Audio in it of itself is not a bad idea, but having it standardized to AI voices and not allowing creators to offer this feature on their own for their own content is a bad idea. Should writers' content always be used on their dime to provide training data to companies that benefit from these experiments, including AI?