how to make grammar model in HTK for using with HVite

Speech Recognition Engines

Flat

User: harsh1kumar
Date: 5/12/2010 2:02 am

Views: 6575
Rating: 0

I have trained HMMs using HTK.

Now I want to test and get the results using the trained HMMs. The speech files I have can have anything spoken in them with no particular pattern.

I want to use HVite to recognize these files. For that I will need wdnet file. To get wdnet, I want a grammar file. So in this case where the spoken file can have anything in it, how can I make grammar model.

Thanks in advance.

--- (Edited on 5/12/2010 2:02 am [GMT-0500] by ) ---

Re: how to make grammar model in HTK for using with HVite

User: harsh1kumar
Date: 5/14/2010 1:05 am

Views: 114
Rating: 2

Please tell me if this can be done or not.

Is this possible that using HTK we can recognize phonemes in any random spoken sentences? These sentences are not subject to any grammar. That means anything can be spoken.

I don't care about the performance. I just want to know if it can be done. And if it can be done, then how??

--- (Edited on 5/14/2010 1:05 am [GMT-0500] by harsh1kumar) ---

Re: how to make grammar model in HTK for using with HVite

User: kmaclean
Date: 6/9/2010 9:11 pm

Views: 2201
Rating: 0

>And if it can be done, then how??

Yes.

You need a language model and an acoustic model. The HTK book describes how to create a language model.

What you are looking for, I think is to create a dictation application, rather than a command and control application (which is the focus of VoxForge).

--- (Edited on 6/9/2010 10:11 pm [GMT-0400] by kmaclean) ---

Previous • Next •


Username	Password