[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Module for Kali TTS
Module for Kali TTS
Fri, 18 Nov 2016 10:32:45 +1100
Hi Rapha?l, thanks for your mail, my reply is inline.
On Fri, Nov 04, 2016 at 05:05:40AM AEDT, Rapha?l POITEVIN wrote:
> Dear team,
> I work for Hypra, a french company which develops a Universal Access System.
> I wrote a module for Kali, a french and english TTS:
> Unfortunately there is no english page about it.
> Kali is a non free software, but it's prety good quality for a french
> listener and we want to integrate it into our system.
I also think the English voices are acceptable, although I certainly think
there are better options, but French is the focus, which is fine.
> This module should be modified later because for the moment, Kali writes a
> wave file which is read by module_play_file(). A kind of kali_text_to_wave,
> like flite, has been recently implemented but I didn't have the time to
> rewrite the module.
Ok, having quickly read this email previously, and now having gotten around
to patching my local master branch copy of Speech Dispatcher with your
supplied patches, I wonder whether we should put this into a WIP branch
first, so that we can properly fix up the above to use Speech Dispatcher's
audio subsystem before we release it.
> Kali is written in C++, therefore I had to modify some files to compile.
Yes, and it builds fine here, although this does produce a lot of compiler
warnings, and I think this is best addressed in the longer term by the
planned re-architecting of modules, which will mean that you will eventually
be able to maintain your own module independantly of the Speech Dispatcher
> Please find my differents patches:
> - ./configure.ac
> - src/modules/Makefile.am
> - src/modules/module_utils.h
> - config/modules/kali.conf
> - config/modules/Makefile.am
> And of course the cpp module.
Thanks. As I have indicated above I have all of this working now just fine. I
do wonder what version of Speech Dispatcher you were patching against. I had
to apply a few of the patches by hand since there are differences between git
master and what you were working with. I think I may have suggested this in a
private email to you, but it is easier to integrate git formatted patches
that apply against master, or if you feel you must, against the
speech-dispatcher-0-8 branch, which is for the 0.8 release series. You can
find the Speech Dispatcher git repository at http://git.freebsoft.org. The
primary reason is for attribution to you, the patch and code author. For this
time around I can work something out to properly credit you for the patches,
but I ask that you take this into consideration for the future. If you are
unsure as to how to use git, I am sure I, or others could suggest some good
resources on learning how to work with git.
> I hope that module could be added to the project. Don't hesitate to tell me
> if some parts of the code don't respect the general modules structure.
As per the above, given that the module functionality is incomplete, it may
be worth putting this code into a WIP branch, so we can clean it up and add
direct audio support, at which point we can then merge it into master, and
probably 0.8 as well, and of course support for more synthesizers is welcome.
I also maintain a repository on github that contains all WIP branches that
are currently being worked on. It is on github to more easily facilitate
collaboration with others. If you have a github account, you will be able to
fork the repository, make changes to a branch, and request a pull request, at
which point it can be reviewed, and merged. If not thats ok as well, just
sending patches to me, or pointing me to a git repository that you use
elsewhere is also fine.
I will get back to you once I have the kali code in a WIp branch and pushed.