Building a training data-set of kids voices

These solutions are a good start but they are too expensive, time consuming, this would require a dedicated team just to handle this issue.

I prefer a solution that is simple, fast, effective, low maintenance. There is a solution, we just don’t see it yet.

Oh wow, very impressive. Killer phrases at its best. Are these proposals too diverse ???

And that is a growing part of “discussion culture” here on discourse and also sometimes on github:
Oh no, low resources. …
Oh no, low.bandwith…
Oh i do not know what the guys from (insert the actual sponsor here) …
Oh no, this is not possible, because this means major changes…
But hey, thanks for the “discussion”.

It can be so easy in zero and one land…especially with so many contributers involved…from different parts of the world.
.
This works until courts are creating facts and not fiction.
FB also wildly declared in their terms what was easy for them, but did not mention some other things, but hey FB is for free!!!. One law student and a european court had another opinion on this. Resulting in paying big bucks.
“Digital Rights” to ring a bell. (and i do not mean the rights of FB)

Think of the children :innocent:
and
Where is the money???
Pathetic

Also when someone (cv/moz) start a project on this scale (worldwide) then cv/moz can handle it on a worldwide basis. The cheapest is not the best way to handle this on a worldwide scale. That is no rocket science to find this out. Playing sitting duck and pretending everything is well will NOT solve this and missing many (possible) recordings and/or finding ways to use child clips legally on an international basis. Common Voice is too important to fall for that.

General thoughts on getting kid/underage voices:

The basic idea is:
Starting a pilot project in selected school(s).
Locally in the school, online via internet to local cv server, both ways???)
(with official funding by the state/governnent???, funding by moz foundation???)

I learned from some reports that there are reading contests in the U.S. (reading bee, if i remember correctly).
Not the contest itself, but the time of preparation for these reading contests would be interesting for cv.
CC0 sentences are presented to the underage contributor from a local cv server in the school and saved on a local cv server in the school.
Cv/Moz is creating a form to fill out for the parents to give permission that their kids are allowed to contribute to cv.
If this pilot project is offically funded by state/government this filled out permissions stay in the school (teacher who is supervising, or school office)
If not officially funded, it is saved (encryted) in local cv server (proof of permission for cv) Two mature email accounts (parents) are required to activate the account for the underage contributor (child/teeny for contributing online at home)
In a later process a decision is required:
Release this as kids database only or include the underage clips to cv main corpus.

As stated in some posts in this thread before:
Kids voices are also natural. They also use voice assistants with (actual) mixed results. Avoiding (possible) problems is not the solution!

It is also very clear to me the efforts to start this pilot project within cv and state/government involved would be MASSIVE!

This proposal is in direction to the “legal team” and/or moz foundation.

This could be a way to save and release underage clips LEGALY!

Hello everybody,

Any advances on this? Or any official position by Mozilla Legal?

Thanks!