Cadaver747: I have no idea what is "MTL games" and Google search results makes we wonder.
MTL from what i see tend to be
Machine Translated Language games. Either you rip the entire language content, dump it in google translate, bring it back, and either put it back in untouched or modify it to be cleaner. For untranslated games, you get the raw (
google's best guess of) text, which can be a little confusing but 95% of the content gets across from what i see. But seeing as it's already being sent to google, getting that text back and sending it to get an AI voice would just be another step in the process.
heh actually i remember a fan translated game where items descriptions were not translated, put it in, and there was quite a bit of lore, too much to fit in the english version though, and definitely rough. Might finish that side project on that game i started eventually...
Cadaver747: So far I wasn't successful at finding any old game enhanced with AI voicing. There is a small mod for TES: Morrowind with some AI voice, sounds really cool, a bit emotionless but it suits well Dagoth Ur.
https://youtu.be/xn2vs196_Js As for the old games from DOS era I think adding voice acting would be a pain and might require some serious reverse engineering, I wish AI could do that as well. There is a fork to ScummVM called ScummVMX by Milo, theoretically every game supported by ScummVM can be *easily* enhanced with voice. As of now all I could find were some videos with upscaled textures, some were hideous at butchering characters' moving mouth.
In games where text pops up (
preferably in a interactive window or conversation) finding said function, i don't see it being that difficult. you'd add a call hijacking the original function. for DosBox type games probably add a new interrupt or special function which sends requests for a cached audio from text (
and if doesn't exist to download/generate it), then play it, returning to the original function.
Mind you this has it's own flaws, like suddenly getting onslaught with all the conversation replies at the same time, all using the same voice (
when it should be changing) one after the other even if you didn't select one... It would also likely be game specific, so each game would have to have this process done to it, or if it's part of an engine, modifying the engine may work. As you said, some reverse engineering would be needed, and who knows may be worse than unvoiced at all. Or maybe fan updated versions of the game(s) or engines would allow you to add in the extra functionality.
Then again maybe going forward devs and programmers will simply include a few new functions (
empty and unused probably) for grabbing synthesized voices at a later date; Letting you inject the modified/fixed version later making all future games future proof for that. But that is just a general thought, I'm not expecting that to happen any time soon.
Who knows, maybe AI will get so good it will get bored and decide to do all the major work for us with updated bug-free games, voice acted, working with the latest OSes, and many other features like good up-scaling of graphics quality improved sound and many other things. While nice to think about in a Star-Trek utopia 'past our problems' it's hard to see it happen in my lifetime. Then again, if the AI scales up and in another 30 years it might just be competent enough to do it (
assuming it doesn't Skynet and nuke us all, though it seems a lot of AI's when let to scour their own information and grow, tend to be based and anti-left, so maybe there's hope in that)