Initial experiments with Estonian speech recognition

Ragni, A. orcid.org/0000-0003-0634-4456 (2007) Initial experiments with Estonian speech recognition. In: Nivre, J., Kaalep, H.-J., Muischnek, K. and Koit, M., (eds.) Proceedings of the 16th Nordic Conference of Computational Linguistics (NODALIDA 2007). 16th Nordic Conference of Computational Linguistics (NODALIDA 2007), 25-26 May 2007, Tartu, Estonia. University of Tartu, Estonia , pp. 249-252. ISBN 9789985405147

Abstract

This paper presents a short description of work recently done at University of Tartu to construct a word–based speech recognition system. Simple bigram and trigram language models with cross–word triphone acoustic models are used by a one–pass best hypothesis recognizer to perform decoding of test data. The lowest word error rate of 37.5% reported in this paper is a common figure for word–based speech recognition of languages like Estonian.

Metadata

Authors/Creators:	Ragni, A. https://orcid.org/0000-0003-0634-4456
Copyright, Publisher and Additional Information:	© 2007 ACL. Reproduced under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 International License (https://creativecommons.org/licenses/by-nc-sa/3.0/).
Dates:	Published: 1 May 2007
Institution:	The University of Sheffield
Academic Units:	The University of Sheffield > Faculty of Engineering (Sheffield) > Department of Computer Science (Sheffield)
Depositing User:	Symplectic Sheffield
Date Deposited:	18 Nov 2019 15:55
Last Modified:	19 Nov 2019 07:30
Published Version:	https://www.aclweb.org/anthology/W07-2437
Status:	Published
Publisher:	University of Tartu, Estonia
Refereed:	Yes
Related URLs:	Publisher

Download

Published Version

Filename: W07-2437.pdf

Licence: CC-BY-NC-SA 3.0

CLICK TO DOWNLOAD

CORE (COnnecting REpositories)

Initial experiments with Estonian speech recognition

Abstract

Metadata

Download

Published Version

Export

Statistics