Open303 - open source 303 emulation project - collaborators wanted

[mistertoast]

in TB-303 the (gated) step length is either 0.5 or 1.5 (slided) or 2.5 (two slides) etc..
not sure about some rumors say about some "triplet mode" or similar thing, i really don't have clues/evidence of it

i'm talking about the normal (wide-spread) sequencing mode ;]

I don't remember hearing triplet stuff concerning the 303, I thought that was just the drum machines. I think some of the Roland drum machines do 48 slots per bar.

[mistertoast]

I'm in the camp that there has to be a sequencer to sound right (to be able to look ahead for the slide) but it should not be a recreation of the TB-303 one (which I think is appallingly painful).

[Music Engineer]

I'm in the camp that there has to be a sequencer to sound right (to be able to look ahead for the slide)

look ahead! good point. although it is possible to position the slide-target notes a bit early in a standard host piano roll, this would be tedious and probably not all users would be aware of it (or just too lazy - i'm guilty of that myself). does someone know how much the 303 actually looks ahead? the devil-fish doc says that the slide time is 60 ms in the standard 303 - would that also be the look ahead time then?

[mistertoast]

I'm in the camp that there has to be a sequencer to sound right (to be able to look ahead for the slide)

look ahead! good point. although it is possible to position the slide-target notes a bit early in a standard host piano roll, this would be tedious and probably not all users would be aware of it (or just too lazy - i'm guilty of that myself). does someone know how much the 303 actually looks ahead? the devil-fish doc says that the slide time is 60 ms in the standard 303 - would that also be the look ahead time then?

I'll bet you anything antto's already figured that out.

[mistertoast]

Hey Robin, do you have a link to that devilfish documentation you're reading?

Is this the one you're reading?

http://www.firstpr.com.au/rwi/dfish/release.html

"The Slide Time pot. Normally the slide time is 60 ms (milliseconds). In the Devil Fish, the Slide Time pot varies the time from 60 to 360 ms, when running from the internal sequencer. When running from an external CV, the time is between 2 and 300 ms."

[antto]

well, IMO, the slide time is relative to the tempo, not some miliseconds constant
otherwise slides would sound like a bad singer ;]
the lookahead is probably 1 step
in my implementation, it's a pretty big mess to get the same effects, so it looks into the next step, the current step, and the previous step (grrrr)
the original might work simpler, looking into only the current and next steps, not sure

anyway, the slide is pretty odd, the DevilFish docs only confused me, it was different than what i had to do to get the slides right

when step01 has no slide, it's length is 0.5
if it has a slide, but step02 has no GATE, the length of step01 is 1.0 (and continues on step02 with length 0.5, so it's 1.5 overall)
..but the pitch is constant (of step01, step02's pitch is ignored)

when step02 HAS gate, length stuff is the same, only difference is that the pitch changes (it slides)
the way it slides is somewhat interesting
the pitch starts changing right from the begining of step02 (which has only 0.5 time to slide to the pitch of step02)
the curve of this pitch change is not linear, it looks like a punchy exponent (starts fast, it does look like you've fed a step voltage to a 1pole LP filter)
..but i think the time changes with the tempo, someone has to test this ;]

[antto]

um, and things get more nasty when you got slide on step02 too (and all three steps are with different pitch, this is one critical spot where bad emulation might fail just by hearing it with a naked ear)

things get even more complicated when you add the Accent and envelopes to all of this
so trying to do what the sequencer does to make the synth do acid, in a piano roll of a normal DAW, is like um.. don't waste your time and brain-cells ;]

Robin: even if you nailed the sound of all the aspects of the TB-303 (osc, filter, envelopes, everything) without a proper "driver" you would still believe that it's missing something, while if you add the sequencer, you will hear a huge difference instantly (this is what happened with me)

[antto]

respect to the guy who made the DevilFish mod, but there is some wrong/incomplete information there in his docs IMO

this is true for almost all other "docs" and "articles" i've read about the TB-303 that was thinking are real, using them as a blind guide only got me into wrong directions
i figured they are wrong only when i started analyzing audio samples

[mystran]

well, IMO, the slide time is relative to the tempo, not some miliseconds constant
otherwise slides would sound like a bad singer ;]
the lookahead is probably 1 step

You don't need any lookahead. The slide is set to a note that should be tied to the next one, and that's all the lookahead you need already.. you just need to keep the gate up, and then trigger a slide on the next note.. consider a 2 step sequence:

1: C+slide
2: D

Now, sequencer outputs CV for C, notices that there is a slide, so it keeps the gate up, and then triggers the slide during the next note (D in this example), so during step 2 the CV to the oscillator will slide from C to D.

You can also do the same perfectly well using external sequencer and overlapping notes (because the external sequencer then controls the gate directly, and you just need to trigger a slide when the gate was already high when a note-on is received).

I agree with you though that tempo-synced slides are cool. I tried that last night and it's lovely.

[brambos]

You don't need any lookahead. The slide is set to a note that should be tied to the next one, and that's all the lookahead you need already.. you just need to keep the gate up, and then trigger a slide on the next note.. consider a 2 step sequence:

1: C+slide
2: D

Now, sequencer outputs CV for C, notices that there is a slide, so it keeps the gate up, and then triggers the slide during the next note (D in this example), so during step 2 the CV to the oscillator will slide from C to D.

Does the slide also affect the envelope behavior? Or does the decay kick in as usual and closes the filter during the slide?

[mystran]

Does the slide also affect the envelope behavior? Or does the decay kick in as usual and closes the filter during the slide?

There doesn't seem to be a separate trigger, but I actually I just realized I have no idea whether the sequencer allows you to do a slide with holding gate.. though I'm under the impression that slide will also cause a hold, thereby not retriggering the envelopes.. curious thing is, TB sequencer (or at least the user interface) does have a separate hold function..

edit: actually I guess it's just a user interface feature no(?) clone (with more sane programming) seems to have a way of inputting separate holds forcing you to use the slide for longer notes... what I wonder though is what happens when you program an accent on a slide note (that is, a step immediately following a slide)

edit2: seems Revolution actually supports sustained notes separately from glides.. but since they don't support sustained notes wrapping around pattern boundary, they suggest you use a glide on the last note (with the first note having the same pitch) instead...

[autodafe]

Hi Robin and Everyone:

here are some samples of my TB303:

www.autodafe.net/tmp/TB303_samples.zip

This zip contains several wav fies.
My TB303 is Midified with a Kenton retrofit
so, same MIDI sequence (just C-F notes on four octaves)

They are named along this scheme:

"TB303_saw_C100_R0_E0_D0_A100.wav"

It's pretty easy but let me explain this scheme..

TB303_is...well, you know...

_saw_ or _square_ is the basic waveform

_C100_ Means Cutoff 100% (Fully open)

_R0_ or _R100_ is the resonance (0 or 100%)

And similarly, E is the EnvMod, D is Decay and A is Accent (always, 0 or 100%)


I am not completely sure about the accented/unaccented notes...I have an external input for accents and don't know if the knob is affected or not (I cannot hear big dfferences between notes with accent 100% and 0% honestly)

[phankiejankie]

Thumbs up for this project, spent some time with the plug, loved the control you have on the accents even at this very early stages of development. You have my moral support and keep up the good work .

[mistertoast]

Hi Robin and Everyone:

here are some samples of my TB303:

www.autodafe.net/tmp/TB303_samples.zip

This zip contains several wav fies.
My TB303 is Midified with a Kenton retrofit
so, same MIDI sequence (just C-F notes on four octaves)

They are named along this scheme:

"TB303_saw_C100_R0_E0_D0_A100.wav"

It's pretty easy but let me explain this scheme..

TB303_is...well, you know...

_saw_ or _square_ is the basic waveform

_C100_ Means Cutoff 100% (Fully open)

_R0_ or _R100_ is the resonance (0 or 100%)

And similarly, E is the EnvMod, D is Decay and A is Accent (always, 0 or 100%)


I am not completely sure about the accented/unaccented notes...I have an external input for accents and don't know if the knob is affected or not (I cannot hear big dfferences between notes with accent 100% and 0% honestly)

Cool. I'd like to see the difference between this midi driven audio and a similar pattern-driven audio to see what the slide is doing.

I'm intrigued by antto's assertion that the slide to the next note begins immediately. I really want to know what that curve to the next note looks like.

[Music Engineer]

Hi Robin and Everyone:

here are some samples of my TB303:

www.autodafe.net/tmp/TB303_samples.zip

much thanks for the samples. i have investigated them briefly now and will do more thoroughly in the course of this afternoon. what i can say so far is that the amplitude envelope has apparently two stages of decay - something i already observed in the ABL - i currently model this by adding a fast exponential decay (created by a leaky integrator) to the main amp-env. i will now tweak this further to match your samples.

I am not completely sure about the accented/unaccented notes...I have an external input for accents and don't know if the knob is affected or not (I cannot hear big dfferences between notes with accent 100% and 0% honestly)

this is indeed strange - they sound (and also look) indeed almost identical. this is in contrast with ABL where there seems to be an even faster decay added for accented notes (which looks linear) such that the envelope essentialy becomes a 3-stage decay. mmmhhh.