It should be like music licensing. Copyright Office sets a fee schedule and you can copy other artists' songs as long as you pay.
Since AI merges data together it'd probably have to be a spotify-like deal where everybody's voice that's in the model gets a check so unless you want the overhead of cutting a 1¢ check to a billion people the models will be specific to individuals. Or if you have it speak lines "as Jennifer Hale" then maybe she gets the majority of the money.
It should be like music licensing. Copyright Office sets a fee schedule and you can copy other artists' songs as long as you pay.
Since AI merges data together it'd probably have to be a spotify-like deal where everybody's voice that's in the model gets a check so unless you want the overhead of cutting a 1¢ check to a billion people the models will be specific to individuals. Or if you have it speak lines "as Jennifer Hale" then maybe she gets the majority of the money.
You'll have Synth like knobs to change the original voice into something not copyrightable