This paper gives an overview of current understanding of acoustic-phonetic issues arising when trying to recognize speech from non-native speakers. Regional accents can be modeled by systematic shifts in pronunciation. These can often better be represented by multiple models, than by pronunciation variants in the dictionary. The problem of non-native speech is much more difficult because it is influenced both by native and spoken language, making a multi-model approach inappropriate. It is also characterized by a much higher speaker variability due to different levels of proficiency. A few language-pair specific rules describing the prototyical nativised pronunciation was found to be useful both in general speech recognition as in dedicated applications. However, due to the nature of the errors and the mappings, non-native speech recognition will remain inherently much harder. Moreover, the trend in speech recognition towards more detailed modeling is counterproductive for the recognition of non-natives.
Cite as: Compernolle, D.v. (1999) Speech recognition by goats, wolves, sheep and ... non-natives. Proc. Multi-Lingual Interoperability in Speech Technology, 1-7
@inproceedings{compernolle99_mist,
author={Dirk van Compernolle},
title={{Speech recognition by goats, wolves, sheep and ... non-natives}},
year=1999,
booktitle={Proc. Multi-Lingual Interoperability in Speech Technology},
pages={1--7}
}