LÖVE

I've been wondering how one generates sounds for talking, though their gibberish(like the sims), with different pitches and deepness, like Reigns, so it makes the character seem more real.

I don't think The Sims uses generated voices.

You could try with some sound synthesis software, feeding it generated words. I wrote a word generator for Spanish-sounding words (post in Spanish): http://orden-y-concierto.blogspot.com/2 ... abras.html

Record short voice samples, then play them randomly in rapid succession.

Actual AI generated gibberish is a new technology that hasn't matured yet. A couple years out.

Thnx for replies...

pgimeno wrote: ↑Mon May 03, 2021 8:19 pm I don't think The Sims uses generated voices.

True, I was using it as an example of gibberish talk.

I saw this soundex thing. Does that work? Can I change how deep the voice samples are through code?

Is it not as simple as taking a range of 3 or 4 seconds snippets of vocals and randomly changing speed? Or is that the challenge - working out how to do that in Love?

togFox wrote: ↑Tue May 04, 2021 12:41 am Is it not as simple as taking a range of 3 or 4 seconds snippets of vocals and randomly changing speed? Or is that the challenge - working out how to do that in Love?

No, I don't think it's that simple (though you may be on to something there, a cheap method of doing it...just might work). The Soundex and gibberish algorithms I'm looking for generate a sound that's closer to an actual language. I don't have a link but search for gameplay of "Reigns" and see how the characters talk, they sound like they're speaking a different language ans have different personalities just due tothe way they changed pitch. Something like Animal Crossing too, but less annoying.

Can I manipulate sound files in löve(You know, like an mp3 file that wants to marry a .wav file, no ignore that) to change amplitude, pitch ,etc?

I see. I just watched 30 seconds of game play on YT. You can go down the algorithm route or have a range of 'syllables' that are very short and play a random sequence back to back. Include 'blank' with male and female variants.

Or invent a cool algorithm.

Making beeps and boops is simplest.
Sliding their pitch like in Celeste is a bit less simple.
Using a sample instead of a generated sound is also a simple modification.
Trying to change the pitch of vocal samples past some percentage from the original will sound bad (with simple resampling, that is)
Trying to change the quality of vocal samples to be more male/female is hard (since pitching them simply will affect formants, you know, the vowel components and stuff, and you'd want to just modify the formants, not the base pitch, for this i mean; you might also want to separately also change the base pitch as well.)

Gunroar:Cannon() wrote: ↑Tue May 04, 2021 1:06 am Can I manipulate sound files in löve to change amplitude, pitch ,etc?

If you read the wiki once in a while, you'll find such neat functions as Source:setVolume and Source:setPitch.

Xii wrote: ↑Mon May 03, 2021 8:42 pm Actual AI generated gibberish is a new technology that hasn't matured yet. A couple years out.

Agreed to a point; gibberish is somewhat doable if you already have both a decent dataset and processing power that's not run locally on one machine.

togFox wrote: ↑Tue May 04, 2021 4:37 am Include 'blank' with male and female variants...

What do you mean here?

togFox wrote: ↑Tue May 04, 2021 4:37 am.
..Or invent a cool algorithm.

Sorry, I'm not that smart

Thnx for the elaborations....

zorg wrote: ↑Tue May 04, 2021 5:31 am Making beeps and boops is simplest.
Sliding their pitch like in Celeste is a bit less simple.
Using a sample instead of a generated sound is also a simple modification.
Trying to change the pitch of vocal samples past some percentage from the original will sound bad (with simple resampling, that is)
Trying to change the quality of vocal samples to be more male/female is hard (since pitching them simply will affect formants, you know, the vowel components and stuff, and you'd want to just modify the formants, not the base pitch, for this i mean; you might also want to separately also change the base pitch as well.)

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).

If you take my suggestion to have a lot of short voice segments and stitch them together to create a 'language' then make sure some of those segments are 'pauses' and make feminine varieties for your female persona's.