[Sugar-devel] Regarding the Sugar Labs Text to Speech feature.

Gonzalo Odiard godiard at sugarlabs.org
Fri Oct 3 15:29:13 EDT 2014


Hi Nurendra,
Certainly there are many thinks we can do connected with text to speech.
We use the "espeak" [1] utility, included in the linux distributions,
and a gstreamer plugin to use it from python.
Text to speech is used in many activitieas, and also in Sugar itself,
you can select any text and press a icon in the frame or press a shortcut
and the text is played. You can see the code here [2]
We call that "Global text to speech" [3]
Right now, I want move part of that code to the toolkit (the library shared
by Sugar,
and the activities, to avoid code duplication) [4]

Where we can get help?
* Better voices: espeak is very robotic.There are other free voices with
better quality?
Can be used in devices with modest capabilities like the xo?

* I found a project on debian repository to compile svox-pico,
used in Android. The voices are very nice, but I don't know if is possible
distribute it.
You can compile it in Fedora doing:

git clone -b upstream+patches git://git.debian.org/collab-maint/svox.git
svox-pico
git checkout -f origin/debian-sid
cd pico
./autogen.sh
./configure
make
make install

And can create a wav file wih the text doing:

pico2wave   --wave=/tmp/test.wav "hello world"

To be able to use this, we would need package all this are rpms,
and create a gstreamer pluggin to use it.
Not so easy, but a nice project ;)

* Other project could be improve our text to speech feature to be able
to highlight word by word while are played, like in "sing along" videos [5]
At least in part was implemented on Read activity when reading text files,
but need implementation in PDF and epub files.
I tried implementing in epub files but was very slow and lost sync between
the audio and the view.

There more things to do if you are interested.
We found text to speech is a powerful tool. Kids enjoy simple activities
as Speak (or no so simple if look at the robot part), and we even use
TTS in our word processor as a way to allow kids to "debug writing"
as Claudia Urrea say.

If you are interested, let me know and I will try tto help.

Regards,

Gonzalo


[1] http://espeak.sourceforge.net/
[2]
https://github.com/sugarlabs/sugar/blob/master/src/jarabe/model/speech.py
[3] http://wiki.sugarlabs.org/go/Features/Global_Text_To_Speech
[4] http://wiki.sugarlabs.org/go/Features/Text_To_Speech_In_The_Toolkit
[5] https://www.youtube.com/watch?v=J1QeShToq-E

On Fri, Oct 3, 2014 at 3:34 PM, Nurendra Choudhary <
nurendrachoudhary31 at gmail.com> wrote:

> Hi,
> I am a Computational Linguistics student at IIIT hyderabad and can code in
> python, c,c++ and also work with various frameworks.
> I would like to work on the Text to Speech feature. Can you help me start?
>
> - Nurendra Choudhary
>
>


-- 
Gonzalo Odiard

SugarLabs - Software for children learning
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.sugarlabs.org/archive/sugar-devel/attachments/20141003/65d14747/attachment.html>


More information about the Sugar-devel mailing list