>I would like to produce an Acoustic Model for French, but I would like to have all voices.

There is backlog of submissions - these have not been processed (i.e. downsampled or vetted for spam submissions) if you want them, I can give you a link, but you'll have to do the processing yourself...

>Also how to only get the French voices from a specific region (Quebec, France)? Maybe there's a database with this information?

Check the readme file for each submission - you should be able to create a script to automate this.

Usually, for a quick list, you would ordinarily go to the listen page for your language and search for what you are looking for - giving you a listing of all those submissions containing the search term... but nothing seems to come up for 'Québécois'.  The accents might be giving the search routine some problems...

Note there only about 9 hours of French processed speech.

--- (Edited on 2/8/2012 4:50 pm [GMT-0500] by kmaclean) ---

Hi,

That would be nice, I can produce a script (python, shell or whatever) to separate (by region, gender) & process them. So I guess I will end up with more than 9 hours.

If you can send me an access to download all the french submissions (processed & not processed), that would be nice.

If my work (scripted or manually) can be of any use... let me know.

Thanks

--- (Edited on 2/8/2012 4:45 pm [GMT-0600] by psc) ---

> an access to download all the french submissions (processed & not processed),

here is a copy of the unprocessed French submissons: rawSpeechSubmissions

>If my work (scripted or manually) can be of any use... let me know.

Please post your results here (or provide a link to your blog/website).  If there is a script, you can post it here or send it to me via email and I can set it up as an attachment to this thread.

thanks,

Ken

--- (Edited on 2/9/2012 7:13 pm [GMT-0500] by kmaclean) ---