Open303 - open source 303 emulation project - collaborators wanted

[antto]

antto, the venom is sounding very nice now!
totally different from beta2. c'mon: the people
need beta3, hahaha...
-----------------------------------------------
and t'be fair, the x0xware could be trimmed down,
with stuff dropped, to fit on orig' cpu?

thanks ;]
well i already dropped all of the original x0x features, the chip has just 16KB for code, and 0.5KB has already been taken for the bootloader (which is a must)

[mztk]

yeah, the atmega whatnot, 162?...did you get a board yet?

so...hey: all the more reason for not using the onboard sequencer at all
(kinda defeats the purpose), and have a clone sequencer in software, as a
plugin? be good for testing, anyway, even with no sound.but i guess you'd
have to do it all over again once the board arrives...if it were just a
VSTi, you've probbly finished it already, right?

[antto]

well sure one could make just a sequencer which sends uhm cv/gate/slide or midinotes.. but it's not the same ;]

no, the cpu mod is still not available..

[ParadoxUn]

I took note of that the original 303 filter needed much more feedback to self-oscillate. I emulated this by taking a mix of the 2nd and 3rd pole of the filter of the feedback instead, of the fourth. (2nd does not selfoscillate, 3rd does). It seems to affect the sound of the filter somewhat. The interesting thing is that it needs more headroom now, so saturation/clipping plays a large part in the sound of this.

[mztk]

well you could 'litter the internet' with this,
nobody would mind if there were different versions!

[rola]

Does anyone still have the samples rv0 recorded?

[antto]

yes, i do
btw, i also got a x0xb0x now (which in most aspects is exactly the same thing as a 303)

[rola]

How big are the files? Any chance you could upload them somewhere?

[antto]

most of them are wav files, and some are 88.2KHz stereo
i also have newer recordings which aren't mentioned in this thread
and also have my own recordings from the x0xb0x (and i can make more if anyone really wants)

older files: 89MB - most of them are mentioned here, and are ogg/mp3 and some are wav
test1: 175MB - wav only, testing envelopes, accents, slide/tie, slow/fast tempo
metro: 14MB - mp3/wav, testing the metronome sound

total: 278MB
i can't upload this to box.net, it'll be too big
i hang around in irc.freenode.net #synthedit and i can send it via DCC if that suits you
RARed: 110MB

[rola]

That would be great. I don't need any new recordings right now. I'm just testing some stuff in reaktor.

[mztk]

little bump.

[DuX]

I love your AciDevil version of this synth, Robin. Drive gives it nice bite. You said you will *see if you can find and post* some newer build of it? That would be awesome! Maybe some of these unused parameters will get used in this newer build? LOL

I love VST plugins without the GUI. It makes you really hear the sound, what you're doing. Only when there are too many parameters, I think GUI would be in order. AciDevil doesn't have so much parameters, and it's so nicely tweakable without any GUI. Love it! GUIs usually make sound crackle and make problems of every kind, too. I would make all plugins GUIless! LOL [joking, as I said some have too many parameters for that... just look at Voyager]

Thank you in advance, cheers!

[antto]

i was working on the good old ladder filter (from kunn)
i also messed around with the good old moog ladder (from mystran with Robin's tuning coeffs) for another synth
and i figured my resonance gain compensation was wrong all of the time
so i thinkered a bit and i think i got a better formula now
and a few more things pop up with that

here's my logic, you know the ladder filter, 2 coefficients a0 and k
i have mapped k to be a function of a0
and then a0 is properly "tuned" to actual frequency

remember that, for a moog ladder, k=4 at max resonance, which for the digital variant is only 4.0 at 0Hz and must be boosted with frequency (because of the 1-sample delay issues)
and because of that, the output gain also drops <- THAT is what i try to compensate

and here's the new formula i use, which *looks* right

Code: Select all

a0 = <approx as a function of (Fc/Fs)>
k = <approx as a function of a0>
// k is always greater than 4.0
rgc = (1.0 + r * (k/4.0 - 1.0)); // Resonance Gain Compensation ;P~
k = k*r; // now K is the feedback level based on the Resonance param

for the diode ladder, it's k/17.0, in any case, this constant is supposed to be the same thing that the approximation for K gives at 0Hz
i apply this to the filter simply at the output like this:
yy = y4*rgc;
out = yy;

now, a few more things arised after i did this
i noticed that the level of my nonlinearity was wrong too
i had to adjust it to make it "seem" equal at all frequencies
so i did:

Code: Select all

fb = x0 - k*y4;
fb = shape(fb*rgc)/rgc;

and it looks much better now
but i begin to notice that it can be done even better
i'm thinking, making K 0.0 to 17.0 (or 4.0) at all times, and only compensating the gain on "yy" .. then taking the feedback from "yy" instead of y4
i think this would be equivalent of doing k*y4 where "k" has compensation
so now it will be more like r*17.0*yy <- and yy=y4*rgc

PS: sorry for bumping the thread, i didn't want to start a new one, and this thing is related
PS^2: if anyone is reading on this topic, i have been silent for a long time which means i have accumulated lots of new clues

[andrea993]

I'm improving open303, I have improved the patterns and now I'm working to a GUI.
I've develop an open303 application that works alone without using vst. The GUI is developed in Qt (so it's multi-platform) and make a porting a Qt gui to VST is easy.

In the application the audio is managed by Jack in order to have a low latency.

And now open303 can also import and export the patterns in ABL2 file format.

[xoxos]

casual enquiry as i haven't researched the matter if anyone knows - what do the pitch glides look like on the tb-303? i'm guessing the method may not yield a linear ^2 response however it was accomplished..?