Title. And this will also affect non-AI imitations of voices?
Copyrighting voices to defeat AI would achieve nothing. Modern AI’s can be overtrained to the point where they strongly resemble their training data, but that is a problem that will be fixed within the next 5 years, which is way earlier than we will see any legislation regarding AI if our geriatric government is anything to go by. After that, AI will generate its “own” content that will be legally protected as its intellectual property.
I base this off of the fact that every media from humans is inspired by previous media. Fundamentally, once it stops directly plagiarising, there is no legal distinction between what a human would be doing and what an AI would be doing, unless we want to come up with a legal classification of “human” that explicitly rejects AI.
That opens up a whole new can of worms though. If you define human to be having human dna, does that mean the 3 babies they have been genetically altered are not human? Or are they, because they comtain at least some human dna? Does that mean i can give my AI a vestigial organ and now its legally protected? Does it have to run off a brain? What is a brain? If i duplicate the neural connections in a brain with mosfets down to every single connection, that is indisputably a human intelligence running on possibly non human hardware depending on the word of law. Or is it a human intelligence? They would react the exact same way to their organic counterpart, down to having the same memories and emotions. Does their non-biological hardware preempt them from being human? Does a pacemaker? Or neuralink?
Theres a lot to be worked out here, but it seems to me to be much less problematic to target the people wanting to misuse AI rather than targetting the tool itself.
Ai models should contain a database of all the data it was trained on, where it came from, and through what license did it get perms to train on it. That way you don’t get the issue of comparing outputs to real life work, because the issue isn’t with the output, it’s with the use of the data in the first place without permission to train the AI model, and no it’s not similar to humans learning. The scale and accuracy of reproduction from a software is so much higher than a human that it’s not even a comparison, before we get to what it means to allow humans to grow as people by enriching themselves Vs what it means for a corporate owned churner.
All those dead people in the laugh tracks are gonna make bank
In a nutshell, voices are not eligible for copyright protection under USA law, whose hegemony results in most of the world conforming to the same. The principal idea for copyright is that it only protects the rendition of some work or act. A writer’s manuscript, an artist’s early sketches, a software engineer’s source code, and a vocalist’s audition recording, are all things that imbue their creator with a valid copyright, but only for that particular product of their efforts.
It is not permissible to copyright the idea of a space opera, nor a style of painting, nor an algorithm for a computer routine, nor one’s own voice. Basically, pure thoughts cannot be copyrighted, nor things which are insufficiently creative like a copyright on the number 42, nor natural traits or phenomenon.
If we did change the law to allow the copyright of a human voice, then any satire or mockery that involves doing a good impression of someone speaking would suddenly be a copyright violation. This is nuts, because it would also deny someone else who – by no fault of their own – happens to have an identical voice. Would they just not be allowed to speak ever? Although intellectual property rights stem from the USA Constitution, so too do First Amendment speech rights, and the direct collision of the two would have strange and unusual contours.
For when ideas can be protected by law, see patents. And for when voices can be protected, see soundmarks/trademarks and brand rights, the latter stemming from rights of association. Such protections generally only hold when the voice or sound in question is an artificial product, like the sound of Ronald McDonald, and the protection only limits direct competitors from using the voice or sound improperly; everyone else is free to do impressions if they want.
So for the titular questions, the hypothesis posed simply will not occur under current law, and it’s hard to see how it would be practical if the law did permit it.
satire or mockery
What about fair use?
whose hegemony results in most of the world conforming to the same
Does it really? I know there are significant differences between US and EU copyright law, fair use comes to mind
While you are correct about copyright on this subject, the more applicable topic here is Right of Publicity. It is state law in over half of US states, intended to protect the use of a person’s voice likeness.
Essentially, if an imitation voice is used in such a way that it could cause confusion about whether it is really the imitated person, then it is illegal to use it in any commercial context. I understand that the question here was about non-commercial contexts, but that line can get blurry when social media views can create followings that then translate into commercial success. I am not a lawyer by any means, I’ve just been researching this for my own AI voices applications and want to protect myself from accidentally imitating anyone.
For example, I need to be able to transform my voice into many other character voices, since I have so many lines to record it would be cost prohibitive to hire actors. The worst move would be to download a voice model of a known actor and use that directly. Very sketchy, both legally and ethically.
So, the next best move is to find three or four voice models and merge them into one with combined tensor data from all three. But I was still quite concerned about this, worried that in the many thousands of voice lines I make, some recognizable actor voices would slip through.
So, I came up with the following pattern that I feel much more comfortable with, both legally and ethically:
I download several voice models that have some quality in common - an accent, vocal timbre, or style of speaking. Then, I merge them to make a model that focuses on that trait. And I record myself saying a line with a lot of phoneme variety, trying to match the vocal trait as close as possible. Then, that merged vocal trait model is used to transform the recording of my voice into the new voice. Then, I use this transformed recording to train a new voice model. And I take a few of these generalized models (e.g. an accent, a tone, a speaking style) and use them to create the final character voice, which should in theory be far removed from any of the actors who contributed.
I’m not sure what OP’s use case is, if it’s truly non-commercial, this method might be overkill. But if anyone wants to try using AI voices in projects but is nervous about legal ramifications, this is one way to try to insulate created voices from the specific training data. YMMV.
Maybe? It depends on how the laws get written?
Yes.
It is hygge to read about Denmark doing good work in this area.
The proposal establishes legal definitions for unauthorized digital reproductions, specifically targeting “very realistic digital representation of a person, including their appearance and voice.”
Creative works such as parody and satirical content remain exempt from these restrictions.
Does this only affect digital reproductions? Because it’s sometimes easy to reproduce someone’s else voice especially if you are a voice actor. And what if you happen to have a voice almost identical to another person?
Does this only affect digital reproductions?
The article seems focused on the digital angle, but I’m not sure.
Because it’s sometimes easy to reproduce someone’s else voice especially if you are a voice actor.
Is that widespread? My impression is that misrepresentation through fake likeness has exploded from AI generated deepfakes?
And what if you happen to have a voice almost identical to another person?
Then you have the right to your voice and they have the right to their voice. If neither uses the likeness for misrepresentation or other nefarious means then nothing will come of it. If either do there are probably already possibilities within current legal frameworks to prosecute.
Is that widespread? My impression is that misrepresentation through fake likeness has exploded from AI generated deepfakes?
Voice imitations without AI are possible if you are skilled enough, but using AI is much easier because you can use recordings of people’s voices directly rather than trying to imitate the voices yourself.
