speech recognition for DSL-N needing some help!

Hi guys!

If someone could help me, as explained in the dsl4x forum topic "speech synthesis & recognition", i made a "fast prototyping" murgalua 0.5.5 application script which is fully operating in French (not much tested in US-English, but i'm "natively" French, so my US-English accent is horrible!) in a DSL-NOT powered system.
Other systems are fully functional like MS-Windows XP, (MS W2K is not compatible with pocketsphinx), Debian4 (recognition engine is CMU-Sphinx2, present in the default repositories). These OS are supporting MBrola voices for espeak speech synthesis output...

So, i ask for your help for these two problems:

- searching, including and testing a good US-English language model for "Alfreid.lua" script (both versions Sphinx2 & pocketphinx)

- enabling MBrola support for DSL-Not (i tested espeak with default "robotic" voices, but no way for making the system speak with mbrola voices, "piping" the output of espeak to mbrola, then to alsaplayer (aplay). aplay is not present in the default install of DSL-N, i could notice that default sound driver was OSS (not compatible to my knowledge with espeak)

Once done, backporting to DSL 4x (Sphinx2 is in the default woody repositories) should be interesting...
Take a look at my homesite, where you will find all needed info and scripts (updated this 26/02/2013)

http://michelterras.perso.sfr.fr/index.html?n=15

Script Alfreid.lua includes a diagnostic part which will help you finding what's missing for a fully usage (either data or binary, Linux or Windows)

Greetings from France to all of you DSL & DSL-N users!

Long live DSL and DSL-Not!