KVR Audio

antto · Post by **antto** » Sat Sep 12, 2009 5:01 pm

Robin from www.rs-met.com wrote:O.K. but this two-stage decay amp-env is also in good agreeement with ABL. could you upload your reference samples somewhere? or point to the original sources?

will try to, there is some stupid restriction to not upload files bigger than 20 or 25mb (not sure) ..uhm, in box.net

i think, when fully open with zero resonance, the filter should not mess up the amp-env too much. i know there's the non-zero env-mod even when the dial is counterclockwise (unless one has a mod that allows to set it to true zero, like devilfish), but that should then only affect the transient - the rest of the envelope (the loong exponential decay) should still be visible.

if you look at the image i posted (which you have to click first, damn, no direct URL) you'll see that when the filter is "fully open" cutoff will fall down to 711Hz, which isn't very high

(which is far from a square)
maybe look at the filter thread, i posted a .rar file with a comparison of my synth with a sample of the TB-303
blah.. here it is: http://www.box.net/shared/bdf8v582cv
aha thanks for that - i suppose the first part is the real thing there?

the left channel is my synth, right channel is the TB-303 audio which can be found on the ladyada website (IIRC)

rlahalla · Post by **rlahalla** » Sat Sep 12, 2009 5:03 pm

antto wrote:will try to, there is some stupid restriction to not upload files bigger than 20 or 25mb (not sure) ..uhm, in box.net

If you want some webspace, just let me know and I'll set you up and account for free.

antto · Post by **antto** » Sat Sep 12, 2009 5:12 pm

mistertoast wrote:What is "gate" on the TB-303, besides an output from the box?
Also, antto, what notes can be produced from a standard TB-303? How many octaves? Which MIDI notes?

not sure if i understand the first question, it's probably the "Gate" button on my sequencer (which determines if the current step is ON or OFF) or the Gate signal inside the synth
blah

well, the second question is kind of tricky
first of all, the TB-303 is not tuned well (i mean the osc) .. this is why there is a big TUNE knob, so you can adjust it to sound equal to your other (crappy old) analog gear ;]

the sequencer usualy has the range of 3 octaves and a C on a 4th octave (like from C1 to C5)
but there is also transpose per track which is kind of pretty Japan stuff ;]
gotta look at the TB-303 docs for that

if matched to modern software synths, the most popular software emulations agree that the "root" octave of the sequencer is C2
when the step is "Down" it goes C1, and if it's "Up" it goes C3
but the synth has a high C key on it's panel, so you can actually have a C4 in your pattern

btw, my sequencer has one octave bonus, but actually there is a trick and you can extend a pattern to use like all 10 octaves (which is probably useless)
also, i think i didn't restrict the FreeBee patterns importer so you can load a FreeBee pattern which uses anything from octave 0 to octave 9 ;]

mistertoast · Post by **mistertoast** » Sat Sep 12, 2009 5:34 pm

antto, you used the word "gate" in some of your descriptions of how slide worked. Just wondered what you were talking about. Is that your internal variable that keeps track of whether the note should trigger the envelopes or something?

antto · Post by **antto** » Sat Sep 12, 2009 6:47 pm

oh, that..
in that manner - yes
but, it's actually more complicated
first, there is "gate" variable in the sequencer (in the patterns i mean, per each step)
then, there is "gate" output of the sequencer, that triggers the envelopes
and another "gate" is the one that opens/closes the VCA (the thing that makes the gaps between steps)

antto · Post by **antto** » Sat Sep 12, 2009 10:00 pm

damn, whole day i spent in trying to get the filter envelope curve
darn, it turns out it's pretty shitty, it's not a simple exponent, not a power of something, i even tryied to do something like 1/sqrt(x*something) but it failed too
finaly, i made a high-resolution screenshot of the FFT and used some math to convert from pixels (screen coordinates) to values which i fed to CurveExpert and the result was a Logistic model, nailed it ;]
i am lucky that my FFT has linear scale (of the frequencies)
it also turned out that with the curve i approximated, the 341ms (minimum) decay time looks 99.9% accurate compared to the audio ;]
so, that's a multikill ;P~
i'm still going to seek any mistakes since i am again speaking a little too early, but it looks i'm correct

mistertoast · Post by **mistertoast** » Sat Sep 12, 2009 10:25 pm

I assume by "filter envelope curve," you mean how the cutoff changes through time.

antto · Post by **antto** » Sat Sep 12, 2009 10:33 pm

yeah, since it only has a Decay (as opposed to the wide spread ADSR) there is only 1 curve for it ;]

my previous curve was bad, i actually didn't spent enough time to look closely at the FFT, i just matched the sharp edge in the begining (Phoscyon seems to do the same, which is um, wrong too, tho it might not be "wrong" if we speak of a devilfish mod)

anyway, the exact curve is not only important to the sound, but to the gimmick circuit, since in the analog synth, the gimmick is based on some magic coming out of the filter envelope (so the exact shape is vital)

mistertoast · Post by **mistertoast** » Sat Sep 12, 2009 10:42 pm

And the gimmick is used on the accented notes to make them different from the unaccented notes, right?

mistertoast · Post by **mistertoast** » Sat Sep 12, 2009 10:48 pm

So what did you do to catch the shape of the curve through time? FFT snapshots with the resonance turned up?

antto · Post by **antto** » Sat Sep 12, 2009 11:00 pm

um, no
accented notes (notes programmed with the sequencer having the Accent button pressed) are already different from normal ones
decay time on accented notes is fixed and you can't control it (there is no such knob or anything)
the decay time itself is the same as the decay time of normal notes when the decay knob is at 0%
so this means that when decay knob is set to 0% and Accent knob is 0% too, there is no difference between normal and accented notes, and that's where all the fun begins, when you start tweaking the thing

anyway, the first difference is that when decay is longer, accented notes remain with the short decay time and can be spotted

the whole fun in the accent system is when you actually increase the Accent knob
first of all, accented notes will be louder now (punchy) this is easy
then comes the "gimmick" effect
it's a circuit that modifies the filter modulation (it adds to it) creating a very odd curve
the DevilFish docs (which Robin is reading) are accurate about it in most aspects, there is also some ASCII art there showing how it'll look on a FFT

to my ears, the synth starts "barking" or something
it's like in some genres of music, when the vocalist (actually not singing) is repeating a short part of a sentance or something, and each time he changes his accent more and more (and you like it so much, you wanna jump out the window) ;P~

antto · Post by **antto** » Sat Sep 12, 2009 11:14 pm

mistertoast wrote:So what did you do to catch the shape of the curve through time? FFT snapshots with the resonance turned up?

yes, i tryied looking at the shape on the FFT, and trying to play with numbers then compile/run/test to see how both the synth and the original audio sample look on a stereo FFT (over the other)
what ever i did, it was close, but not exactly, sometimes altho it was looking so similar, it sounded fake and it was so obvious i think even a no-303-specialist could tell it's fake ;]
hm.. now it looks right and sounds right ;]
i was too lazy to convert the screenshot of the FFT into curve data (to import into CurveExpert) but i wasted the whole day trying to be a hero ;]
i finaly did the dirty work in about 15 minutes and there it is, perfect

it also seems that my mesurements of the envelope low/high peaks at different knob settings is quite accurate against the original audio file (now i look at both my synth and the audio sample, on the same FFT, so the reading on the FFT might be wrong, but if my mesurements were right - both curves will be be moving in sync)
the only thing i'm not very sure about is the 70Hz in case4, i still can't tell by the FFT
will do a waveform test for that

btw, i just found out a nice distortion plugin, i think was meant for guitars (it grabbed my attention from the KVR-random-plugin-of-the-hour thing)
TubeBaby..

mistertoast · Post by **mistertoast** » Sat Sep 12, 2009 11:26 pm

Hmm. So you had to do multiple FFTs through time to get the peak at different times, right? Still not sure how you pulled that off. What was your procedure? Automated screenshots?

antto · Post by **antto** » Sat Sep 12, 2009 11:39 pm

naah, the curve is the same no matter what the Decay time is
it's like a shaper

mystran · Post by **mystran** » Sun Sep 13, 2009 1:17 am

Less arguing, more dancing (neither fish nor hardware were harmed making the above clip).

Open303 - open source 303 emulation project - collaborators wanted