Paper Info
Title
Transcription Of Broadcast News - Some Recent Improvements To Ibm'S Lvcsr System
Abstract
This paper describes extensions and improvements to IBM's large vocabulary continuous speech recognition (LVCSR) system for transcription of broadcast news. The recognizer uses an additional 35 hours of training data over the one used in the 1996 Hub4 evaluation [7]. It includes a number of new features: optimal feature space for acoustic modeling (in training and/or testing), filler-word modeling, Bayesian Information Criterion (BIC) based segment clustering, an improved implementation of iterative MLLR and 4-gram language models. Results using the 1996 DARPA Hub4 evaluation data set are presented.
Year
DOI
Venue
1998
10.1109/ICASSP.1998.675411
PROCEEDINGS OF THE 1998 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING, VOLS 1-6
Keywords
Field
DocType
broadcasting,information theory,testing,speech recognition,training data,telephony,bayesian information criterion,speech synthesis,language model,feature space,grammars,bandwidth
Rule-based machine translation,Speech synthesis,Feature vector,IBM,Bayesian information criterion,Pattern recognition,Computer science,Speech recognition,Artificial intelligence,Cluster analysis,Vocabulary,Language model
Conference
ISSN
Citations 
PageRank 
1520-6149
6
3.27
References 
Authors
7
6
Name
Order
Citations
PageRank
L. Polymenakos1266.44
P. Olsen28510.62
D. Kanvesky3254.61
R. A. Gopinath441748.03
P. S. Gopalakrishnan55110.24
Stanley F. Chen61723219.64