Acoustic Model Discussions

Flat
alternative pronountiations in mlf file?
User: igor_odriozola
Date: 4/1/2011 10:39 am
Views: 4799
Rating: 3

Hi everybody!

I am trying to obtain HMMs from some recordings of VoxForge, using HTK. I am creating the phonene-mlf file, and I would like to know whether there is any way to take into account the alternative pronunciations of the dict file when writing the mlf file, or is it mandatory to use only one transcription for each audio file?


Thank you,

Igor

--- (Edited on 4/1/2011 10:39 am [GMT-0500] by igor_odriozola) ---

--- (Edited on 4/1/2011 11:12 am [GMT-0500] by igor_odriozola) ---

--- (Edited on 4/1/2011 11:13 am [GMT-0500] by igor_odriozola) ---

Re: alternative pronountiations in mlf file?
User: kmaclean
Date: 4/11/2011 8:28 am
Views: 2140
Rating: 3

>take into account the alternative pronunciations of the dict

>file when writing the mlf file

I don't think so.  

When you get to Step 8 - Realigning the Training Data, you will use HVite to match you recorded training data to the best pronunciation in yor pronunciation dictioanary. 

 

--- (Edited on 4/11/2011 9:28 am [GMT-0400] by kmaclean) ---

PreviousNext