[Sugar-devel] Gsoc proposal: Speech Synthesis

[James Simmons]

Chirag,

I still haven't run your code, but I did take a good look at it.  I 
expected it to look quite a bit like the code I and Aleksey Lim came up 
with for Read Etexts.  I was surprised to find that it didn't, but parts 
of it did look familiar to me because it looks like you're trying an 
approach that I tried and was forced to give up on.  What it looks like 
is you're sending the words to espeak one at a time, after highlighting 
them in the text viewer.  If that's what you're doing then you're 
launching espeak for each and every word, creating a .WAV file for that 
word, and then using aplay to play the word.  On a sufficiently powerful 
machine that works but sounds awful.  On an XO it doesn't work at all.

If this is what you are doing then have a look at the code for Read 
Etexts.  In that code I make a version of the text that has markup to 
indicate the beginning of each word.  Originally speech-dispatcher used 
that markup to do callbacks into my code, telling it which word to 
highlight.  Aleksey Lim wrote a gstreamer plugin for espeak that 
replaced speech-dispatcher but did the same thing.  His plugin works 
better and requires no configuration.

I didn't come up with this myself; someone from the speech-dispatcher 
mailing list suggested it.  It isn't perfect, but I'm pretty sure it 
works better than what you are attempting.

James Simmons


James Simmons wrote:
> Chirag,
>
> I won't be able to try out your code for awhile, but I did look at it 
> and noticed that while you refer to it as an "activity" it is not in 
> fact packaged as an "Activity".  Even if you intend for this code to 
> wind up being a part of Sugar itself, there is no reason you couldn't 
> make it an Activity now, and there would be advantages to doing that.  
> For one thing, it would be easier to try out.  The easier something is 
> to test, the more testing is done, and the better quality of testing 
> is done.  Plus the Activity could be used later by those unwilling to 
> update their XO's to the latest Sugar.
>
> Other than creating an SVG icon with Inkscape it wouldn't take much 
> work to make this a real Activity.
>
> James Simmons
>
>
>> Date: Thu, 11 Jun 2009 12:33:26 +0530
>> From: chirag jain <chiragjain1989 at gmail.com>
>> Subject: [Sugar-devel] Gsoc proposal: Speech Synthesis
>> To: sugar-devel <sugar-devel at lists.sugarlabs.org>
>> Message-ID:
>>     <e116096a0906110003mf9a841el3fe3da562f7b427b at mail.gmail.com>
>> Content-Type: text/plain; charset=ISO-8859-1
>>
>> Hi,
>>
>> I proposed for the speech-synthesis in Gsoc 09. My proposal can be 
>> viewed at :
>>
>> http://wiki.sugarlabs.org/go/speech-synthesis
>>
>> As a first phase of my development, I have implemented the speech and
>> karoke style coloring of the text. A basic speech configuration
>> manager has also been implemented to alter the volume, pitch and rate
>> of the speech.
>>
>> It would be great if you can test the activity.
>> Please download the speech-synthesis.zip from the link:
>>
>> http://code.google.com/p/speech-synthesis/downloads/list
>>
>> I have also included a detailed documentation of the activity.
>>
>> It would be great if you can send some feedbacks to me so that I can
>> improve upon his activity.
>>
>> Regards
>>   
>
>