ISCA Archive Eurospeech 1997
ISCA Archive Eurospeech 1997

A study of multilingual speech recognition

Fuliang Weng, Harry Bratt, Leonardo Neumeyer, Andreas Stolcke

This paper describes our work in developing multilingual (Swedish and English) speech recognition systems in the ATIS domain. The acoustic component of the multilingual systems is realized through sharing Gaussian codebooks across Swedish and English allophones. The language model (LM) components are constructed by training a statistical bigram model, with a common backoff node, on bilingual texts, and by combining two monolingual LMs into a probabilistic finite state grammar. This system uses a single decoder for Swedish and English sentences, and is capable of recognizing sentences with words from both languages. Preliminary experiments show that sharing acoustic models across the two languages has not resulted in improved performance, while sharing a backoff node at the LM component provides flexibility and ease in recognizing bilingual sentences at the expense of a slight increase in word error rate in some cases. As a by-product, the bilingual decoder also achieves good performance on language identification (LID).

doi: 10.21437/Eurospeech.1997-138

Cite as: Weng, F., Bratt, H., Neumeyer, L., Stolcke, A. (1997) A study of multilingual speech recognition. Proc. 5th European Conference on Speech Communication and Technology (Eurospeech 1997), 359-362, doi: 10.21437/Eurospeech.1997-138

  author={Fuliang Weng and Harry Bratt and Leonardo Neumeyer and Andreas Stolcke},
  title={{A study of multilingual speech recognition}},
  booktitle={Proc. 5th European Conference on Speech Communication and Technology (Eurospeech 1997)},