, Carnegie Mellon Univ., USA

TAB.1 Adaptation of Time Differentiated Cepstrum for Noisy Speech Recognition 1075

Tai-Hwei Hwang, *Lee-Min Lee, Hsiao-Chuan Wang

National Tsing-Hua Univ., ROChina

*Mingchi Institute of Technology, ROChina

TAB.2 On The Importance of Various Modulation Frequencies for Speech Recognition 1079

†Noboru Kanedera, *Takayuki Arai, Hynek Hermansky, Misha Pavel

Oregon Graduate Institute of Science and Technology, USA

*International Computer Science Institute, USA

†Ishikawa National College of Technology,Japan

TAB.3 A Robust RNN-Based Pre-Classification for Noisy Mandarin Speech Recognition 1083

Wei-Tyng Hong, Sin-Horng Chen

National Chiao Tung Univ., ROChina

TAB.4 A Parallel Environment Model (PEM) for Speech Recognition and Adaptation 1087

Mazin Rahim

AT&T Labs, USA

TAB.5 Adaptive Model Combination for Robust Speech Recognition in Car Environments 1091

Volker Schless, Fritz Class

Daimler Benz AG, Germany

TAB.6 A Comparative Study of Speech Detection Methods

Stefaan Gerven Van, Fei Xie 1095

KULEUVEN-ESAT, Belgium

TAB.7 Voice Activity Detection Using Source Separation Techniques 1099

Nikos Doukas, Patrick Naylor, Tania Stathaki

Imperial College, UK

TAB.8 Applying Blind Signal Separation to the Recognition of Overlapped Speech 1103

Tomohiko Taniguchi, Shoji Kajita, Kazuya Takeda, Fumitada Itakura

Nagoya Univ., Japan

TAB.9 Multiresolution Channel Normalization for ASR in Reverberant Environments 1107

Carlos Avendano, Sangita Tibrewala, Hynek Hermansky

Oregon Graduate Institute of Science and Technology, USA

TAB.10 A Speech Pre-Processing Technique for End-Point Detection in Highly Non-Stationary Environments

Rafael Martinez, Agustin Alvarez, Vilda Pedro Gomez, Mercedes Perez, Victor Nieto, Victoria Rodellar

Universidad Politecnica de Madrid, Spain 1111

TAB.11 Application of Several Channel and Noise Compensation Techiques for Robust Speaker Recognition

Laura Docio-Fernandez, Carmen Garcia-Mateo

Univ. of Vigo, Spain 1115

TAB.12 Knowing the Wheat from the Weeds in Noisy

Speech 1119

Hany Agaiby, Thomas J. Moir

Univ. of Paisley, UK

TAB.13 Model-Based Approach for Robust Speech Recognition in Noisy Environements with Multiple Noise Sources 1123

Do Yeong Kim, *Nam Soo Kim, Chong Kwan Un

KAIST, Korea

*SAIT, Korea

TAB.14 Normalization of Speaker Variability by Spectrum Warping for Robust Speech Recognition 1127

Y.C. Chu, Charlie Jie, Vincent Tung, Ben Lin, Richard Lee

Philips Taiwan, ROChina

TAB.15 LPC Poles Tracker for Music/Speech/Noise Segmentation and Music Cancellation 1131

Stephane H. Maes

IBM, USA

TAB.16 Comparative Evaluations of Several Front-Ends for Robust Speech Recognition 1135

Doh-Suk Kim, Jae-Hoon Jeong, Soo-Young Lee, Rhee M. Kil

Korean Advanced Instiute of Science and Technology, Korea

TAB.17 Speaker Normalization Through Formant-Based Warping of the Frequency Scale 1139

Evandro B. Gouvea, Richard M. Stern

Carnegie Mellon Univ., USA

TAB.18 The Use of Cepstral Means in Conversational Speech Recognition 1143

Martin Westphal

Univ. Karlsruhe, Germany

TAB.19 Compensation for Environmental and Speaker Variability by Normalization of Pole Locations 1147

Juan M. Huerta, Richard M. Stern

Carnegie Mellon Univ., USA

TAB.20 Cellular Phone Speech Recognition: Noise Compensation vs. Robust Architectures 1151

Jean-Baptiste Puel, Regine Andrι-Obrecht

IRIT - Universitaire Paul Sabatier, France

TAB.21 Speech Recognition in Noise Using On-Line HMM Adaptation 1155

TungHui Chiang

Advanced Technology Center (ATC), Computer and Communication Labs (CCL), Industrial Technology Research Instiitute (ITRI), ROChina

SESSION: TAC

Acoustic Modelling II

Chair: Vassilios Digalakis, Technical Univ. of Crete, Greece

TAC.1 Incorporating Linguistic Knowledge and Automatic Baseform Generation in Acoustic Subword Unit Based Speech Recognition 1159

Trym Holter, TorBjφrn Svendsen

The Norwegian Univ. of Science and Technology (NTNU), Norway

TAC.2 Modeling and Decoding of Crossword Context Dependent Phones in the Philips Large Vocabulary Continuous Speech Recognition System 1163

Peter Beyerlein, Meinhard Ullrich, Patricia Wilcox

Philips GmbH, Germany

TAC.3 Modelling Inter-Frame Dependence with Preceeding and Succeeding Frames 1167

Philip Hanna, Ji Ming, Peter O'Boyle, F.Jack Smith

Queen's Univ. of Belfast, N. Ireland

TAC.4 Continuous Speech Recognition Using Syllables

Rhys James Jones, *Simon Downey, John S. Mason

Univ. of Wales Swansea, UK 1171

*BT Laboratories, UK

TAC.5 A New Approach to Generalized Mixture Tying for Continuous HMM-Based Speech Recognition 1175

Daniel Willett, Gerhard Rigoll

Gerhard-Mercator-Univ. Duisburg, Germany

TAC.6 State Tying for Context Dependent Phoneme

Models 1179

Klaus Beulen, Elmar Bransch, Hermann Ney

RWTH Aachen Univ. of Technology, Germany

TAC.7 A Novel Node Splitting Criterion in Decision Tree Construction for Semi-Continuous HMMS 1183

Jacques Duchateau, Kris Demuynck, Dirk Van Compernolle Katholieke Univ. Leuven-E.S.A.T, Belgium

TAC.8 Creating Unseen Triphones by Phone Concatenation in the Spectral, Cepstral and Formant Domains 1187

Mats Blomberg

KTH, Sweden

TAC.9 Creating Large Subword Units for Speech Recognition 1191

Thilo Pfau, Manfred Beham, *W. Reichl, Gόnther Ruske

Muenchen Univ. of Technology, Germany

*Bell Laboratories, USA

TAC.10 Segmental Modeling Using a Continuous Mixture of Non-Parametric Models 1195

Jacob Goldberger, David Burshtein, *Horacio Franco

Tel Aviv Univ., Israel

*SRI International, USA

TAC.11 Segmentation and Modeling in Segment-Based Recognition 1199

Jane W. Chang, James R. Glass

MIT, USA

TAC.12 Using Syllables in a Hybrid HMM-ANN Recognition System 1203

Alfred Hauenstein

Siemens AG, Germany

TAC.13 Noise Robust Segment-Based Word Recognition Using Vector Quantisation 1207

Ramalingam Hariharan, Juha Hakkinen, Kari Laurila, Janne Suontausta

Nokia Research Center, Finland

TAC.14 Viterbi Based Splitting of Phoneme HMM's 1211

Luis Javier Rodriguez, *Ines M. Torres

Universidad del Pais Vasco., Spain

*UPV/EHU, Spain

TAC.15 The Demiphone: An efficient Subword Unit for Continuous Speech Recognition 1215

Jose B. Marino, A. Nogueiras, Antonio Bonafonte

Universitδt Poliitecnica de Catalunya, Spain

TAC.16 Organizing Phone Models Based on Piecewise Linear Segment Lattices of Speech Samples 1219

Hiroaki Kojima, Kazuyo Tanaka

Electrotechnical Lab, Japan

TAC.17 Automatic Architecture Design by Likelihood-Based Context Clustering with Crossvalidation 1223

Ivica Rogina

Univ. Karlsruhe, Germany

TAC.18 Towards Articulatory Speech Recognition: Learning Smooth Maps to Recover Articulator Information 1227

Sam Roweis, *Abeer Alwan

California Institute of Technology, USA

*UCLA, USA

TAC.19 Selection of the Most Effective Set of Subword Units for an HMM-Based Speech Recognition System

Anastasios Tsopanoglou, *Nikos Fakotakis

KNOWLEDGE SA, Greece 1231

*Univ. of Patras, Greece

TAC.20 Multi-Band Continuous Speech Recognition

Christophe Cerisara, Jean-Paul Haton, Jean-Francois Mari, Dominique Fohr

CRIN-CNRS & INRIA Lorraine, France 1235

TAC.21 The Design of Acoustic Parameters for Speaker-Independent Speech Recognition 1239

Nabil N. Bitar, Carol Y. Espy-Wilson

Boston Univ., USA

SESSION: TAD

Speech Coding II

Chair: John Mourjopoulos, Univ. of Patras, Greece

TAD.1 High Quality Split-Band LPC Vocoder and its Fixed Point Real Time Implementation 1243

Stephane Villette, Milos Stefanovic, Ian Atkinson, Ahmet Kondoz

Univ. of Surrey, UK

TAD.2 Missing Packet Recovery Techniques for DM Coded Speech 1247

Wen-Whei Chang, *Hwai-Tsu Chang, *Wan-Yu Meng

National Chiao-Tung Univ., ROChina

*Industrial Technology Research Institute, ROChina

TAD.3 Spectral Sensitivity of LSP Parameters and Their Transformed Coefficients 1251

Vu Hai Le, Laszlo Lois

Technical Univ. of Budapest, Hungary

TAD.4 Reducing the Complexity of the LPC Vector Quantizer Using the K-D Tree Search Algorithm 1255

V. Ramasubramanian, K.K. Paliwal

ATR Interpreting Telecommunications Res. Labs., Japan

TAD.5 LPC Quantization Using Wavelet Based Temporal Decomposition of the LSF 1259

Aweke N. Lemma, *W.Bastiaan Kleijn, Ed F. Deprettere

Delft Univ. of Technology, The Netherlands

*KTH, Sweden

TAD.6 A Novel 1.7/2.4 KB/S DCT Based Prototype Interpolation Speech Coding System 1263

Costas S. Xydeas, Gokhan H. Ilk

Univ. of Manchester, UK

TAD.7 Improved Regular Pulse VSELP Coding of Speech at Low Bit-Rates 1267

Yong-Soo Choi, *Hong-Goo Kang, Sang-Wook Park, †Jae-Ha Yoo, Dae-Hee Youn

Yonsei University, Korea

*AT&T Labs, USA

†LG Electronic Inc., Korea

TAD.8 Joint Estimation of Pitch,Band Magnitudes, and V\UV Decisions for MBE Vocoder 1271

Yong Duk Cho, Hong Kook Kim, Moo Young Kim, Sang Ryong Kim

Samsung Advanced Institute of Technology, South Korea

TAD.9 A New Distance Measure in LPC Coding: Application for Real Time Situations 1275

Balazs Kovesi, Samir Saoudi, Jean Marc Boucher, *Gabor Horvath

ENST-Br, France

*Technological Univ. of Budapest, Hungary

TAD.10 Consideration of Processing Strategies for Very-Low-Rate Compression of Wideband Speech Signals with known Text Transcription 1279

Peter Vepyek, Alan B. Bradley

RMIT, Australia

TAD.11 Zero-Redundancy Error Protection for CELP Speech Codecs 1283

Norbert Gortz

Univ. of Kiel, Germany

TAD.12 Low Bit Rate Speech Coding Using an Improved HSX Model 1287

Ridha Matmti, Milan Jelinek, Jean-Pierre Adoul

Univ. of Sherbrooke, Canada

TAD.13 Phonetic Vocoding with Speaker Adaptation

Carlos M. Ribeiro, Isabel Trancoso

INESC, Portugal 1291

TAD.14 Quantization of Spectral Sequences Using Variable Length Spectral Segments for Speech Coding at Very Low Bit Rate 1295

Geneviθve Baudoin, *Jan Cernocky, †Gerard Chollet

ESIEE, France

*FEIVUT, France

†ENST, France

TAD.15 On Modeling Event Functions in Temporal Decomposition Based Speech Coding 1299

Shahrokh Ghaemmaghami, Mohamed Deriche, Boualem Boashash

Queensland Univ. of Technology, Australia

TAD.16 Phase Quantization by Pitch-Cycle Waveform Coding in Low Bit Rate Sinusoidal Coders 1303

Soledad Torres, *Javier F Casajus-Quiros

Universidad de Valladolid, Spain

*Universidad Politecnica de Madrid, Spain

TAD.17 A Perceptual Study of the Greek Vowel Space Using Synthetic Stimuli 1307

Antonis Botinis, *Marios Fourakis, †John W. Hawks

Athens Univ., Greece

*The Ohio State Univ., USA

†Kent State Univ., USA

TAD.18 Mixed Multi-Band Excitation Coder Using Frequency Domain Mixture Function (FDMF) for a Low-Bit Rate Speech Coding 1311

Woo-Jin Han, Sung-Joo Kim, Yung-Hwan Oh

KAIST, Korea

TAD.19 Robust GSM Speech Decoding Using the Channel Decoder's Soft Output 1315

Tim Fingscheidt, Olaf Scheufen

Aachen Univ. of Technology, Germany

TAD.20 A Low-Bit-Rate Speech Coder Using Adaptive Line Spectral Frequency Prediction 1319

Carl W. Seymour, Tony A. Robinson

Cambridge Univ., UK

SESSION: W1A

Dialogue Systems:Applications

Chair: Norman Fraser, Univ. of Surrey, UK

W1A.1 Experiments in Spoken Queries for Document Retrieval 1323

James Barnett, Steve Anderson, *John Broglio, Mona Singh, †R. Hudson, †S.W. Kuo

Dragon Systems, USA

*Univ. of Massachusetts, USA

†Intermetrics Inc., USA

W1A.2 Towards an Automated Directory Information System 1327

Frank Seide, *Andreas Kellner

Philips Research Laboratories Taipei, Taiwan

*Philips GmbH Aachen, Germany

W1A.3 A Strategy for Mixed-Initiative Dialogue Control

Lars Bo Larsen

Aalborg Univ., Denmark 1331

W1A.4 On the Design of Effective Speech-Based Interfaces for Desktop Applications 1335

Jim Hugunin, Victor Zue

MIT, USA

W1A.5 Dialogue Strategies Guiding Users to their Communicative Goals 1339

Matthias Denecke, Alex Waibel

Carnegie Mellon Univ., USA

W1A.6 A Speech Interface for Forms on WWW 1343

Sunil Issar

Carnegie Mellon Univ., USA

SESSION: W1B

Speech Production Modelling

Chair: Michael D. Riley, AT&T Labs, USA

W1B.1 Voice Conversion by Codebook Mapping of Line Spectral Frequencies and Excitation Spectrum 1347

Levent M Arslan, David Talkin

Entropic Research Laboratory, USA

W1B.2 Optimal State Dependent Spectral Represetation for HMM Modeling: A New Theoretical Framework

Chafic Mokbel, *Guillaume Gravier, *Gerard Chollet

France Telecom, France

*ENST, France 1351

W1B.3 Speech Analysis and Systems Using an AM-FM Molulation Model 1355

Alexandros Potamianos, *Petros Maragos

AT&T Labs-Research, USA

*ILSP & Georgia Tech, Greece & USA

W1B.4 Synthesis of Fricative Consonants by Audiovisual-to-Articulatory Inversion 1359

Khaled Mawass, Pierre Badin, Gerard Bailly

ICP, INPG, France

W1B.5 New Transformations of Cepstral Parameters for Automatic Vocal Tract Length Normalization in Speech Recognition 1363

Tom Claes, *Ioannis Dologlou, Louis ten Bosch, Dirk Van Compernolle

Lernout & Hauspie Speech Products, Belgium

*K.U Leuven-E.S.A.T, Belgium

W1B.6 A Multiresolutionally Oriented Approach for Determination of Cepstral Features in Speech Recognition

Simon Dobrisek, France Mihelic, Nikola Pavesic

Univ. of Ljubljana, Slovenia 1367

SESSION: W1C

Speaker Recognition II

Chair: Aaron Rosenberg, AT & T Labs, USA

W1C.1 Speaker Identification with User-Selected Password Phrases 1371

Aaron E. Rosenberg, S. Parthasarathy

AT&T Labs, USA

W1C.2 Speaker Verification Based on Phonetic Decision Making 1375

Jesper Olsen

Aalborg Univ., Denmark

W1C.3 Analysis and Comparison of Score Normalisation Methods for Text-Dependent Speaker Verification 1379

A.M. Ariyaeeinia, P. Sivakumaran

Univ. of Hertfordshire, UK

W1C.4 Automatic Speaker Recognition on a Vocoder

Link 1383

Frederic Jauquet, Patrick Verlinde, Claude Vloeberghs

Royal Military Academy, Belgium

W1C.5 Likelihood Ratio Adjustment for the Compensa-tion of Model Mismatch in Speaker Verification 1387

Frederic Bimbot, *Dominique Genoud

ENST/CNRS, France

*IDIAP, Switzerland

W1C.6 A Lognormal Tied Mixture Model of Pitch for Prosody-Based Speaker Recognition 1391

Kemal M. Sonmez, Larry Heck, Mitchel Weintraub, Elizabeth Shriberg

SRI International, USA

SESSION: W1D

Speech Enhancement I

Chair: Hynek Hermansky, Oregon Graduate Inst. of Science and Tech., USA

W1D.1 Residual Noise Suppression Using Psychoacoustic Criteria 1395

Tim Haulick, Klaus Linhard, Peter Schrogmeier

Daimler Benz AG, Germany

W1D.2 Processing Linear Prediction Residual for Speech Enhancement 1399

*B. Yegnanarayana, Carlos Avendano, Hynek Hermansky, *P.Satyanarayana Murthy

Oregon Graduate Institute of Science and Technology, USA

*ITT MADRAS, India

W1D.3 Combined Acoustic Echo Control and Noise Reduction for Mobile Communications 1403

Stefan Gustafsson, Rainer Martin

Aachen Univ. of Technology, Germany

W1D.4 A Nonstationary Autoregressive HMM and its Application to Speech Enhancement 1407

Ki Yong Lee, Yeol Rheem

Changwon National Univ., Korea