8th International Conference on Spoken Language Processing

Jeju Island, Korea
October 4-8, 2004

Rapid Acoustic Model Development using Gaussian Mixture Clustering and Language Adaptation

Nikos Chatzichrisafis (1), Vasilios Digalakis (2), Vasilios Diakoloukas (2), Costas Harizakis (1)

(1) Dialogos Communications SA, Greece
(2) Technical University of Crete, Greece

This work presents techniques for improved cross-language transfer of speech recognition systems to new, previously undeveloped, languages. Such techniques are particularly useful for target languages where minimal amounts of training data are available. We describe a novel method to produce a language-independent system by combining acoustic models from a number of source languages. This intermediate language-independent acoustic model is used to bootstrap a target-language system by applying language adaptation. For our experiments we use acoustic models of seven source languages to develop a target Greek acoustic model. We show that our technique significantly outperforms a system trained from scratch when less than 8 hours of read speech is available.

Full Paper

Bibliographic reference.  Chatzichrisafis, Nikos / Digalakis, Vasilios / Diakoloukas, Vasilios / Harizakis, Costas (2004): "Rapid acoustic model development using Gaussian mixture clustering and language adaptation", In INTERSPEECH-2004, 17-20.