Please use this identifier to cite or link to this item: https://doi.org/10.21256/zhaw-23889
Publication type: Conference paper
Type of review: Editorial review
Title: ZHAW-CAI : ensemble method for Swiss German speech to Standard German text
Authors: Ulasik, Malgorzata Anna
Hürlimann, Manuela
Dubel, Bogumila
Kaufmann, Yves
Rudolf, Silas
Deriu, Jan Milan
Mlynchyk, Katsiaryna
Hutter, Hans-Peter
Cieliebak, Mark
et. al: No
DOI: 10.21256/zhaw-23889
Proceedings: Proceedings of the Swiss Text Analytics Conference 2021
Editors of the parent work: Benites de Azevedo e Souza, Fernando
Tuggener, Don
Hürlimann, Manuela
Cieliebak, Mark
Vogel, Manfred
Conference details: Swiss Text Analytics Conference – SwissText 2021, Online, 14-16 June 2021
Issue Date: 2021
Publisher / Ed. Institution: CEUR Workshop Proceedings
ISSN: 1613-0073
Language: English
Subjects: Natural language processing; Swiss German
Subject (DDC): 006: Special computer methods
410.285: Computational linguistics
Abstract: This paper presents the contribution of ZHAW-CAI to the Shared Task ”Swiss German Speech to Standard German Text” at the SwissText 2021 conference. Our approach combines three models based on the Fairseq, Jasper and Wav2vec architectures trained on multilingual, German and Swiss German data. We applied an ensembling algorithm on the predictions of the three models in order to retrieve the most reliable candidate out of the provided translations for each spoken utterance. With the ensembling output, we achieved a BLEU score of 39.39 on the private test set, which gave us the third place out of four contributors in the competition.
URI: http://ceur-ws.org/Vol-2957/sg_paper3.pdf
https://digitalcollection.zhaw.ch/handle/11475/23889
Fulltext version: Published version
License (according to publishing contract): CC BY 4.0: Attribution 4.0 International
Departement: Applied Linguistics
School of Engineering
Organisational Unit: Centre for Artificial Intelligence (CAI)
Institute of Language Competence (ILC)
Published as part of the ZHAW project: Interscriber: Turning Dialogues into Actionable Insights
Appears in collections:Publikationen School of Engineering

Files in This Item:
File Description SizeFormat 
2021_Ulasik-etal_ZHAW-CAI-Ensemble-method-Swiss-Standard-German.pdf241.79 kBAdobe PDFThumbnail
View/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.