Language Model Representations For Beam-Search Decoding

Language Model Representations For Beam-Search Decoding (1995)

Cached

Download Links

[poseidon.itc.it:7117]

by Giuliano Antoniol , Fabio Brugnara , Mauro Cettolo , Marcello Federico

Venue:	In Proceedings of the ICASSP'95
Citations:	17 - 1 self

Versions

Version History

Metadata Version 1

Datum	Value	Source
TITLE	Language Model Representations for Beam-Search Decoding	INFERENCE
AUTHOR NAME	Giuliano Antoniol	SVM HeaderParse 0.1
AUTHOR AFFIL	IRST-Istituto per la Ricerca Scientifica e Tecnologica	SVM HeaderParse 0.2
AUTHOR ADDR	I-38050 Povo (Trento), Italy	SVM HeaderParse 0.1
AUTHOR NAME	Fabio Brugnara	SVM HeaderParse 0.1
AUTHOR AFFIL	IRST-Istituto per la Ricerca Scientifica e Tecnologica	SVM HeaderParse 0.2
AUTHOR ADDR	I-38050 Povo (Trento), Italy	SVM HeaderParse 0.1
AUTHOR NAME	Mauro Cettolo	SVM HeaderParse 0.1
AUTHOR AFFIL	IRST-Istituto per la Ricerca Scientifica e Tecnologica	SVM HeaderParse 0.2
AUTHOR ADDR	I-38050 Povo (Trento), Italy	SVM HeaderParse 0.1
AUTHOR NAME	Marcello Federico	SVM HeaderParse 0.1
AUTHOR AFFIL	IRST-Istituto per la Ricerca Scientifica e Tecnologica	SVM HeaderParse 0.2
AUTHOR ADDR	I-38050 Povo (Trento), Italy	SVM HeaderParse 0.1
ABSTRACT	This paper presents an efficient way of representing a bigram language model for a beam-search based, continuous speech, large vocabulary HMM recognizer. The tree-based topology considered takes advantage of a factorization of the bigram probability derived from the bigram interpolation scheme, and of a tree organization of all the words that can follow a given one. Moreover, an optimization algorithm is used to considerably reduce the space requirements of the language model. Experimental results are provided for two 10,000-word dictation tasks: radiological reporting (perplexity 27) and newspaper dictation (perplexity 120). In the former domain 93% word accuracy is achieved with real-time response and 23 Mb process space. In the newspaper dictation domain, 88.1% word accuracy is achieved with 1:41 real-time response and 38 Mb process space. All recognition tests were performed on an HP-735 workstation. 1. INTRODUCTION Many current ASR systems generate initial hypotheses through a b...	user correction - Legacy Corrections
YEAR	1995	INFERENCE
VENUE	In Proceedings of the ICASSP'95	INFERENCE
VENUE TYPE	CONFERENCE	INFERENCE
PAGES	588--591	INFERENCE
VOLUME	1	INFERENCE
CITATIONS	9 found	ParsCit 1.0