Multilingual Informedia Project Partners:
Multilingual Informedia Project Partners Informedia (English)
CMU Informedia Group (Howard Wactlar, Alex Hauptmann, Ricky Houghton, et al.)
CMU Sphinx Group
Multilingual Speech Recognition
CMU/UKA Interactive Systems Labs - JanusRTk (Alex Waibel, Michael Finke, Petra Geutner, Peter Scheytt)
Translation/Cross Language Retrieval
CMU Language Technologies Institute (Jaime Carbonell, Eric Nyberg, Bob Frederking, Paul Kennedy, et al.)
DARPA N66001-97-D-8502 Delivery Order 0001 Introduction and Motivation
Slide3:
Introduction and Motivation
Serbo-Croatian BN Speech Performance:
Serbo-Croatian BN Speech Performance Broadcast News System
Vocabulary Growth Per Broadcast:
Vocabulary Growth Per Broadcast Broadcast News System
Language Normalization:
Language Normalization Transcription and Pronunciation Variants
One Single Form for Serbian/Croatian Variants
Unique Orthography for Foreign Words
Transcription Errors
Vocabulary: 48 K
OOV Rate: 7.9% (w/o Normal.: 10.2%)
Interpolated Language Models:
Higher Improvement Than w/o Normal. Broadcast News System
Hypothesis Driven Lexicon Adaptation:
Hypothesis Driven Lexicon Adaptation Broadcast News System
Information:
Information Publications and Presentation:
ISL: http://www.is.cs.cmu.edu
INFORMEDIA: http://www.informedia.cs.cmu.edu
LTI: http://www.lti.cs.cmu.edu
http://www.cs.cmu.edu/~scheytt
Email:
scheytt@cs.cmu.edu
pgeutner@ira.uka.de
finkem@cs.cmu.edu
waibel@cs.cmu.edu