[sugar] An Update about Speech Synthesis for Sugar

[Edward Cherlin]

On Feb 18, 2008 6:22 AM, Hemant Goyal <goyal.hemant at gmail.com> wrote:
> Hi,
>
> It s great to see many other developers sharing the idea we have been trying
> to implement right within the Sugar Environment.

Yes, thanks to all.

> We have been working on integrating speech-synthesis into Sugar for quite
> some time now. You can check out our ideas here :
> http://wiki.laptop.org/go/Screen_Reader
>
> We are also documenting all our ideas and requirements with respect to
> Speech Synthesis in this Requirements Analysis Document here :
> http://www.nsitonline.in/hemant/stuff/Speech%20Synthesis%20on%20XO%20-%20Requirements%20Analysis%20v0.3.5.pdf
>
> It outlines some of our immediate as well as long term goals wrt
> speech-synthesis on the XO. Your ideas, comments and suggestions are
> welcome.
>
> I'd like to update the list about our progress:
>
> speech-dispatcher has been selected as a speech synthesis server which will
> accept all incoming speech synthesis requests from any sugar activity
> (example: Talk N Type, Speak etc)
> speech-dispatcher provides a very simple to use API and client specific
> configuration management.So whats causing the delays?

I have a few questions. Let's see what the InterWebs tell us.

* How many languages does speech-dispatcher support?

http://www.freebsoft.org/doc/speechd/speech-dispatcher_5.html#SEC8

SD works with Festival http://www.cstr.ed.ac.uk/projects/festival/
English, Czech, Italian, Spanish, Russian, Polish...

What is the mechanism for adding additional languages? Phoneset
recording, dictionary, and what?

http://www.freebsoft.org/doc/speechd/speech-dispatcher_23.html#SEC82
Develop new voices and language definitions for Festival: In the world
of Free Software, currently Festival is the most promising interface
for Text-to-Speech processing and speech synthesis. It's an extensible
and highly configurable platform for developing synthetic voices. If
there is a lack of synthetic voices or no voices at all for some
language, we believe the wisest solution is to try to develop a voice
in Festival. It's certainly not advisable to develop your own
synthesizer if the goal is producing a quality voice system in a
reasonable time. Festival developers provide nice documentation about
how to develop a voice and a lot of tools that help doing this. We
found that some language definitions can be constructed by
canibalizing the already existing definitions and can be tuned later.
As for the voice samples, one can temporarily use the MBROLA project
voices. But please note that, although they are downloadable for free
(as price), they are not Free Software and it would be wonderful if we
could replace them by Free Software alternatives as soon as possible.
See http://www.cstr.ed.ac.uk/projects/festival/.

Which in turn says:

Externally configurable language independent modules:

    * phonesets
    * lexicons
    * letter-to-sound rules
    * tokenizing
    * part of speech tagging
    * intonation and duration

That answers most of my technical questions, including how (in
principle, anyway) we are going to support tonal languages such as
Yoruba. Now for organization.

Where should we put TTS projects for language support? Can we create
http://dev.laptop.org/tts? Who should be in charge? What sort of
process should we have for creating projects? Should we just
automatically create a TTS project for every translate project?

> speech-dispatcher is not packaged as an RPM for Fedora,

I see Debian packages. Is there a converter?

>so at present I am
> mostly making a RPM package so that it can be accepted by the Fedora
> community and ultimately be dropped into the OLPC Builds. You can track the
> progress here : https://bugzilla.redhat.com/show_bug.cgi?id=432259 I am not
> an expert at RPM packaging and hence its taking some time at my end. I'd
> welcome anyone to assist me and help speed up the process.
> dotconf packages which speech-dispatcher is being packaged by my team mate
> Assim. You can check its progress here :
> https://bugzilla.redhat.com/show_bug.cgi?id=433253
>  Some immediate tasks that we plan to carry out once speech-dispatcher is
> packaged and dropped into the OLPC builds are :
>
> Provide the "much needed" play button, with text highlight features as
> discussed by Edward.

Thank you.

> Port an AI Chatbot to the XO and hack it enough to make it speak to the
> child :).
> Encourage other developers to make use of speech-synthesis to make their
> activities as lively and child friendly as possible :)
> Explore orca and other issues to make the XO more friendly for
> blind/low-vision students

Have you looked at Oralux, the Linux distro for the blind and visually-impaired?
http://oralux.net/
We should invite them to join our efforts.

> @James : We envision that speech-synthesis will surely get integrated with
> Read in due time. I think it would be great if maybe Gutenberg text could be
> loaded right from Read only?
>
> > I was not planning on anything so fancy.  Basically, I was frustrated
> > that I had a device that would be wonderfully suited to reading
> > Gutenberg etexts and no suitable program to do it with.  I have written
> > such an Activity and am putting the finishing touches on it.  As I see
> > it, the selling points of the Activity will be that it can display
> > etexts one page at a time in a readable proportional font and remember
> > what page you were on when you resume the activity.  The child can find
> > his book using the Gutenberg site, save the Zip file version to the
> > Journal, rename it, resume it, and start reading.  It will also be good
> > sample code for new Activity developers to look at, even children,
> > because it is easy to understand yet it does something that is actually
> > useful.  I have written another Activity which lets you browse through a
> > bunch of image files stored in a Zip file, and it also would be good
> > sample code for a new developer, as well as being useful.
>
> Warm Regards,
> Hemant
>
> _______________________________________________
> Sugar mailing list
> Sugar at lists.laptop.org
> http://lists.laptop.org/listinfo/sugar

-- 
Edward Cherlin
End Poverty at a Profit by teaching children business
http://www.EarthTreasury.org/
"The best way to predict the future is to invent it."--Alan Kay