This is the GPL Dutch acoustic model trained from IFA corpus:

http://www.fon.hum.uva.nl/IFA-SpokenLanguageCorpora/IFAcorpus/

You need one of the decoders from sphinx family to use it:

http://cmusphinx.sourceforge.net

For a usage demo, look into the script in the test subdirectory.
It requires only sphinx2 and demonstrates continuous Dutch digit
recognition.

To rebuild model from new voice data:

0. Download and convert audio files from IFA corpus
1. update etc/sl_speech_dutch_train.fileids
2. update etc/sl_speech_dutch_train.transcription
3. update etc/celex.dic if required
4. setup sphinxtrain task and copy etc directory there
5. run ./scripts/make_feats.pl -ctl etc/sl_sphinx_dutch_train.fileids
6. run ./scripts/RunAll.pl

