Hi,

I'm developing a speech recognition system using Julius. There I use a small acoustic model developed by me. It is consists of tamil characters in unicode. I could create this acoustic model without any problem, even it contains tamil characters in unicode.

When I used this acoustic model in Julius sample app it shows the result in some abnormal font(Which is unable to understand).

• Do Julius support unicode and can display result in tamil or,
• I have to rewrite all tamil words in english letters

--- (Edited on 3/4/2010 7:04 am [GMT-0600] by ) ---

>When I used this acoustic model in Julius sample app it shows the

>result in some abnormal font

does it work with HTK's HVite? 

>Do Julius support unicode and can display result in tamil or,

Julius should work with unicode since it was originally developed for Japanese speech recognition.

Ken

--- (Edited on 3/8/2010 11:25 pm [GMT-0500] by kmaclean) ---

Hello! I have a different approach for you. You could try to import your Tamil dictionary into simon (e.g. HTK format is supported). At least, I was able to import 60.000 Tamil words into simon (PLS format). It would be interesting to know whether you would be successful with this approach.

"rewrite all tamil words in english letters" - you can try that. Encoding issues can be pretty difficult to solve. With English letters, you don't have this problem.

--- (Edited on 2010-03-20 2:56 pm [GMT-0500] by ralfherzog) ---