Saiya6Cit wrote: ↑Wed Apr 13, 2022 12:16 pm
Frist of all, I want to thank TheBlackPaladin for sharing his knowledge on the subject, I learned a lot.
Second, I wanted to know your opinion but what I personally think is that AI voice synthesizer is OK as long as the voice actors get a percentage everytime it is used, probably cents, but it would be a compensation which in my opinion they are entitled to get. I think the closest to that would be how spotify works, artists get paid for their music, but it's not just them, it's everyone involved in the music industry. Could this mean that in a similar way that we faced copyright issues with images since 2018 now we will face it for digital voice clips? Deep regulations It would make sense honestly.
Well thank you, I'm glad I was able to provide some helpful info.
As for compensating actors for their synthesized AI voices, that's a tough to answer for a number of reasons (not the least of which is that I'm just one person, so I don't mean to speak for all actors). Personally, my thought is that how an actor should be compensated for their AI voices should vary on a case-by-case basis depending on how widely used the voice is. Like, if it's just used to play a phrase like, "Welcome to the library" whenever somebody enters a public library in some small town in Wisconsin, that should be a relatively small fee, but if it's used as the voice in a commercial that plays across the entire country, that should require far more money. Spotify is probably not the best economic model for AI usage anyway since it--in my view--
screws the musical artists with how low the royalty rates are. As this article notes, a song would need to be streamed 314 times for the royalty payment to even reach a single dollar.
So, there is no one-size-fits-all solution, economically speaking. The closest thing that has been proposed to an official rate structure for AI usage of an actor's voice never came to pass. A synthesized AI voice company approached SAG-AFTRA (the US actors union) about creating a contract and rate structure, and one was indeed created...but the company decided at the last second, for reasons that are not clear, to bail on it.
Honestly though, the far bigger issue beyond financial compensation for an AI voice is the
usage of the voice. In other words, having an app that can make anybody convincingly sound like they said anything...has some serious potential consequences. It's one thing to get an AI voice say funny or silly phrases, but there's a far darker flip-side to that: it could also be used to make the actor sound like they said something horrible. For example, what's to stop somebody from using an AI voice to make an actor sound like they said something horribly racist? Or misogynist? Or homophobic? Or advocating a political view or political candidate that they don't actually support? Or making a bomb threat?
See what I mean? There are some
enormous ethical dilemmas around usage. There are certain things that, I don't care how much I was paid, I would never want an AI voice of mine to say. Especially as the technology advances and becomes more convincing, as it does every day, and people could end up mistaking the AI voice for something that the actor actually said in real life. Any smart actor who agrees to have an AI copy of their voice made should insist that their contracts include a clause saying that the actor has to personally approve their voice in every instance that it's used. All the more reason that Mario and Rene were 100% right to ask that their voices be removed from FakeYou.
Saiya6Cit wrote: ↑Wed Apr 13, 2022 12:16 pm
Two other very important subjects have been exposed here:
TheBlackPaladin wrote:
1) Toei could one day say, "No more paying humans to dub, we're just going to use an app to synthesize the Japanese voice actors in another language."
You know that is very terrifying. I have worked as a live interpreter (like a protocol droid C3PO) and people would constantly joke sayuing they would rather use Google translate
There is no way machines could ever get the level of complexity and emotion that can be printed into human voice. Voice actors work really hard, humas should never be replaced for that.
I agree that they shouldn't, but unfortunately, the technology is getting better at emoting. Some genres of voice-over like E-Learning, audiobooks, and audio description are already putting AI to use and getting rid of human narrators. Especially in cases where tons of emoting isn't necessarily required. Even in cases where emoting is required though, the technology is getting better. Obsidian, the video game company,
recently started using AI voices for the early stages of their game development cycles. At the moment, even Obsidian will tell you that they're only using AI for the early stages of game design--basically using AI as "placeholders," to eventually be replaced by real human voice actors--but even a couple years ago, that would have sounded absurd. That's the nature of technology: it's constantly improving.
Heh, so yeah, I'm not a fan. Can you tell?
In any event, the future is not necessarily all doom-and-gloom. More and more laws are being created to address issues like these. In California, for example, it is explicitly illegal to use AI voices in political ads. The cynical part of me wants to say, "Wow, as soon as something threatens a politician, look how quickly something is done about it," but hey, it's something. In addition, as the threat of visual and audio deepfakes increases, more and more software programs are being developed specifically to detect deepfakes. Finally, when it comes to actors, plenty of actors unions are discussing this issue and drawing up protective clauses.
I remain cautiously optimistic that we can reach a future where AI is used in an ethical manner, and that cases where it isn't can be mitigated and exposed.
