Hello all,

I'm currently in the process of building a translator service using PocketSphinx. Right now I have a successful English -> Spanish (using festival) translator, and would like to encorporate Bidirectional speech.

I found the CMU mirror hosting the Spanish Language & Acoustic model:

http://www.speech.cs.cmu.edu/sphinx/models/hub4spanish_itesm/

And am having a bit of trouble setting it up (using Linux Fedora 12 2.6.32.11-99.fc12.i686): Where exactly do I place these files?

I placed the Acoustic Model, Language Model, and Model Definitions in:

• Acoustic Models (using 8g)
• Model Definition file (H4.2500.mdef)
• Language Models (H4.arpa.Z.DMP)

/usr/local/share/pocketsphinx/model/hmm/es/H4.cd_continuous_8gau/
I placed the

1. Dictionary
2. List of phonemes
3. Control file
4. Label file
5. Filler file

In a seperate directory (~/Sphinx/es).

Using the standard PocketSphinx execution:

pocketsphinx_continuous -hmm <DIR> -lm <LMFILE> -dict <DICTFILE>

What are the values I include for these parameters? Am I missing any files?

--- (Edited on 4/18/2010 4:18 pm [GMT-0500] by Ashlander) ---

pocketsphinx_continuous -hmm /usr/local/share/pocketsphinx/model/hmm/es/H4.cd_continuous_8gau -lm /usr/local/share/pocketsphinx/model/hmm/es/H4.cd_continuous_8gau/H4.arpa.Z.DMP -mdef /usr/local/share/pocketsphinx/model/hmm/es/H4.cd_continuous_8gau/H4.2500.mdef -dict ~/Sphinx/es/h4.dict -feat s3_1x39

--- (Edited on 4/21/2010 00:20 [GMT+0400] by nsh) ---