The main problem for us is a recordings anyhow, the rest can be created rather easily. If you'd like to help, submit recordings of your voice and voices of your friend and relatives. Once you'll have more than two hours of subscribed speech we can think about model building.

Bonjour Olivier,

All contributions are welcome!

I created a new forum where you can submit French speech.  You'll need to create your own prompts, and record audio as shown in the SubmitSpeech section.

Once we have accumulated enough speech (a couple of hours is our current threshold), we can look at creating acoustic models. 

If you can find an existing French Speech Corpus this would help speed things up.  We have some links to resources for other languages on the Dev Wiki (feel free to update this page if you like).

thanks for your interest in VoxForge!

Ken