TikTok Lawsuit Highlights How AI Is Screwing Over Voice Actors (vice.com) 93
An anonymous reader quotes a report from Motherboard: With only 30 minutes of audio, companies can now create a digital clone of your voice and make it say words you never said. Using machine learning, voice AI companies like VocaliD can create synthetic voices from a person's recorded speech -- adopting unique qualities like speaking rhythm, pronunciation of consonants and vowels, and intonation. For tech companies, the ability to generate any sentence with a realistic-sounding human voice is an exciting, cost-saving frontier. But for the voice actors whose recordings form the foundation of text-to-speech (TTS) voices, this technology threatens to disrupt their livelihoods, raising questions about fair compensation and human agency in the age of AI.
At the center of this reckoning is voice actress Bev Standing, who is suing TikTok after alleging the company used her voice for its text-to-speech feature without compensation or consent. This is not the first case like this; voice actress Susan Bennett discovered that audio she recorded for another company was repurposed to be the voice of Siri after Apple launched the feature in 2011. She was paid for the initial recording session but not for being Siri. Rallying behind Standing, voice actors donated to a GoFundMe that has raised nearly $7,000 towards her legal expenses and posted TikTok videos under the #StandingWithBev hashtag warning users about the feature. Standing's supporters say the TikTok lawsuit is not just about Standing's voice -- it's about the future of an entire industry attempting to adapt to new advancements in the field of machine learning.
Standing's case materializes some performers' worst fears about the control this technology gives companies over their voices. Her lawsuit claims TikTok did not pay or notify her to use her likeness for its text-to-speech feature, and that some videos using it voiced "foul and offensive language" causing "irreparable harm" to her reputation. Brands advertising on TikTok also had the text-to-speech voice at their disposal, meaning her voice could be used for explicitly commercial purposes. [...] Laws protecting individuals from unauthorized clones of their voices are also in their infancy. Standing's lawsuit invokes her right of publicity, which grants individuals the right to control commercial uses of their likeness, including their voice. In November 2020, New York became the first state to apply this right to digital replicas after years of advocacy from SAG-AFTRA, a performers' union. "We look to make sure that state rights of publicity are as strong as they can be, that any limitations on people being able to protect their image and voice are very narrowly drawn on first amendment lines," Jeffrey Bennett, a general counsel for SAG-AFTRA, told Motherboard. "We look at this as a potentially great right of publicity case for this voice professional whose voice is being used in a commercial manner without her consent."
At the center of this reckoning is voice actress Bev Standing, who is suing TikTok after alleging the company used her voice for its text-to-speech feature without compensation or consent. This is not the first case like this; voice actress Susan Bennett discovered that audio she recorded for another company was repurposed to be the voice of Siri after Apple launched the feature in 2011. She was paid for the initial recording session but not for being Siri. Rallying behind Standing, voice actors donated to a GoFundMe that has raised nearly $7,000 towards her legal expenses and posted TikTok videos under the #StandingWithBev hashtag warning users about the feature. Standing's supporters say the TikTok lawsuit is not just about Standing's voice -- it's about the future of an entire industry attempting to adapt to new advancements in the field of machine learning.
Standing's case materializes some performers' worst fears about the control this technology gives companies over their voices. Her lawsuit claims TikTok did not pay or notify her to use her likeness for its text-to-speech feature, and that some videos using it voiced "foul and offensive language" causing "irreparable harm" to her reputation. Brands advertising on TikTok also had the text-to-speech voice at their disposal, meaning her voice could be used for explicitly commercial purposes. [...] Laws protecting individuals from unauthorized clones of their voices are also in their infancy. Standing's lawsuit invokes her right of publicity, which grants individuals the right to control commercial uses of their likeness, including their voice. In November 2020, New York became the first state to apply this right to digital replicas after years of advocacy from SAG-AFTRA, a performers' union. "We look to make sure that state rights of publicity are as strong as they can be, that any limitations on people being able to protect their image and voice are very narrowly drawn on first amendment lines," Jeffrey Bennett, a general counsel for SAG-AFTRA, told Motherboard. "We look at this as a potentially great right of publicity case for this voice professional whose voice is being used in a commercial manner without her consent."
Re: (Score:2)
Funny how it's still meatbags cleaning toilets.
Re: (Score:3, Funny)
That's because it's meatbags that need toilets. Robots have shiny metal asses.
Re: (Score:1)
Say what?? (Score:2)
"We look at this as a potentially great right of publicity case for this voice professional whose voice is being used in a commercial manner without her consent."
Just think of how different the world would have been if we had this technology during the previous administration.
Re: (Score:2)
That depends a lot on what jurisdiction you're in. Around here, I couldn't even take a picture where you're in it and publish it without your consent, and I would expect this to be similar with voices. Interesting question, though.
Re: (Score:2)
But if they paid you for your picture, they could publish it as they probably got consent at the time of payment.
That said, there are (or should be) limits on how much they can alter or doctor it (putting your face in a racist rally or on a naked person shouldn't be kosher.)
Re: (Score:2)
Who owns your DNA? It’s not who you think [geneticlit...roject.org]
I welcome the day... (Score:4, Funny)
When my phone talks to me as Scarlett Johansson...
Re: (Score:3)
Re: (Score:2)
Re: (Score:3)
I'd try Christopher Walken. Probably scare the heck out of me initially.
Re: (Score:2)
It'll start out well, but the next thing you know you'll be in a relationship with your phone. It ends badly--ultimately, she just cheats on you and leaves.
Shouldn't matter for long (Score:5, Insightful)
We should quickly advance to being able to use wholly artificial voices and not have to use a specific person to source them. Then this won't be an issue, other than voice actors not being needed (but that's no different to any other obsolete profession).
Re: (Score:2)
Well, at least then the voices would finally match the wooden acting. Right now, the dubbed voices are often way to lively.
Re: (Score:2)
After five years working in live theatre I swore I'd never go back until they'd replaced actors with holograms. Of course at that point they wouldn't need a lighting designer any more either, so the point would be kind of moot. :-)
Re: (Score:3, Interesting)
We should quickly advance to being able to use wholly artificial voices and not have to use a specific person to source them. Then this won't be an issue, other than voice actors not being needed (but that's no different to any other obsolete profession).
You're basically regurgitating the old "buggy whip" argument; but with the rapid pace of technological development, widespread adoption of automation, and concentration of wealth, such arguments are no longer valid.
The time when the vast majority of jobs have been replaced by technology is clearly visible on the horizon, but we don't have any clue at all how people are going to have money to pay for the necessities of life when there are no jobs to be had. Wealth re-distribution from the elites? I don't thi
Re:Shouldn't matter for long (Score:4, Insightful)
The problem is, we have already been through this several times - go look at photos of accounting departments from the 1930s (dozens of people adding up columns in books and then passing the books to another bunch of people to do the same in the next column), or the Littlewoods Pools sorting departments (literally multiple halls with hundreds of people in doing manual checking of betting slips), or typing pools (heard of a business which has a typing pool recently?), or dock workers (containerisation was a bitch for the navvies - no, not that containerisation, the other one).
This isn't something new that we are just suffering now, its been ongoing since the industrial revolution - society has thus far adapted and I see no reason why it wont adapt again.
Re:Shouldn't matter for long (Score:4, Insightful)
We have not been through this several times. It is a cumulative effect, which continues to accumulate. The breaking point lies further still, though it's impossible to predict where exactly.
We've come up with myriad clever arrangements to paper over or mitigate the issue, and they've worked, to an extent. Creating busywork and bullshit jobs. Having "working age" start progressively later in life and end progressively earlier. More people on disability. Extended unemployment benefits. Sequestering would-be (un)employees in forever-war meat grinders. Shorter workweeks. Paid vacation and maternity leave.
It won't work indefinitely. In many countries, worker productivity has already peaked. Workforce participation rate has already peaked. They've been in decline for decades now in the US. Productive (in an economic sense) work is disappearing.
The end-game, without any course correction, is the continuing decline of living standards into an eventual dystopic hellscape. It's hard to predict exactly when, due to the slow and gradual nature of the problem. It may vary country to country, although in light of globalism, a more global view is needed (and beyond the scope of this post.) If we've already hit the peak, then it'll be measured not in centuries, but in decades.
Unless people voluntarily control their uteruses, population control measures are in the future too. In the dystopic scenario, it will just be involuntary culling. Letting people die. Is it any wonder life expectancy has been decreasing for years in the US? It's from deaths of despair - suicides, drug overdoses. The only other alternative is criminality and jail. It used to be mostly black communities that had these issues. (Confining it to the ghetto served as another way to paper-over the problem.) But joblessness is equal-opportunity now. Coming soon to a family near you.
Re: (Score:2, Insightful)
Re: (Score:3)
The farm laborer could go screw on lug nuts when displaced by the tractor, the factory worker could stock shelves at Walmart when robots could do the job better and cheaper. Where are the new low-skill low-education jobs of the future coming from? Any new menial work being produced now is automated immediately.
Oh, I know they can all reeducate themselves and become programmers! /sarcasm
There are a frack of a lot of people who aren't smart enough do do anything more complex than be a security guard or flip
Re: (Score:2)
Re: (Score:3)
I tend to distrust feel-good opinion-based non-solutions for some reason. "Don't worry about it, everything will be fine, we've always dumped our effluent in the river and there's never been a problem." Then the Cuyahoga River caught fire on national television and the EPA had to be created. Just because doing nothing worked in the past isn't a guarantee that it's going to work in the future.
Re: (Score:2)
slow and gradual nature of the problem.
That's about to change, automation appears to be on an exponential curve and we're only now reaching the 'knee' of that curve. Professionals think they're safe, but already an algorithm can scrutinize breast cancer biopsy slides better and faster than an experienced pathologist, only the inertia of the insurance companies keeps pharmacists employed, and while a team of contract lawyers found ~75% of errors inserted in a pile of contracts in 8 hours a trained AI found >80% in under 45 minutes.
Re:Shouldn't matter for long (Score:4, Insightful)
We have never been through this, where "this" is automation with the capability of doing anything a human can do. We're not there yet, or even close, but the pace of progress is accelerating. It will probably happen before anyone is really ready for it, and certainly those claiming it's impossible or just the same old thing again will not be ready.
Re: (Score:1)
From the content consumer point of view, good riddance the narcissistic and blinding greedy actors and hello to a lot more writers and animators and a lot more content. The money wasted on actors spent on more content.
The idiots voice was not used, a timbre and accent similar to the insatiably greedy narcissist was used probably be well over a MILLION others with similar voice timbre and accent.
For the insanely greedy and egoistic idiots voice to be used, the anal idiot, has to actually be the producer of t
Seems fine, if the 'voice model' actually agrees (Score:3, Interesting)
Re: (Score:2)
Except there won't be a next generation of actors and celebrities to feed off of, because the only reason to use stars is because they're already stars (as far as Marketing is concerned). Once that generation dies off, it'll be all AI, all the time.
Re: (Score:3)
Would they really need celebrity voices when they can just generate their own? Just slap a hundred persons with voice samples and let them rate their favorite then feed the data to an "AI", it will guesstimate which voice tone is the most popular with which audience and, just like magic, you don't need pesky humans for this anymore.
Re: (Score:1)
Scan the audience on the way in and modify the voices in real-time to better match the expectations.
No two performances the same.
Re: (Score:2, Insightful)
I can't. This is just silly techno fear-mongering.
Synthetic speech sucks. It's "better" than it was 15 years ago, but it still sucks pretty bad.
Don't give me that "not now, but soon!" line either. If you still buy into that whole "promissory AI" myth after literal decades of snake-oil salesmen over-promising and under-delivering, I don't know how to help you.
Re: (Score:2)
"It's better than it used to be, and don't try to tell me it will continue to improve!"
- narcc
Re: (Score:2)
I guess reading is difficult for some people.
While there have been some improvements, things have not improved to the level that voice actors need to worry. (We saw a lot of this 'computers will take your job' nonsense in the late 70's / early 80's.) Synthetic speech is still terrible. This is just more fear-mongering, like the idiots who think the "singularity" is near and computers are going to destroy humanity like in a science fiction movie.
The claim that AI will ("real soon now") reach some level th
Re: (Score:2)
Celebrities might be able to but I think for general voice actors there probably isn't enough unique about them to make it worth paying them big money. Much of their talent lies in acting and delivering lines as the producer wants, and being reliable. All things that computer generated voices can or will soon be able to do.
It will get to a point a bit like the "create a character" things in video games where you move some sliders to create the exact voice you want. There might be a new job that opens up, th
Re: (Score:2)
I don't see the problem (Score:1, Interesting)
I pay a guy to build me a fence and I give him money and then I can use that fence for as long as it lasts, and I can re-paint it and fix it and change it and I don't owe the guy I paid to build the fence more money.
Re: (Score:3)
Re: (Score:2)
I would say if I paid the architect for their time, to develop the plans then it should be mine, not theirs, just like when I develop software for the company I work for, it doesn't belong to me it belongs to them. If they develop the plans on their own and then sell me a copy that only partially compensates then it can belong to them. Even so not forever, its not like that plan isn't using other ideas that other people have developed in the past.
Legally its about the contract you have at the time. But in t
Re: (Score:2)
Then, that's something you would need to negotiate up front, making it a "work for hire." That's not automatic, and an architect would charge you more for that right.
>when I develop software for the company I work for, it doesn't belong to me it belongs to them.
Because, that's automatically considered a work for hire. Do some research.
Re: (Score:2)
Re: I don't see the problem (Score:4, Interesting)
Well in this case it would be more like I pay you for 10 house designs. Then I pay for an AI system that analyzes the designs, extracts elements of your architectural style, and makes automatically as many variant designs as I want.
It's actually kind of a gray area copyright law wise and I wonder how the courts will try to untangle it
Re: (Score:2)
Re: (Score:2)
Basically the monkey selfie [wikipedia.org] copyright case substituting AI for the monkey.
Re: (Score:3)
That would clearly be a derivative work
But that's how human architects today work, right? Using elements from existing architecture and using them in your own is about as much making it a derivative work as someone being a foreigner, learning English from English novels, and then writing an original novel. Just because you're using the same English words as other novels does not necessarily make your work a derivative one, even if you learned those English words from other copyrighted texts.
Re: I don't see the problem (Score:2)
Re: (Score:2)
Re: (Score:2)
However, the equivalent in this story is even more murky, or clearer in a different way.
Let's say I get people's permission to analyze their voice. I get a million samples of peoples voices in a good 30 second clip or so designed to highlight what I want. I then develop a set of parameters based on those samples to synthesize as many variants as I want... Except there is no person from the original set of people who could claim ownership over my work.
So... (Score:2)
They can't feed (i.e. copy) the audio into an AI without violating copyright. So, there's that.
Re: (Score:2)
Re: (Score:3)
Re: (Score:2)
Re: (Score:2)
They can't feed (i.e. copy) the audio into an AI without violating copyright.
For example, In my jurisdiction, this is not true. Since you're not redistributing the audio but using it personally, copyright law is not involved in that in any way.
Re: (Score:2)
"They can't feed (i.e. copy) the audio into an AI without violating copyright."
Unless the contract the voice actor signed said they owned the copyright. This isn't rocket science. Voice actor contracts will likely include this moving forward. High demand ones may balk, but relative nobodies like the Siri lady will get paid for a day's work and move on. Realistically, should she be making millions from Apple?
No, AI is not screwing over voice actors (Score:2)
Business executives are using tools to screw over employees and contractors, just as they always have. AI is not making that decision, people are.
Who needs actors? (Score:2)
Just computer generate them based on colorized actor images and mannerisms from super old movies (pre-1926) those are not under any copyrights. Actors will always want increasing amounts of money and ownership over your art and creativity. If they aren't providing a useful service they should live off taxation. Being forced to use workers or actors is a form of taxation anyway. It's better to pay that tax to the government than directly to the workers. Let the government deal with their issues.
Re: (Score:1)
And when they come for your job, you'll change your tune 180 degres.
People like you always do.
Re: (Score:2)
Nope, I'd be glad to live on unemployment, which is what I am proposing. Read what I wrote again, twice.
Re: (Score:2)
So should every job be preserved forever?
Related: (Score:4, Interesting)
Actor Crispin Glover sued a movie director for cloning his face into Back to the Future 2 and won a major settlement and a ban on the practice between the Screen Actors Guild and filmmakers.
Plenty of people will give redistribution rights (Score:3)
Even if her contract with the Institute of Acoustics didn't allow redistribution, plenty of people will allow it, for not a lot of money either ... and in the end that's all this is, redistribution, allegedly without a valid license. There are no deeper legal questions, just questions of contracts and strict liability.
ByteDance almost certainly got fucked by the Institute of Acoustics.
Big Fleshy Deal! (Score:2)
As someone who has actually tried to hire a voice actor for a small bit part, I say "YOU HAD IT COMING YOU GREEDY LOSERS!"
When you need 30 seconds of voice for a small time advertisement, and they're asking for nearly 10k, you wind up doing it yourself (I actually wound up getting a friend to do it). The issue I have is this: Intellectual property (in general, not just voice acting) is like a lottery. If you are crazy lucky, you can make millions of dollars off of a few hours of work... or if you are not
Re: (Score:2)
What's unique about automation here? (Score:5, Insightful)
I've heard commercials in which the voice actor sounds like Morgan Freeman but isn't, or sounds like James Earl Jones, but isn't, and similar. So this technology uses ML or similar to perform essentially the same degree of mimicry as one of these impressionists. A human impressionist is, umm, human, of course, but is this really so different? It's legal to emulate the appearance and mannerisms of another person so long as it's not done with the intent or result of representing oneself as that other person, but that doesn't seem to be the standard here.
Re: (Score:2)
Titty sprinkles. [youtube.com]
Re: (Score:1)
Re: (Score:2)
I remember the first time I saw this vine [youtube.com].
Actual background (Score:3)
Which Vice didn’t bother to provide:
“Bev Standing recorded about 10,000 sentences of audio for the state-backed Chinese Institute of Acoustics research body to use in translations, in 2018.”
And now TikTok is using those recordings for a different purpose.
So another case of China stealing and copying it seems.
https://www.bbc.com/news/techn... [bbc.com]
Re: (Score:2)
The company she recorded them for and which sold them on is British. I doubt ByteDance bought them without a license to distribute.
Re: (Score:2)
Why wouldn't they? Nobody ever wins a lawsuit in a Chinese court against a Chinese defendant unless something so blatantly illegal has taken place (like melamine baby food) that the perps are going to be made examples of. So as long as they don't fuck up that hard, they're not going to get penalized for anything else, at least not within China.
Re: (Score:2)
It's not a lie. I had to take the warnings provided by insurance carriers and translate them to plain English for bean counters. One of the things explicitly stated in my source documents was that the chances of winning in a Chinese court against a Chinese defendant were about 1 in 20, regardless of the merits of the case. Also, any foreign operations have to form a joint venture with Chinese nationals, and the Chinese ownership must be no less than 51%. By law.
Re: (Score:3)
Apple did the same thing with Siri. The woman they used for the US voice and the British guy they used for the UK voice had no idea. The British guy has said publicly the first he knew of it was when he heard and iPhone talking in his voice.
Seems most likely that they didn't realize they were signing a contract that allowed this kind of use. It's fairly new, most recordings were used for playback rather than AI training until quite recently.
James Brown : The Man Of Screams (Score:2)
Any idea how often his screams were used on other records?
Re: (Score:3)
"whosampled.com" meticulously catalogue an enormous library of samples. For James Brown, they have a total of 14,124 samples used in other people's songs.
That's a lot of screams.
https://www.whosampled.com/Jam... [whosampled.com]
Beserker novels and voice synthesis (Score:2)
In one Beserker novel (forget which) there is a mention that the synthesized voice used to talk to the humans had been put together from the screams of humans being tortured by the Beserkers (and you could still here echoes of the screams). I think of that sometimes when using telephone voice prompts -- in terms of wondering if the people recording the prompts really wanted to do that or felt pushed into it by economic necessity.
https://en.wikipedia.org/wiki/... [wikipedia.org]
Tough one (Score:2)
In a few specific instances, I think this is a good thing. Like Mark Hamill will always be a top-tier Joker voice and no one else has ever really come close to capturing the essence of the character. On the whole though, with the increase in automation, it's time we just bite the bullet and accept the need for a universal basic income. Enough to cover rent, food, and maybe a little more. You won't be rich, but you also won't be homeless unless you choose to be. The sooner we start the sooner we can work out
Re: (Score:2)
But who would be the best Mexican Joker?
On the Way to the Inevitable (Score:2)
Re: (Score:2)
"Clearly, impersonating someone is illegal and unethical. "
Tell that to Rich Little, Matt Parker, Trey Stone, Jimmy Fallon, 95% of SNLs cast, the thousands of Elvis impersonators, or all the 'Elmo's / Spidermen / Buzz Lightyears / Ironmen / Chers / Doras / Minions' in Time Square.
Re: (Score:2)
$7000 for legal expenses (Score:2)
It will occur eventually (Score:1)
They could sue and win over their own voices, but this problem could be overcome easily in the future, creation new synthesized voices but human-like, and use only interchangeable voice actors as a guide pattern that the AI will turn into the final voice. So the enterprises will own their own created voices, and voice "guiders" could be replaced without big problems.
Very useful for things like games, which will turn free from voice actor time limitations and price renegotiations for extensions (like a DLC o
I read this in Bugs Bunny's voice (Score:2)
Re: (Score:2)
"intellectual" property (Score:2)
If we don't grant exclusive rights, then people won't have an incentive to have a voice! Imagine a world where nobody knows how to speak, because the market has been destroyed.
Won't be long until the voice artists are out (Score:2)
Markup language for kinks in pronunciation, fully computer generated sounds, no human template needed.
Or go a middle route. Hold voice auditions, with minimal compensation for those picked, and pick 10,000. You, TOO could be "advertising famous"! Then you have 10,000 voices to choose from.
Where do I sign? (Score:2)
So, you get unlimited access to my content to do with as you see fit and I get a chance to make easy money? ..
Sure, why not
Funny how often people lol when signing a contract or agreement because all they see are potential dollar signs and somehow think things can ever be anything but positive for them.
Key question? (Score:2)