> Did anyone else experenced that problem?

Is there a problem?

> I would be very greatful if somebody would know the solution.

Map them to existing phonemes if you don't want to retrain the acoustic model. Train new acoustic model if you still want to use new phones.

Phonems that are not in "sl_speech_dutch.phone" file like "nn", "yy", "ss" are not making any problems. Unlike that phonem "er" was a problem because all the word that had that phonem could not be recognised.

Where can I find a list of all phonems that are used in this acoustic model because it seems that those in "sl_speech_dutch.phone" file are not all?

Thank you for your answers!

> Where can I find a list of all phonems that are used in this acoustic model because it seems that those in "sl_speech_dutch.phone" file are not all?

this file is exactly the list of phones used

> Unlike that phonem "er" was a problem because all the word that had that phonem could not be recognised.

You have a misconception that poor recognition can be caused by missing phone. It's not like that. The model can be unsufficiently trained for various reasons. This phone can be underrepresented or stripped during alignment, there could be other issues.

Thank you verry much on your answer.

I understand what you are saying, but what about truly missing phones like "nn", "ss, "yy" that are not in "sl_speech_dutch.phone" but the words with those phones in dictionary are recognized?

Hello

I was wrong, indeed it seems that phone file is not in sync with the model. The real list of phones could be found in the mdef file in the model folder. All phones including nn and er are there.

 I've updated the archive, it should be up-to-date now

https://sourceforge.net/projects/cmusphinx/files/Acoustic and Language Models/Dutch Voxforge/voxforge-nl-0.1.tar.gz

Thank you verry much on your answers! They have been verry useful.