ACL-02 Workshop on
Speech-to-Speech Translation:
Algorithms and Systems
July 11 2002
University of Pennsylvania, Philadelphia
A workshop held as part of the
Association for Computational Linguistics 40th anniversary meeting
ACL-02 (http://www.acl02.org)
Hosted by The Computer and Information Science Department and the
Institute for Research in Cognitive Science
University of Pennsylvania, Philadelphia
July 7-12 2002
FINAL CALL FOR PAPERS
(Please notice the new date for paper submission)
DESCRIPTION:
Facilitation of speech communication across language barriers is a critical problem to solve for a global economy to thrive. Robust systems for speech-to-speech translation (S2S) are clearly necessary to move us towards achieving this goal. However, construction of such systems is clearly extremely complex, involving research in Automatic Speech Recognition (ASR), Text-to-Speech (TTS), Machine Translation (MT), Natural Language Understanding (NLU) and Generation (NLG). Although substantial progress in each of these components individually has been made over the last two decades, simply integrating individual ASR, NLU, MT, NLG, and TTS components to produce S2S systems is not sufficient to produce acceptable results. For example, conventional text-based MT systems have not been designed to cope with the imperfect syntax and transcription errors which characterize automatically transcribed conversational speech. Traditional speech recognizers (ASR component) and speech synthesizers (TTS component) have not been designed to recognize or synthesize speakers' emotional expressions which convey meanings and play an important role in the communications between human beings. Therefore, speech-to-speech translation raises a whole new set of algorithmic challenges over and above those associated with the individual underlying technologies themselves.
We would like to bring together various researchers in the field together to present the current state-of-the-art on speech-to-speech translation and discuss the challenges involved in building a functioning high performance system. We hope to hear about different approaches to the S2S realization and exchange ideas about the advantages and disadvantages of various approaches. The workshop will specifically focus on natural language processing problems which are unique and critical to producing robust speech-to-speech translation systems and components.
We solicit
submissions to the workshop in the following areas, however any other topic
related to the speech-to-speech translation is also acceptable:
Machine
Translation:
·
Algorithms for machine translation
applicable to S2S
·
Algorithms and systems for
application specific and limited domain machine translation
·
Rule-based MT, statistical MT,
template-based MT, interlingua-based MT
Speech
recognition and TTS:
·
Enhancing the performance of ASR in
S2S using natural language processing techniques
·
TTS modules that focus on conveying
meaning and emotions
·
Robust speech recognition algorithms
for S2S
·
Challenges for extracting and
conveying stress, prosody and emotions in speech across languages
NLP:
·
Natural language processing
algorithms for S2S
·
Natural language generation from
meaning representations
Language:
·
Challenges for speech-to-speech
translation across languages due to language characteristics, and suggestion of
solutions
·
Challenges for conveying stress,
prosody and emotions in speech across languages
System
architecture and software integration
·
Component architecture and design of
modular S2S systems
·
Portability of S2S systems to
different languages and domains
·
Implementation issues for robust and
limited resource S2S systems
Multilingual Data Collection and
System Evaluation:
·
Evaluation metrics of spoken
language translation quality
·
Language resources and knowledge
acquisition
SUBMISSIONS:
We invite paper submissions from all researchers in the area of S2S translation, natural language processing, linguistics, and all related topics. All submissions will be reviewed by an international program committee.
Submissions should follow the two-column format of ACL proceedings and should not exceed eight (8) pages, including references. We recommend the use of ACL LaTeX style files or Microsoft Word Style files available at http://www.cs.ualberta.ca/~lindek/acl02/style/. Papers in pdf format must be submitted electronically to: yuqing@us.ibm.com. Deadline for paper submissions has been postponed to April 5, 2002.
WORKSHOPS
REGISTRATION FEES:
The duration of the workshop is one full day. Only ACL-02 conference participants are allowed to register for the workshop. The registration fee is going to be set by the ACL-02 organizing committee. The Proceedings of the Workshop will be published by the ACL-02 organizing committee.
IMPORTANT
DATES:
April 5, 2002: Postponed deadline for workshop paper submissions
April 22,
2002: Notification of acceptance to authors
May 17,
2002: Deadline for camera-ready final version copies
July 11,
2002: S2S workshop in ACL-02 in Philadelphia
ORGANIZERS:
Yuqing Gao
(IBM T. J. Watson Research Center)
Alex
Waibel (Carnegie Mellon University & University of Karlsruhe)
PROGRAM
COMMITTEE:
Yuqing
Gao, Project Lead, Speech-to-Speech Translation Research, IBM T. J. Watson
Research Center
Hakan
Erdogan, Speech-to-Speech Translation Research, IBM T. J. Watson Research
Center
Alex
Waibel, Professor & Director, Interactive Systems Lab, Carnegie Mellon
University (USA) & University Karlsruhe (Germany)
Michael Picheny,
Manager, Speech Recognition Research, IBM T. J. Watson Research Center
Seiichi
Yamamoto, Director, ATR Spoken Language Translation Research Laboratories
(Japan)
Gianni
Lazzari, Vice Director of ITC-irst (Italy)
Taiyi
Huang, Professor, Institute of Automation, Chinese Academy of Sciences