Re: about RHVoice

[Samuel Thibault]

Hello,

Alexander Epaneshnikov, le lun. 08 févr. 2021 17:38:48 +0300, a ecrit:
> 07.02.2021 20:24, Samuel Thibault пишет:
> > илья пащук, le dim. 07 févr. 2021 20:10:55 +0300, a ecrit:
> > > control file auto generation theamed a good idea in the beginning, but if 
> > > we
> > > need to mention every voice in debian/copyright file it may be better to
> > > remove auto generation to prevent non-free voices to appear in the free
> > > repos.
> > If the copyright terms are different for each voice it may be better to
> > handle them by hand indeed.
> > 
> > We will also need to know exactly the list of files which have usage
> > restrictions, to remove them from the debian main package, and only keep
> > them in the non-free package.
> 
> hello Samuel and all. RHVoice maintainer here. we are interested in
> including RHVoice in the debian repository.
> can we help as a upstream in this effort?

One thing I was unsure when I had a look at the repository some time
ago: are natia and talgat the only data/voices/ directories that have
usage restrictions?

About the third-party directory, it would be useful to have a way to
tell cmake / scons to just link against the system-provided versions of
them instead of building them again. Notably, libsonic, rapidxml and
tclap are already packaged in Debian, and we do not want to maintain
both the main packages and a copy within rhvoice.

Now, there is the data: I see .fst files, .dt files, .lts files, .fsm
files, .pdf files, .inf files, .data files. How are these produced? To
be free, we need the tools that allow to edit/compile them (or re-record
them if these are produced from recording human voice, for instance).

Samuel