Spoken language identification based on GMM models
Adam Dustor, Paweł Szwarc · ICSES 2010 International Conference on Signals and Electronic Circuits · 2010
The paper describes application of gaussian mixture models GMM to the task of spoken language identification. The influence of the length of the test utterances on identification error rate was examined. During identification procedure recordings for 15 languages were used, both European and Asian ones. As a language model GMM with full covariance matrix was applied. Obtained results of identification error rate were discussed.