Thanks Nickolay.
Now It became clear for me.But Now I have another question. I searched in hts mailing list and I found that utt2lab is used for converting utt into htk-style format but I've used "make" command to provide inputs for hts training and when gv,mgc,cmp,f0,labels,scp and list folders were built I used (usr/bin/perl scripts/training.pl scripts/config.pm) command to train.I found that for extracting monophone and full context labels from utt files,"dumpfeats" in festival/examples and then init.scm in festival and then ...are called.I can't find where and when is "utt2lab" used???
Best Regards A.A
--- On Fri, 12/11/09, Nickolay V. Shmyrev <nshmyrev@xxxxxxxxx> wrote:
From: Nickolay V.
Shmyrev <nshmyrev@xxxxxxxxx> Subject: Re: [hts-users:02305] question on dumpfeats To: hts-users@xxxxxxxxxxxxxxx Cc: "Angelina Anderson" <angelina_anderson@xxxxxxxxx> Date: Friday, December 11, 2009, 1:43 AM
В Срд, 09/12/2009 в 00:50 -0800, Angelina Anderson пишет: > Hi, > I know Festival's dumpfeats and utt2lab.sh are used to convert > festival-style utterance files into HTK-style labels.
Hello Angelina
> Now I want to know if all the utt information like "f0 131.315 ; pos 1.34664 ; > 130 id _119 ; f0 121.315 ; pos 1.33664 ; > 131 id _117 ; f0 133.147 ; pos 0.862565 ; " is necessary for converting into htk-style label formats??
No, it's not needed
> which information in utt is used for converting into htk-style label??
You can find the list of features used inside utt2lab script
> Does hts use duration
information in utt built in festival?
No
> IS hts training result dependent on which duration method in festival is used?
No
>
|