[Subject Prev][Subject Next][Thread Prev][Thread Next][Date Index][Thread Index]

[hts-users:02969] Re: About tied-clustering questions files


Hi
Thanks a lot for your kind answer. 

Regards,
Guillaume Galou

On Mon, 2011-07-25 at 09:31 +0900, Keiichiro Oura wrote:
> Hi,
> 
> Utterance-level questions (j1-j3) are defined in questions_utt_qst001.hed.
> Please, read lab_format.pdf for details.
> These questions are used for context-clustering of global variance.
> 
> Regards,
> Keiichiro Oura
> 
> 
> 2011/7/20 Guillaume Galou <guillaume.galou@xxxxxxxxx>:
> > Hi all,
> >
> > I successfully ran the arctic and brazilian demos in order to understand
> > the full process. Now, my aim is to create a french voice.
> > Reading the /data directory inside each demo, I saw questions_qts001.hed
> > and questions_utt_qst001.hed. I understand both structure and aim of the
> > first file. My concern is about the second one.
> > May I have an explanation/reference in order to understand how to build
> > such files for my language (especially the second mentionned file)?
> > I read HTK book; not clear enough in my opinion on this point.
> >
> > Best regards,
> >
> > Guillaume
> >
> >
> >
> 


References
[hts-users:02877] Regarding creating HTS voice for Hindi, kulesh kumar
[hts-users:02964] About tied-clustering questions files, Guillaume Galou
[hts-users:02968] Re: About tied-clustering questions files, Keiichiro Oura