In this article
A cloned voice solves real problems. It can add a line to a shipped game without a studio session, keep a long-running character sounding the same as the original actor ages, and localise into languages that were previously cut for budget. Those are genuine wins. The dispute is about the terms.
What performers are asking for
The core demands are consistent across the recent labour agreements: informed consent for any synthetic use, a separate negotiated fee for it, the right to refuse specific uses, and a time limit so a one-off session does not become a permanent licence. None of that is exotic. It mirrors how likeness rights work elsewhere.
Where studios are cautious
The friction is with catalogue work — older recordings made before synthetic use was contemplated, where the contract is silent. Reusing that material to train a model is legally untested and reputationally risky, and most large studios have quietly decided it is not worth it yet.
The likely shape of the answer
Opt-in synthetic voice, negotiated per project, with credit and a fee, is where the standard agreements are heading. Studios that get ahead of it will have an easier time hiring. Studios that try to route around it will spend the savings on the fallout.
