Open303 - open source 303 emulation project - collaborators wanted

[antto]

thanks fliffo, unfortunately, i need a stereo sonogram
the best thing for me would be something like this one, but displaying both Left/Right channels one over the other (Red/Blue additive blended)
tho, this will still be useful (at least it's big)

[antto]

http://www.box.net/shared/qvjtbzmp3i (image) this is what i did so far, it's only the acc/decay envelopes, knob values are correct (tho, the sonogram was taken over a "warmed-up" loop, otherwise it would match
interesting, in this script, the Fs (sampling rate) is 639.80464Hz (muhahaha)
now i can start experimenting with the gimmick here, phew..
EDIT: now the tricky thing about this mIRC script is that i'm gonna "plot" (and process) two versions of the sonogram
it's from the sample "accenttest.ogg" where the same pattern is played at a very low and a normal (kinda fast) tempo with the same settings
i'll take two snapshots of the sonogram of a warmed-up loop, from each version (low/fast tempo) and process both versions with my envelopes as i tweak the thing in the script
this would otherwise require a huge amount of time if i did it as i did till now (in the daw, recompile/rerun/reopen/test/export)
the low-tempo version shows the exact "shape" of the accented decay being smoothened, while the faster one also shows the "acceleration"
if i match both (with the same "circuit") then mission is complete
tho, this whole thing is probably "cheating" because i have no idea how is the gimmick circuit done in the real thing (my LP filters fail when envmod is increased, so it must be done in another way, i'll worry about this later)

[rv0]

Anto (re: previous page)
I checked but I don't think I made any mistakes in those samples.
I think I remember the decay is just not that responsive on these settings.

If you want more samples just ask
I'll be recording the following for Robin, I rather record all of them at once

Code: Select all

ROBIN:
waveform: saw, 
reso: max, envmod: min, decay: min, accent: min 
cutoff should go through the 11 tick-marks on the panel. then for each tick-mark, play a long and low note (the lowest possible, so we can observe the frequency response at a dense spacing of the harmonics). 

- a sample like above but with cutoff, envmod and decay at max and another
- one with cutoff at min to see it in the range below 300 Hz

So lemme know the requests

[antto]

ok, so..

Code: Select all

G G G G
G O G G
G G G G
G G G G

-- -- -- --
-S -- A- -S
A- A- -- SA
A- -- SA --

damn, i'm so in a hurry, this is probably wrong
16 steps pattern, all "c" keys
Tune: mid
resonance = max
cutoff = min
envmod = max
accent = min
decay = max
waveform: saw

while recording

Code: Select all

envmod > min
accent > max
cutoff > max
decay > min
envmod > max
decay > max
cutoff > min

let the synth loop around between each knob tweak
tweak the knobs fast, so i can clearly see where you stopped moving the knobs

because of the envmod+cutoff=max this should be in 88200Hz, tho, i will be OK with a 44100Hz sample too

record this sample two times, at a tempo around 140 (fast one) and ~50 (slow)
thanks

[antto]

ok, got back from work
rv0: another request, very simple
i'm not sure about how a A/D convertor works (especialy yours) at different sampling rates?!
so just a very simple test

Code: Select all

G G G G G G
c(D) d#(DS) c c(S) C C(U)

just 6 steps..
tempo: something around 70
init:

Code: Select all

reso = min
cutoff,envmod,decay = max
tune = <middle>
waveform = saw

while recording:

Code: Select all

waveform > square
reso > max
waveform > saw
reso > min

again, move the knobs fast, and let the thing loop a little between each part
record this twice, at 44100 and 88200

i'm hoping that the HP filter inside the A/D converter is NOT changing in anyway depending on the sampling rate.. but i'm not sure

i am also interested in measuring the knob responses (mainly cutoff, envmod and decay) at each of the "tick marks" but that would be a hard thing to record i guess (time consuming, and you'll probably need a magnifying glass?)
tho, if you like to do it - tell me ;]

[antto]

i finaly wrote the damn script, got two "versions" of the sonogram, processing the envelopes for each, and filters working
this is how it looks with a single lowpass filter at 3Hz (it's not close at all, but you can see how it kinda works)
http://www.box.net/shared/fqye5ii3r6 < image
i'll be messing around with this now

[deraudrl]

...lowpass filter at 3Hz...

I sure hope that's a typo...

[antto]

?! why?
erm, that the normal 303 envelopes, the only "special" thing going on is that on accented notes, the envelope signal is _also_ sent thru this LP filter and the output is combined

[deraudrl]

?! why?
erm, that the normal 303 envelopes, the only "special" thing going on is that on accented notes, the envelope signal is _also_ sent thru this LP filter and the output is combined

Oh, it's filtering the envelope...
never mind.

[antto]

a little report: the step length of the TB-303 is not exactly 0.5 but more like 0.518 or 0.519
EDIT: 0.525

[Mike Janney]

50 pages and total stop? Come on guys, we need the competition.
Mike
AudioRealism

[antto]

hey Mike!
naah, at least i haven't stopped..
i've just been a little quiet lately, waiting for samples, guess we gotta send rv0 a private message, he's probably busy..

i threw away my old envelopes section, redid everything inside a single module, that includes the filter envelope, the gimmick, and the volume envelope
i now use RC-Decay everywhere (except for the filter decay, which is still linear + "my curve" ..)
the gimmick is better now, tho, the "circuit" is kinda nasty, but probably similar to the 303, i've only used 1 pole LP filters and RC Decays
gonna post another audio comparison demo, VB-303 vs the x0xb0x audio sample
just need to finish the volume envelope circuit a little

i also use 1 pole LP filters to smoothen my gate signals
but my sequencer showed some strange behaviour now with the new gimmick circuit, i'll either have to fix it or rewrite the whole thing
now since the envelopes are done inside one single module, a lot of "CV signals" are no longer needed

PS: freebee patterns with comments are now correctly handled, thanks ;]

EDIT: i investigated the sequencer bug (accent on the first step didn't work when the pattern is played the first time, no trigger)
i couldn't figure what was wrong (you know, the code was a complete mess) and got a little mad, fixed it like this: CTRL+A, DEL ;]
now since my envelopes needed only 2 CV signals from the sequencer - Gate and Accent, and the pitch (slides) are handled inside the sequencer itself, i no longer need to push my brain, so i just wrote about 20 lines and it worked
it doesn't look into the next step anymore (that was insane too) it only remembers the previous step, because of the slides, and the looping (this feature is not in the original 303 anyway)
now it'll also use less CPU ;]

btw, i added experimental Shuffle, it desyncs when the value is changed, but it works otherwise (i got an idea how to fix the desync)
it's very funny ;]

[antto]

x0x_pat2_cut01_compare08_2.rar (6.1MB / 13.3MB wav)
there, Left = Venom, with the latest modifications ;]
Right = TB-303 from ladyada
audio rendered at 88200, (externaly) downsampled to 44100Hz

i'm kinda happy ;]
still need to tweak this new gimmick circuit, but it's close, especialy with full resonance (or at least it doesn't act too silly there)
gonna post a demo of the shuffle probably tomorrow ;]

[antto]

http://www.box.net/shared/s9ftqsjnjr <- shuffle ;] (it's working now)

[jobromedia]

http://www.box.net/shared/s9ftqsjnjr <- shuffle ;] (it's working now)

Man! This is plain and simply godlike!