Link: Venom VB-303 VSTi (Beta release)

[musikmachine]

That's already been debated. antto said he would work on it but only after he'll be pleased with the sound and behavior of the synth.

Give him some time

Ok thanks. I thought i might have an old version. So is this going to be freeware? If so that's mighty generous of the dev.

[musikmachine]

Wow this i really nice. Kudos to antto if it is free but would happily send some $ his way. Seems buggy in Live 7 though; i saved a project containing one instance of drumazon and venom and when i pressed play the cpu spiked to 120% and there was no audio. It was fine before i saved it although cpu use was a bit high.

[antto]

GUI will probably be the same size, tho i'll probably make better looking knobs (they'll still look like 303 knobs)
currently, i am working again (i mean a job) so i have like 6 free hours to mess with the synth (or rest from work) and then another 6 hours to sleep, and sunday is free..
i hate that..
GUI takes a lot of time, while i really want to improve the sound and fix the bugs, and add the stuff i wanted to add..

musikmachine: my initial idea was to make myself a 303-like synth that i can use for myself, but then, if it's good enough (if people don't laugh too much when i start talking about it and The Three-o-Three) why not share it
it is freeware, it will be freeware, i also want to show people that freeware is not "bad" SynthEdit creations neither

have no idea why it wouldn't work in Live7 after it worked..
hm..
the initial freeze is (probably) because it is reinitializing (calculating some dumb lookup tables, and sequencer re-reading the whole pattern memory)
try to reproduce the error:
load Live7, load the synth, either create a simple pattern, or load one from a file into a pattern slot (remember which slot it is)
then save the project, close Live7, and reopen it, reopen the project
see if the pattern is there and the synth responds (plays)
there isn't much to do about the CPU Usage at the moment
i could have optimized some of the code in the beta, but it's too late now
while now, i changed some parts of the sequencer, the envelopes, the distortion, and most of all - the filter
now i am affraid it could use more CPU (not sure, because i'm working with all the modules compiled without MMX/SSE, and i haven't optimized anything yet, since i'm still working on it)
but i think i don't give a damn about this, really
the synth is quite hard to immitate, while it's quite simple as an analog machine
the hard thing is that it sounds/works the way it does, because it's analog, and analog devices are NOT perfect, but trying to be, while software (DSP) devices are (sort of) perfect, because 1.0 is 1.0, while 1Volt will allways have some noise, and temperature will do it's thing, and what not
so i don't want to compromise CPU too much, i will do whatever needed (as long as i can) to make it sound closer/better at any cost
the Home PC is getting faster and faster as we talk, right? ;]

i'm still working on it, after i found out this ladder filter code (the one i called a 303-ish moog filter) i can now really compare my synth's output with a real audio sample of a TB-303 and see the similarities/differences
i'm working on some micro-details now levels of gain-loss, low-frequency-loss, exact PW of the "Square" ..
i would need customly recorded samples of a TB-303, i still don't have such

i'm also experimenting with another filter (a new one) since yesterday, kindly provided by kunn, and moddeled after the diode ladder
and it sounds like it!
maybe i'll include both, don't know
it needs some work..
also, i figured both this new filter, and my old 303-ish moog filter can actually work at only 2x oversampling without any (big) difference in sound
tho, 4x oversampling is still good for the distortion, but i guess i could add optional 2x or 4x oversampling switch
still got work with both filters, the gain-loss is (i think) incorrect.. this is how much the overall gain of the filter output drops down when resonance is turned from 0 to max, and is important when used with distortion especialy

[mistertoast]

Who is Kunn and how can I see his filter source code?

[thecontrolcentre]

Great to hear you're still working on Venom. I like it very much. Haven't noticed any issues with Live 8 as yet. Looking forward to the next release

[spacedad]

no sign of a no sse version yet?

[antto]

no sign of a no sse version yet?

well, i think i did rebuild the (beta) modules with no-SSE just to see if it works on my old machine, before i started working on the new version (what i'm doing now)
i don't remember, i guess i keep the DLL somewhere (since i no longer have the beta modules, they were overwritten (i got backups but it's gonna take too much time)
i could <unofficialy> release the no-SSE (beta) DLL if i find it, tho it's pretty bad on CPU Usage, not sure if you'll ever use it

the new version will have both SIMD (MMX&SSE) powered, and non-SIMD version (two DLLs in the zip)

thank you for still being interested in this, even that i am experiencing difficulties right now, i keep seeing that i am not wasting this part of my life right now

[spacedad]

not sure what Non-SIMD is,but don't worry about diging out old versions,i need a new computer soon really,so i'll look fwd to trying it then.

[antto]

MMX, SSE, SSE2, 3DNow! and similar "features" as far as i know are all called SIMD (Single Instruction - Multiple Destinations)
my old Duron 650MHz actually has MMX and 3DNow! (and Enhanced 3DNow!) but that doesn't help, since the modules were compiled for SSE

i'll try to find that DLL tomorow probably
i still advise all of you guys, play with the beta, but don't make serious music with it because it won't work with the new version
.. you'll have to keep the the beta DLL around when you want to edit your old projects (tho you could export the patterns/memory, and probably save the automation from the DAW to external files (like FLStudio can) and "port" your tracks to the new version, but it won't sound quite the same)

i think i'm gonna upload another audio sample showing all the new changes i made to the filter/envelopes lately (i was quite silent here lately, but was actually working on it)

to sum up things, now it sounds as close as i wouldn't dare to dream of before, i'm just trying to improve some microscopic details, tho it's kinda hard, since i really need specific audio samples to analyze/compare with
i still don't like one part of the Volume envelope tail .. which differes from a TB-303, it's accent-slided-to-normal vs accent-slided-to-accent, in the second case the TB-303 volume would Bump and continue to fall off even on the second note (which is accented) otherwise (first case) it would "cut" off the bump tail on the second note (when it's a normal note) my synth does this in both cases..

another thing that i am suspicious about again is the Gimmick, there is a little thing that i don't like about it (the curves) it's pretty small detail, but it differs, and it annoys me when i compare it to an audio sample
the gimmick accumulation (or range) i think isn't right too, i really need clean audio samples for this, i've modeled the accumulation against samples from acidvoice, but i'm not really sure how were the knobs turned.. i got other samples that it seems the accumulation is different..

i also think that the PW of the "square" changes depending on the Osc frequency (since it's actually not a Square wave oscillator, but some nasty waveshaper, as far as i know, a comparator is involved, and probably DC offset makes it assymetric, the raise of the comparator is i guess constant over time, which would make the PW smaller as Osc frequency raises)
i will probably try to immitate this (tho not in that way, but still with bandlimited sawtooths) and will probably add a switch for it (AutoPW - on/off)

[antto]

x0x_pat2_cut01_compare02_8bit.rar (3.1MB / 17.4MB wav)
i did this right now, TB-303 sample on one channel, Venom on the other (guess which is where)
i just dropped down the original audio sample in my DAW, then with automation clips on Venom matched the knob movement to mimic the original live performance.. blah blah
it was done in about hm.. 30 minutes

[musikmachine]

GUI will probably be the same size, tho i'll probably make better looking knobs (they'll still look like 303 knobs)
currently, i am working again (i mean a job) so i have like 6 free hours to mess with the synth (or rest from work) and then another 6 hours to sleep, and sunday is free..
i hate that..
GUI takes a lot of time, while i really want to improve the sound and fix the bugs, and add the stuff i wanted to add..

musikmachine: my initial idea was to make myself a 303-like synth that i can use for myself, but then, if it's good enough (if people don't laugh too much when i start talking about it and The Three-o-Three) why not share it
it is freeware, it will be freeware, i also want to show people that freeware is not "bad" SynthEdit creations neither

have no idea why it wouldn't work in Live7 after it worked..
hm..
the initial freeze is (probably) because it is reinitializing (calculating some dumb lookup tables, and sequencer re-reading the whole pattern memory)
try to reproduce the error:
load Live7, load the synth, either create a simple pattern, or load one from a file into a pattern slot (remember which slot it is)
then save the project, close Live7, and reopen it, reopen the project
see if the pattern is there and the synth responds (plays)
there isn't much to do about the CPU Usage at the moment
i could have optimized some of the code in the beta, but it's too late now
while now, i changed some parts of the sequencer, the envelopes, the distortion, and most of all - the filter
now i am affraid it could use more CPU (not sure, because i'm working with all the modules compiled without MMX/SSE, and i haven't optimized anything yet, since i'm still working on it)
but i think i don't give a damn about this, really
the synth is quite hard to immitate, while it's quite simple as an analog machine
the hard thing is that it sounds/works the way it does, because it's analog, and analog devices are NOT perfect, but trying to be, while software (DSP) devices are (sort of) perfect, because 1.0 is 1.0, while 1Volt will allways have some noise, and temperature will do it's thing, and what not
so i don't want to compromise CPU too much, i will do whatever needed (as long as i can) to make it sound closer/better at any cost
the Home PC is getting faster and faster as we talk, right? ;]

i'm still working on it, after i found out this ladder filter code (the one i called a 303-ish moog filter) i can now really compare my synth's output with a real audio sample of a TB-303 and see the similarities/differences
i'm working on some micro-details now levels of gain-loss, low-frequency-loss, exact PW of the "Square" ..
i would need customly recorded samples of a TB-303, i still don't have such

i'm also experimenting with another filter (a new one) since yesterday, kindly provided by kunn, and moddeled after the diode ladder
and it sounds like it!
maybe i'll include both, don't know
it needs some work..
also, i figured both this new filter, and my old 303-ish moog filter can actually work at only 2x oversampling without any (big) difference in sound
tho, 4x oversampling is still good for the distortion, but i guess i could add optional 2x or 4x oversampling switch
still got work with both filters, the gain-loss is (i think) incorrect.. this is how much the overall gain of the filter output drops down when resonance is turned from 0 to max, and is important when used with distortion especialy

Cheers antto. For free i'll take the quirks np.

Actually a smaller ui is not a bad thing as it takes up less screen space and it's not as bad eyestrainingly bad as i first thought. Also cpu is not too bad, about 6% in Live 8. I did try reloading the project afterwards and it was fine. It does cause a cpu spike when i drop into my project and also when i disable/reenable it in both Live 7 & 8. I should mention i'm on windows 7.

Anyway this a great sounding sounding bassline synth imo. Great stuff and i hope you develop more in the future.

[antto]

thanks for the understanding..
i pretty much hate big GUIs, a lot of space wasted
i wanted this to be compact, whatever monitor one has, he must be able to see each pixel, otherwise the resolution is too much
i switched to 1152x864 on my 17" Philips, i still don't think it's too small at all
a lot of the initialization time is taken by the oscillator, i think there is a way to be optimized a little..

btw, see/hear that wave file i posted, to have a rough idea of the improvements ;]
(and tell me which of the two you think is mine)

[mistertoast]

Do you build your osc tables at init time, antto?

[antto]

yeah? when else?

[mistertoast]

yeah? when else?

Yeah, if you use multiple tables oscs, you either build them or load them.

The other solution is to use blits or minbleps. Or just oversample like a mother.