Anyone understand why multiple demos have voices singing now.. how do they do it ?

Most full demo (no tech or sizelimit) soundtracks since the early 00s are just mp3 streams or alike, size-coded that have soft-synths or retro categories were singing is an issue due to datasize or hardware power often don't (sometimes they do as a technical demonstration).

But I did notice some 64k's and small synths-executables had singing this year, I've added small voice-samples (compressed) but that's just seconds whilst these entries had longer sequences so I'm a tad curious as well.

It has been a thing for some time. Even back in the 16-bit days:

- Movement Mina Omistan (1994): https://youtu.be/Ryq8fz_XAhI

- TBL Tint (1996) second part: https://youtu.be/KcSV4qAG-xk?t=459

I wish there was more info about how they're made but I suppose part of the demo scene has always been secretive and reverse engineer it if you want to know.

The first one said made with Godot and blender so is it using premade assets? (Which I thought was not allowed)

Edit: #22 Heart of Glass says 144kB of WebGL + audio. I assume this means the audio is separate.

Literally just singing! It’s an mp3 :)