Leverage existing community-oriented docker work?

[KathyReid]

Raised by @lissyx in https://github.com/JRMeyer/deepspeech-playbook/issues/15

https://github.com/Common-Voice/commonvoice-fr/blob/master/DeepSpeech/

This is work I have been conducting explicitely to factorize and simplify most of the pipeline to help community build their own models. To the best of my knowledge, several languages are building on top of this, including Italian, Kabyle and Spanish.

I think it would be wise instead of re-inventing the wheel to move this work out of that repo and into a new one, and use that for the Playbook.

[lissyx]

@KathyReid Now that you have the playbook under mozilla org, maybe we should either:

• fork the subdirectory to a new mozilla org repo
• push the docker work into the playbook repo?

[KathyReid]

Thanks for the reminder on this one.

I think we should keep the docker work separate to the PlayBook, but reference it. Suggest the following actions:

• Create a new repo under Mozilla called mozilla/deepspeech-fr and move the docker work here
• Then over time, if this is forked for other languages, and possibly modified (ie with new scorers or other language-specific NLP processing), those repos can be stored by language (ie deepspeech-kab for Kabyle), in a similar way that Common Voice is separated by language.
• The other reason I think it makes sense to separate them by language is that over time, there may be different language versions of the PlayBook (ie fr, de and so on) and then the language version of the PlayBook can reference the DeepSpeech docker work for that language.

[lissyx]

Create a new repo under Mozilla called mozilla/deepspeech-fr and move the docker work here

Right, but no -fr we don't want to make people think bad

Then over time, if this is forked for other languages, and possibly modified (ie with new scorers or other language-specific NLP processing), those repos can be stored by language (ie deepspeech-kab for Kabyle), in a similar way that Common Voice is separated by language.

I'd prefer people to contribute directly to the main repo with their own code

The other reason I think it makes sense to separate them by language is that over time, there may be different language versions of the PlayBook (ie fr, de and so on) and then the language version of the PlayBook can reference the DeepSpeech docker work for that language.

There will always be a lang-specific component at some point, I don't see this as a big deal.

[KathyReid]

This is being discussed in this Discourse thread
https://discourse.mozilla.org/t/thread-discussion-on-how-to-manage-repositories-on-github-for-deepspeech-work-in-other-languages/76518

I'm awaiting some clear direction here to know what the next steps are.