Do I use the way to adapt the default acoustic model to recognise the words not in VoxForge

[Toan Nguyen]

Dear cmusphinx team,

I would like to recognise the 2 words "Anfahrt" and "Streuen" in DE.

So I downloaded German acoustic model from VoxForge

http://www.voxforge.org/de/downloads

Of course, I already added the 2 words above to the phonetic dictionary and
build the new language model.

But my Android application does not yet recognise them.

Which method is best to solve my problem?

Adapting the default acoustic model

or

Training Acoustic Model For CMUSphinx

Yours sincerely, Toan

[Nickolay V. Shmyrev]

To get help on the accuracy you need to provide the recordings you are trying to recognize and the full setup you are using:

http://cmusphinx.sourceforge.net/wiki/faq#qwhy_my_accuracy_is_poor