Efficient Search Using Posterior Phone Probability Estimates

Efficient Search Using Posterior Phone Probability Estimates (1995)

Cached

Download Links

[ftp.dcs.shef.ac.uk]

by Steve Renals , Mike Hochberg

Venue:	In Proc. ICASSP
Citations:	30 - 8 self

Versions

Version History

Metadata Version 1

Datum	Value	Source
TITLE	Efficient search using posterior phone probability estimates	INFERENCE
AUTHOR NAME	Steve Renals	user correction - Legacy Corrections
AUTHOR AFFIL	Department of Computer Science; University of Sheffield	user correction - Legacy Corrections
AUTHOR ADDR	Sheffield S1 4DP, England	user correction - Legacy Corrections
AUTHOR NAME	Mike Hochberg	user correction - Legacy Corrections
AUTHOR AFFIL	Department of Engineering; University of Cambridge	user correction - Legacy Corrections
AUTHOR ADDR	Cambridge CB2 1PZ, England	user correction - Legacy Corrections
ABSTRACT	In this paper we present a novel, efficient search strategy for large vocabulary continuous speech recognition (LVCSR). The search algorithm, based on stack decoding, uses posterior phone probability estimates to substantially increase its efficiency with minimal effect on accuracy. In particular, the search space is dramatically reduced by phone deactivation pruning where phones with a small local posterior probability are deactivated. This approach is particularly well-suited to hybrid connectionist/hidden Markov model systems because posterior phone probabilities are directly computed by the acoustic model. On large vocabulary tasks, using a trigram language model, this increased the search speed by an order of magnitude, with 2% or less relative search error. Results from a hybrid system are presented using the Wall Street Journal LVCSR database for a 20,000 word task using a backed-off trigram languagemodel. For this task, our single-pass decodertook around 15× realtime on an HP73...	user correction - Legacy Corrections
YEAR	1995	INFERENCE
VENUE	In Proc. ICASSP	INFERENCE
VENUE TYPE	CONFERENCE	INFERENCE
PAGES	596--599	INFERENCE
VOLUME	1	INFERENCE
CITATIONS	15 found	ParsCit 1.0