rezaee - 2017-12-11

Hi there
I tried the default continous.c program, but it seems it's so far from our accent?! or maybe it has very bad language model!

When I test this words I get this results:

me:hi
pocketsphinx: i
m: how old are you
p: hold on you
m:where do you live
p: good you leave/ where the you/ where do you/ when you leave

So, if I want to get better results, which way is better?

  1. creating a ristricted dictionary and buld a new language model for that(although I don' know if I have not the acoustig model, cam I do that or not?)

  2. trying to recording middle eastern people voice when they try to speak english?

  3. changing the default acoustic model and language model to other ready to use package? (But which one?)

the second option I think it's the best but it's more defficult and I have no ability to doing that right now! Are there any other good option?