The Part of the Voice That Says No

I used to lie in bed after a shift and calculate how much of me the job would need the following night. Legs, certainly. Hands. Enough of a brain to put the letters where they belonged. The rest of me could stay home, if only I could work out how to leave it there. I never managed the separation. Even my bad mood had to report for duty.

So I understand the attraction of a copy. Send the useful bastard. Let him have the fluorescent lights and the supervisor. I would keep the bed.

I had the payment worked out too. It came to me. So did the decision about where the copy went tomorrow, and whether he went anywhere at all. I had invented a labor-saving device, not resigned from my own life.

Now I look at what happened around Kenjiro Tsuda’s voice and find my lovely arrangement running with the payment details changed.

Tsuda is the Japanese actor who voices Kento Nanami in Jujutsu Kaisen. He said an anonymous TikTok account was using an AI copy of his voice to narrate videos about urban legends, the occult, and conspiracy theories. There were 188 videos. TikTok disputed the resemblance, calling the sound a “generic male voice” and the perceived similarity subjective.

On Wednesday, a Tokyo district court recognized that a performer’s voice can be protected by publicity rights—the right to control commercial use of a person’s identity—when used exclusively for its commercial appeal. The judge compared the voice to a portrait as a symbol of personality. She rejected the request to make TikTok remove the videos because the uploader had already deleted them.

I haven’t heard the disputed recordings and cannot settle the resemblance myself. TikTok says it will study the judgment before deciding its next steps, and keep working to balance technology development with rights protection.

What gets under my skin is not the possibility that a machine can sound convincing. I can grant the machine a beautiful performance. Give it the timing, the breath, the little hesitation that makes a listener lean closer. Let it pass every test of the ear. There is still a part missing which an audience may never hear: the performer deciding not to take the job.

I have turned down work for noble reasons and for reasons that would embarrass a sick dog. Too early. Too far. Didn’t like the man asking. Wanted to sit somewhere doing nothing until the money ran out. A right that applies only when I have a magnificent explanation is not much use to the person I have actually been.

When somebody offers to buy a performance, the offer contains a small admission. The person making it cannot have the performance yet. He needs another person to agree. All the flattering words about talent eventually have to squeeze through that unpleasant little doorway.

A copied voice offers a way around the conversation. Not in every use; I can imagine an actor willingly licensing a synthetic version and being delighted not to travel to a recording booth. I would not drag him there by the collar in defense of authentic suffering. I wanted to stay in bed myself. But the arrangement changes completely when the copy is useful because the person is recognizable and convenient because he need not be consulted.

Then the business hasn’t merely found a cheaper voice. It has found a version of a worker without the troublesome interval in which he considers the offer.

That phrase, generic male voice, stays with me. Spend years becoming recognizable. Then prepare to explain why you aren’t interchangeable after all. I can hear a whole second career in the explaining.

Yuko Sasaki of the Japan Actors Union calls a voice the result of years of training and apprenticeship. I believe her. I also dislike the idea that a person must first establish how hard he worked before his refusal deserves attention. If the voice came easily, would that make it available? If the actor enjoyed the training, do we reduce the amount of permission required?

I have leaned on the suffering argument myself. The writer paid for those sentences with his miserable life. The singer earned that note through a thousand lousy evenings. Fine. But I don’t want to invoice the misery before I can say who gets to speak for me. The cheerful bastard ought to be allowed to refuse too.

The court was deciding a question about commercial appeal. I am also thinking of the voice nobody would pay to hear. Mine, on a bad morning. I don’t own every similar growl, and I can survive an impression. But making a recognizable imitation perform a new script is not the same thing as getting me to say it.

That distinction can survive a flawless imitation. In fact, the better the imitation, the more I need it.

Tsuda’s lawyers described the videos’ content as sordid. Their objection is specific. Mine would survive a change of script. Make the copied voice announce a charity concert. Make it praise something admirable. I still want the question about agreement answered before the goodness of the cause is offered as payment.

In my old fantasy I could always call the thing home. I had taken that for granted, the way I took for granted that the bed would still be mine when I got into it. I never imagined waking to discover that somebody else had sent my duplicate out to explain a position I despised, and that the explanation was getting excellent reviews.

The ugliest part would not be hearing him fail to sound like me. It would be hearing him succeed.

I could refuse the next assignment. I could refuse all of them. Somewhere the work would continue, with my familiar impatience, my pauses, even a convincing hint that I would rather be somewhere else. They could put that in the performance too. A man apparently sick of the job, doing it forever.

Additional reporting: AI tool that copied actor’s ‘lustrous’ voice violated his rights, Tokyo court rules

Source: Tokyo court rules the human voice is protected, in a TikTok AI clone case