LÖVE

Gunroar:Cannon() wrote: ↑Tue May 04, 2021 8:05 am What do you mean, how I change the base pitch(with setPitch?), then how do I change the formants, that should be with setPitch I think).

With what löve gives you, you can only change how fast a sample is playing, that's resampling; with longer samples, if you setPitch(a) where a > 1.0, it will play with a higher pitch, but also faster; if 0.0 < a < 1.0 then it will play with a lower pitch, but also slower.
This is the simplest method, and the only one anyone can do with what löve offers you out of the box.

Any other solution is by comparison very hard to do.

I've checked Reigns gameplay; it looks to me that these are also pre-recorded from human actors, like The Sims, except the snippets are shorter and there is some "mix and match" (concatenation) of snippets.

You can (probably) record some voice samples, get the Fourier transform and then generate some random simplified versions of this samples.

darkfrei wrote: ↑Tue May 04, 2021 3:54 pm You can (probably) record some voice samples, get the Fourier transform and then generate some random simplified versions of this samples.

https://github.com/zorggn/zorg-fft
That exists if you are brave enough to try that, and don't want to mess with non-lua libraries.

pgimeno wrote: ↑Tue May 04, 2021 3:14 pm I've checked Reigns gameplay; it looks to me that these are also pre-recorded from human actors, like The Sims, except the snippets are shorter and there is some "mix and match" (concatenation) of snippets.

Yeah, I figured that, this was all about making sounds like that (that priest says "ash-kelem" too much), though if it wasn't and they used an algorithm like soundex(not really random but certain sounds for certain words) they would all say the same thing for each scenario, since the scenarios' texts are always the same.

zorg wrote: ↑Tue May 04, 2021 4:31 pm

darkfrei wrote: ↑Tue May 04, 2021 3:54 pm You can (probably) record some voice samples, get the Fourier transform and then generate some random simplified versions of this samples.

https://github.com/zorggn/zorg-fft
That exists if you are brave enough to try that, and don't want to mess with non-lua libraries.

Seems easy enough to use, thnx, though I'll go for something by recording certain samples then randomizing them.
And I was thinking, if I group words into ... groups(like soundex algorithm, ban and bam both "sound the same" making something like b119), then play the sounds for each of those letters/queues for a certain sound to play, would that work or does it seem like to much trouble for a game?