Hi,

The Voxforge models packaged in Ubuntu are those from 0.1.1~build726 which dates back to 2007, so I'm wondering, should I update the package to a new daily build, or is there some reason why the old one is better?

If the answer is to keep it, do you plan to release a new version anytime soon?

Thanks

--- (Edited on 6/5/2009 12:54 pm [GMT-0500] by RainCT) ---

Hi RainCT,

>should I update the package to a new daily build, or is there some

>reason why the old one is better?

The nightly builds include some submissions with problems in their  transcriptions - I have not had the chance to correct these yet. 

The old release is only 'better' in the sense that it has been tested. 

The nightly builds should work, but there are no guarantees (not that there was any guarantee with the old release...).  Give it a try, and let us know if it is OK.  Theoretically it should be better because there is much more speech audio, but the problem submissions might negate any advantage of the larger amount of speech audio used to train the acoustic model.

>If the answer is to keep it, do you plan to release a new version

>anytime soon?

Depends on what you mean by soon.  If you mean in the next month or so, no. 

A new version should be out in the next 3-4 months.

Ken

--- (Edited on 6/10/2009 1:23 pm [GMT-0400] by kmaclean) ---

I've been updating the package in Ubuntu Karmic and now it has the daily build from 26/08/2009; I won't be changing it anymore as we are now in Feature Freeze. In case you can get a new "stable" release out this month (or early October) I'll request a freeze exception for it and try to get it updated to it.

Cheers :)

--- (Edited on 9/1/2009 9:44 am [GMT-0500] by RainCT) ---