ISCA Archive SLTU 2012
ISCA Archive SLTU 2012

Boosting under-resourced speech recognizers by exploiting out-of-language data - case study on Afrikaans

David Imseng, Hervé Bourlard, Philip N. Garner

Under-resourced speech recognizers may benefit from data in languages other than the target language. In this paper, we boost the performance of an Afrikaans speech recognizer by using already available data from other languages. To successfully exploit available multilingual resources, we use posterior features, estimated by multilayer perceptrons that are trained on similar languages. For two different acoustic modeling techniques, Tandem and Kullback-Leibler divergence based HMMs, the proposed multilingual system yields more than 10% relative improvement compared to the corresponding monolingual systems only trained on Afrikaans.

Index Terms: Multilingual speech recognition, posterior features, under-resourced languages, Afrikaans


Cite as: Imseng, D., Bourlard, H., Garner, P.N. (2012) Boosting under-resourced speech recognizers by exploiting out-of-language data - case study on Afrikaans. Proc. 3rd Workshop on Spoken Language Technologies for Under-Resourced Languages (SLTU 2012), 60-67

@inproceedings{imseng12_sltu,
  author={David Imseng and Hervé Bourlard and Philip N. Garner},
  title={{Boosting under-resourced speech recognizers by exploiting out-of-language data - case study on Afrikaans}},
  year=2012,
  booktitle={Proc. 3rd Workshop on Spoken Language Technologies for Under-Resourced Languages (SLTU 2012)},
  pages={60--67}
}