TCTS Lab Research Groups


[ FPMs > TCTS > ASR group > Projects > DEMOSTHENES ]





The DEMOSTHENES project has a duration of 12 months (09/98 - 09/99) and is funded by the Ministry of Technology, Research and Energy of the Region Wallonne.

See the Official web site for complete information about this project.



Acquiring a good command of spoken Dutch is a non-trivial task for most French speaking learners of the language. Traditional audio-visual aids -- in the classroom or the language laboratory -- have shown their limitations in correcting pronunciation (lack of systematic feed-back in a non-invidualized environment). On the other hand, current techniques in continuous speech processing make it possible to develop multimedia tools that analyse and correct the foreign language learner's pronunciation in a consistent and individualized approach.

In this prospect, two Belgian research teams have joined their expertise in speech recognition and software development for foreign language learning to launch the DEMOSTHENES project.

DEMOSTHENES will result in a full-fledged multimedia courseware for Dutch pronunciation, which detects and corrects the typical errors made by French speaking learners. Thanks to the well-known hybrid HMM/ANN systems which aim at combining Hidden Markov Models and Artificial Neural Network, the final product will discriminate pronunciation errors at the phoneme level. Such a fine-grained approach distinguishes the DEMOSTHENES application from products currently available on the market, which provide feed-back in a graphic, non-linguistic format.