Adapting Slovak ASR for native Germans speaking Slovak

Authors:
Štefan Beňuš;Miloš Cerňak;Milan Rusko;Marián Trnka;Sachia Darjaa
Affiliations:
Institute of Informatics, Slovak Academy of Sciences, Bratislava, Slovakia, and Constantine the Philosopher University, Nitra, Slovakia;Institute of Informatics, Slovak Academy of Sciences, Bratislava, Slovakia;Institute of Informatics, Slovak Academy of Sciences, Bratislava, Slovakia;Institute of Informatics, Slovak Academy of Sciences, Bratislava, Slovakia;Institute of Informatics, Slovak Academy of Sciences, Bratislava, Slovakia
Venue:
DIALECTS '11 Proceedings of the First Workshop on Algorithms and Resources for Modelling of Dialects and Language Varieties
Year:
2011

Citing 3
Cited 1

Algorithms on strings, trees, and sequences: computer science and computational biology

Algorithms on strings, trees, and sequences: computer science and computational biology
Transcriber: Development and use of a tool for assisting speech corpora production

Speech Communication - Special issue on speech annotation and corpus tools
Acoustic model adaptation based on pronunciation variability analysis for non-native speech recognition

Speech Communication

Rule-Based morphological tagger for an inflectional language

COST'11 Proceedings of the 2011 international conference on Cognitive Behavioural Systems

Quantified Score

Hi-index	0.00

Visualization

Abstract

We explore variability involved in speech with a non-native accent. We first employ a combination of knowledge-based and data-driven approaches for the analysis of pronunciation variants between L1 (German) and target L2 (Slovak). Knowledge gained in this two-step process is then used in adapting acoustic models and the lexicon. We focus on modifications in the pronunciation dictionary and speech rate. Our results show that the recognition of German-accented Slovak is significantly improved with techniques modeling slow L2 speech, and that the adaptation of the pronunciation dictionary yields only insignificant gains.