pouët.net

Virgill: Actually no, the frequency distribution of the bands is wildly guessed at best. Will try!

good lord, dat cables.....

Here´s another example what 64klang2 is capable of:
physical modeled trumpets, timpani, percussions, brasses and strings in different articulations. Granular synthesis used for the piano.
https://soundcloud.com/virgill/physical-symphony
next thing i gonna try is modal synthesis. :)

mind. blown.

question: How big is the songdata+player for something like that song?

nevermind, there is a link.

Virgill and Gopher, you fucking guys are insane :)

Virgill and Gopher, you fucking guys are insane :)

This is so tempting... where is the download for 64klang2? Please, don't try to convince us that this amazing tool is not yet finished enough!

I remember doing some simple "physically-based" flute synth based on P.R.Cook research for 64k back in the days - you could even "overblow" it, but I never got this far - great job!

tomkh. Yes, the slide flute algorithm is nice.. Basically it´s simple but getting the right parameters is not that easy ;)

Virgill: sure, so you basically use similar approach? i.e. a feedback loop with LP and delay controlled with note fequency (plus reed/mouthpiece model).

tomkh,yes ;)

Speaking of which, would be nice to do sound synthesis based on actual object interaction/collision in 4k or 64k intro, like in those siggraph papers, shouldn't be so hard I guess, i.e. a ball falling down a grate, something breaks apart or a hammer hits a cowbell in different places and makes slightly different sounds.

omg! it's awesome

the above .exe is packed with kkrunchy (asm07 i think).
the only size optimization from my side was skipping code from nodes that are not used at all.
optimization of the patch and song data as well as skipping node-internal unused code was not included there yet.

i just started with that just a couple of days ago.

the following current stats are from crinkler because it gives nicer and more reliable output than kkrunchy. if you dont count in the decompressor overhead i found crinkler and kkrunchy to be comparable in net packed size.

init/player code (optimized):
1200 bytes

synth code (almost optimized):
6930 bytes

patch data (optimized):
3550 bytes

song data (almost unoptimized):
5310 bytes

general synth data (unoptimized):
830 bytes

total:
17820 bytes

song data optimization is next on my list and there's probably 1k to gain there.
let's see :)

Another 64klang2 example by pOWL :
the last sequencer
The glitch module is great :)

that's awesome!

the voices are the beginning, there are compressed samples right?

more physical modeling with 64klang2:
https://soundcloud.com/virgill/binary-rhapsody

Why not simply give us 64klang2 once for all?

So many amazing examples and, still, never "finished for the public". It's been almost a year since "Computers are Gay" was released.

Will it be published for real? When? Come on!

ham, it´s still work in progress ;)

At this rate its better off writing your own synth like everyone else.

maybe you should apply for beta-testing...

Nah, I should learn basic music composition and code my own synth. There is plenty of audio DSP info around and I really should learn the MIDI format.