Cross-Language Experiment

Abstract

The contribution addresses the cross-language experiment. The aimwas to test the possibility of the conversion French phoneme modelsinto Czech ones. This model conversion uses the Hidden Markov Models(HMM) classification procedure. The first step consists of theiterative mapping of French models to Czech ones. The mapping is givenby the analysis the confusion matrix. The second step is the Baum-Welchre-estimation resulting in the final models for Czech language. Despiteof the differences between French and Czech languages the finalrecognition score reaches 64% for the phoneme recognition and 74% fordigit recognition. Relatively low recognition accuracy is caused by theinadequate noise model. The experiences gained with the cross-languageexperiment were utilized for the classification of simple human bodymovements. The solution of this problem and results are described inthe second part of this contribution under the title EEG SignalsClassification-Introduction to the Problem.