Hi,
I have just committed changes to address your comments and I have uploaded
the package again to mentors:
Again, you can fetch the package using:
dget -x
https://mentors.debian.net/debian/pool/main/f/festvox-ca-ona-hts/festvox-ca-ona-hts_1.3-1.dsc
Further info is available here:
https://mentors.debian.net/package/festvox-ca-ona-hts
Comments to your review follow here:
2016-09-05 23:32 GMT+02:00 Guillem Jover <guil...@debian.org>:
Sure! Here's the review, please:
>
> * Document changes for Vcs-* fields.
* Document dependency version bumps and additions.
>
* Minor stylistic nit, add a space between the >= and the version. :)
>
All these three have been fixed.
> * The copyright years and holders might need an update?
>
I have updated the copyright years and holders for the files that have
changed among versions.
> * I have no clue how these voices work internally so you'll have
> to give me a hand here. What's the source for upc_ca_ona.htsvoice?
> How do you generate that? Do we have the tools to do so in Debian?
> Otherwise we might need to put this in contrib. :/
>
I would understand moving the package to contrib if that is where it
belongs.
There is a summarised description of the process of generating the htsvoice
file in the README.source file. Basically, the htsvoice file contains the
voice model (so no "code", just "data").
To generate this model, we used around 3 weeks of computer processing power
and up to 5GB of RAM in some parts of the process. I don't think it makes
much sense to regenerate it in Debian.
Apart from the computer power, you will need the recordings with their
phonetic segmentation. For this model, they can be downloaded from
http://festcat.talp.cat/download/data, distributed with a CC-BY-SA license
(the recordings are almost 1GB large). You will also need some tools, all
but one of them are free. The one that is non-free is called HTK. It can be
downloaded gratis and is open-source but is non-free because it is not
redistributable. You can see the license terms here:
http://htk.eng.cam.ac.uk/docs/license.shtml.
Other packages (such as festlex-poslex) do provide trained models without
the training the data from scratch and are in main. I found a thread
https://lists.debian.org/debian-legal/2009/05/msg00028.html where a similar
case is discussed, but I did not see a final consensus on the issue. Also,
the ftp masters authorised the previous upload of this package.
I (as part of upstream) cannot improve the situation for the htsvoice
models as we don't have the resources to develop a fully free alternative
to HTK. Providing the raw data should allow anyone to train better models
and it is already more than what other packages have done.
> * The package is pretty much lintian clean, except for a pedantic
> tag about missing an upstream changelog, which we can ignore.
>
> I've built and tested these and they work fine, thanks!
>
Cool!
>
> > In the past it was a bit hard to find a sponsor and knowing that I had to
> > go through the process of finding one was demotivating me a bit.
>
> Please feel free to contact me anytime you need a sponsor, at least for
> any Catalan related package. In case I'm not responsive you could also
> try the debian-l10n-cata...@lists.debian.org mailing list, or just Cc
> it when you mail me. :)
Thanks, I will do that for sure!
Sergio
