All of it. Genuinely. Don't missinterpret this as me being pro-AI, because no, absolutely not.

But from a purely technical point of view modern AI is an amazing achievement. The way we are doing it is stupid and unethical and destructive to the planet, but the tech is quite a feat.

It's like if the Wright brothers pioneered flying for the sole purpose of bombing countries.

It's like if Watt built the steam machine for the sole purpose of causing the climate crisis.

it's baffling.

Being a techie myself I'm just sitting here like

"This is a genuinely great feat of software engineering, but do we really have to burn the planet down for this? Also, do we have to steal artists's works for this and then try to put them out of their job? Also, that whole misinformation thing..."

(Also, I already said that in the beginning, there are a ton of ethical problems with genAI as it is. Like... just don't use it. Period.)

At the end of it that's the big catch for Voice AIs in particular. The technology we have there is genuinely amazing.

it is just used in the worst way possible.

Because companies see AI as a tool, but as a way to reduce labour costs. plain and simple.

But also also also genuinely just paying voice actors or working with friends or volunteers for voice acting is probably the less labour intensive solution at the end of the day.

But as there is no ethically sources voice AI, and since nobody is actually putting in the work to finetune the dialogue, that entire point is just moot.

(Also this doesn't work for big companies. Like, they can afford to pay people)

Like, if i was creating a game for funsies, and there was something like open source voices and an ethically sourced model, I'd genuinely see no evil in using those as a workbasis for voices in the game if I put in the effort to make it sound good.

Like, you could even make the argument in this specific case, if an ethically sourced AI would exist, a small creator putting in hours of work to fine tune dialogue for his creation using ethically sourced voices could be a fine example of AI actually being used as a tool to create art.

Hell, you could even argue that that's just in line with the fine tradition of audio directing and editing (even if, arguably, it is a bastardized version of that tradition)

That's basically completly contradictory to the common AI mindset: That you can just write the prompt and it'll spit out the finished work. Maybe you'll finetune the prompt a little and that's it.

But sitting down, editing dialogue in fine detail? That's actual work.

BUT:

You need to put in effort. That's the big thing. Every tool gives you lots of possibilities in defining the tone of voice, the speed, even the emotions. You just have to go in and try and error yourself into what's sounding good.

And that's the catch.

And people will be like "Oh, the quality will improve", but.... The quality already has improved.

I'm serious. TTS has essentially been a solved thing for a long while. The tools on the market are already extremely potent and you can do some really naturaly sounding dialogue with them.

So, the fascinating thing about Voice AI is to me.... that basically nobody uses them to the full extend of their functionality.

I'm serious.

Like, if you watch a clip from basically any AI dub, it will suck. The acting usually just won't match the scene.

Big mood

Iwie vermenschlicht das unsere Vergangenheit auch auf eine schöne Art. Einfach eine Familie mit Kind und Hund.

Iwie ist viel gelehrte Geschichte über Staaten und Herrscher, dass man fast vergisst, dass da auch einfach Menschen waren, die geliebt und gelacht haben.

A lesbian orgy you say? 👀