Comment on I am abandoning Arc Raiders because Embark is stealing voices through AI
realitaetsverlust@piefed.zip 18 hours ago
Watkins, speaking to PCGamesN, elaborated that the text-to-speech always starts with a voice actor: “It’s part of their contract that we use it [AI] for this purpose, and that allows us to do things like our ping system, where it’s capable of saying every single item name, every single location name, and compass directions. That’s how we can get that without needing to have someone come in every time we create a new item for the game.”
So no, they are not “stealing” voices. Their contracts explicity states that they are training the model. So they are getting paid, which in conclusion rules out “stealing”.
Also, from your video:
But to my understanding
Rarely ever good if a sentence starts with “To my understanding”
the AI tools Embark uses to then synthesize the rest of the performance come from models that are trained on millions of other voice actors that have been stolen from in the way that all generative AI models steal from artists.
No, that’s the whole point of models that are trained on a single voice - you do NOT use other voice actors because that would completely muddy the voice. The models are trained on a singular voice to mimic that person perfectly. Using other voices is like asking someone to cook a potato soup for you and then you toss in tomato and paprika.
Chef6652@lemmy.world 17 hours ago
Thank you for the sourced answer.
First of all, I’m not the author of the video. I’m only a concerned customer/fan.
Secondly, I am already boycotting Windows for those kind of reason. And I have been a Linux user for the better part of the last 10 years. You are right, we should fight this kind of stance. But this is a different beast imho.
Now to answer: I will copy/paste most of my answer to kate above.
I find this really cynical. Because:
To back this term (“stealing”). I consider that the genAI technology as a whole has been built on the stolen work of the whole world. ChatGPT has been built upon thousands of Github projects without their consent, I suppose Sora, DALL-E, Stable Diffusion, etc have been made in the same way. Training data has been stolen.
Even though it’s a bit far-stretched, using this kind of technology to remove any future human collaboration is stealing to me.
I personally do understand how genAI work. But thank you for clarifying what is usually a black box for most people. Let me phrase in detail what my understanding is. And feel free to correct me if I’m wrong.
What Embark did is to train a specific model on a specific voice. I agree. But the base of this model is to be capable of understanding how any voice works in order to copy how a specific voice work. In other words. This means that the technology/model has first been built using thousands of recordings to “understand” how human voice works. And then there has been a last layer to copy a specific voice. I am simplifying of course. But I want to express how much I am disagreeing with the “they only used one specific voice”. Technology isn’t magic, we as a society don’t understand how to make a program that can copy a single voice to make it say anything perfectly. We can use an amalgam of thousands of them to build a neural network (what some call AI) to statistically mimic what a human voice is. And then feed it a single voice to mimic.
Sorry for the long response.
realitaetsverlust@piefed.zip 15 hours ago
The didn’t have to, but it certainly makes it easier. And I find it silly to not use a technology that makes something easier if you have it available. That’s like a farmer plowing his field by hand instead of using a tractor.
First of all, we had Text-To-Speech way before any kind of generative AI. In germany, we had speech synthesized announcements on railway stations for like 15 years at least. Like this here: https://www.youtube.com/shorts/AuIkJ_UGltI. We also had vocaloids for decades now. So it’s wrong to assume we had no idea how voices work before AI.
Second, I get your stance on “I’m not using AI because somewhere up the chain it was developed by morally ambiguous ways”, but I don’t think that makes anything better. You should rate the current use-case, not something that happened earlier in the production chain. AI in itself is not bad. If used properly, it’s an incredibly helpful tool. There’s other and much better hills to die on imo.
Chef6652@lemmy.world 15 hours ago
To follow your analogy:
“The tractor isn’t made off of other people living wage” is what I was starting to answer but… it does a bit, so you are right in a sense. Previously, people where hired to plow the field I guess. And I don’t see this as big of a problem as genAI. But this subject touches more on mass production and capitalism.
I would like to intentionally bend your analogy by stating that it’s not inherently true: if anybody could access a technology that helps them by magically destroying lives in another country far away, would you say the same thing? “It would be silly to ignore it as it makes things easier for me” seems quite short-sighted to me.
stephen01king@piefed.zip 15 hours ago
You’re already using something convenient to you at the cost of exploiting other lives far away. It’s your smartphone.
realitaetsverlust@piefed.zip 12 hours ago
Might be cruel to say it, but that’s called “progress”. The world needs to continue to evolve - latching to old jobs seems silly. We got rid off of blacksmiths because we don’t have the need anymore. Europe once had a huge horse stable industry spanning the entirety of central and western europe. We don’t have that anymore either, because we now have cars. We also don’t have any telegraph operators or switchboard operators (necessary for long distance communication back then), elevator operators or laundry washwomen - these jobs have all been made obsolete by technical advancements.
I think quite the opposite - it’s the long-sighted better option. Progress is never good for those negatively affected in the short term, but we can’t keep jobs around that aren’t really necessary anymore just for the sake of those people having a job.
And in this particular case, there’s not even any loss involved. They used their voice to train an AI, it was explicity part of the contract and they got paid for it. I honestly do not see the problem.