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

Re: Release of free software speech synthesizer framework / MBROLA replacement





On 24.03.2015 00:54, Samuel Thibault wrote:
Hello,

This can be interesting indeed.

Tobias Platen, le Sat 21 Mar 2015 11:54:31 +0100, a écrit :
an implementation of an MBROLA clone that I released under GPL v3.
How is it related to MBROLA?  Only the algorithm?  I see that there is
some code from the university of Mons, did they release that part as
GPLv3?  We have to make sure the copyright is properly respected.
MBROLA uses an Algorithm that is similar to Spectral Model Synthesis[1] to remove aperiodicity from a speech signal. Spectral Model Synthesis has also been used in other synthesizers such as VOCALOID. However the algorithms of WORLD are newer.
The code is from Mage/pHTS[2] which they released under the GPLv3.

[1]http://mtg.upf.edu/technologies/sms
[2]https://github.com/numediart/mage-legacy

Tobias Platen

I also prepared a Debian source package which will import in a git
repository soon.
It looks fine, except that libsekai0.dirs is not needed (and introduces
a spurious /usr/include directory), and the copyright should be more
precise: the files from M. Morise are actually BSD-3 (the copying.txt
file is actually confusing, perhaps this part of the source should be in
a separate directory, to make the various licences clear).

Thanks,
Samuel


Reply to: