Repro-1 (out now)

[spunkmuffin]

edit: actually both 1 AND 2 could be true. Now that would be clever!

What about 3.): Maybe you are just slightly paranoid in this case and Urs had not noticed this yet and just found it while he had some time to re-check the code...

Or 4. I'm having some fun

edit: Or 5. I'm paranoid AND having some fun.

[fluffy_little_something]

It must be a lot of code, written over months, maybe even including pieces of code taken from other synths.

[Ingonator]

As mentioned earlier i had alraedy bought Repro-1 and really love it so far. Besides creating some of my own patches from scratch i also had already re-programmed some Pro One patch sheets i found.

Besides that i also own Arturia Prophet V (current version is v3) since it was released 10 years ago and always liked the sounds i could get from this. With patches that missed some low end i used an EQ or a Bass enhancement plugin which worked quite nicely.
Anyway it was quite obvious to try also reprogramming some of the patches from this in Repro-1 so i started with the 40 original Prophet 5 presets included with Arturia Prophet V3.
Biggest problem there is the bigger Cutoff range of the Pro One filter so al filter related paramter s like e.g. Cutoff, envelope amount, filter sustain level etc. ned quite a bit of tweaking.
While i am not finished with the P5 presets yet i have re-programmed 8 of them so far and i am surpised how close sounding i could get both synths.
Of course i would have to exclude P5 presets that use Unison as this is not included in Repro-1 (while those maybe also sound good enough without it).

I know it is not 100% accurate to use a plugin as a reference but due to lacking a real Prophet 5 here it is diffuclt to 100% nail it to the sounds in the real thing. Anyway if i find proer audio demos or samples i will use them for fine tuning (i got a little experience with re-programming Minimoog patch sheets in The Legend and Monark). At the moment the patches in Arturia Prophet V seem to be a good starting point.
I will not invest a huge amount of time in this, also as i got other stuff to do (inclduing doing more of my own patches from scratch). When i decide to publish the patches here others could further optimize them too if they like.

UPDATE:
Like expected audio demos of the P5 factory presets are found at Synthmania:
http://www.synthmania.com/Prophet-5%20(Rev.3).htm

[Ah_Dziz]

In my "play around with repro 1" project in bigwig, I can run 15 repro1s before I run into trouble. That's on my laptop. I loose one instance in cubase. All of these have my standard mutonic, dramatic, battery, and kontakt loaded up along side. Seems reasonable for such an insanely accurate emulation. I'll probably be playing this thing live most of the time and recording the output due to the stiff CPU but it's manageable. This is on my laptop not my main workstation. Also with it being the beta I'll bet there will be all sorts of optimization between now and when the 1.2-3 versions come out. I was inches from buying diva, but this is cheaper and much more fun.

[wagtunes]

In my "play around with repro 1" project in bigwig, I can run 15 repro1s before I run into trouble. That's on my laptop. I loose one instance in cubase. All of these have my standard mutonic, dramatic, battery, and kontakt loaded up along side. Seems reasonable for such an insanely accurate emulation. I'll probably be playing this thing live most of the time and recording the output due to the stiff CPU but it's manageable. This is on my laptop not my main workstation. Also with it being the beta I'll bet there will be all sorts of optimization between now and when the 1.2-3 versions come out. I was inches from buying diva, but this is cheaper and much more fun.

Plus, in a project, once you have the sound just the way you want it, you can always just freeze the track. That's what I do with my CPU hog plugs.

[Vortifex]

..

[Urs]

edit: actually both 1 AND 2 could be true. Now that would be clever!

What about 3.): Maybe you are just slightly paranoid in this case and Urs had not noticed this yet and just found it while he had some time to re-check the code...

Or 4. I'm having some fun

edit: Or 5. I'm paranoid AND having some fun.

It's indeed so that I haven't touched the filter code since June? July? - I have only ever tweaked its additional 25 or so parameters to match a Pro-One. I was like "Oh, shit, bwahaha" when I saw the remnant trig functions in the code today. Unexpected, as I normally weed these things out quickly.

Anyhow, I've meanwhile written Jaws, Lyrebird, Resq and Sonic Conditioner from scratch, re-invented oversampling, re-imagined a spiffy interpolation method from Maple, designed, written and run a genetic algorithm for criteria-based filter design (part of oversampling also) and fixed a couple of bugs. On the side I also run a company. I also wrote a demo song and started a promo video.

So hehehe, I sometimes forget what exactly I coded a quarter earlier.

[xalama qo]

^^^ Love the way you're so "sh*t happens" about being human! A dev with a more fragile ego would've jumped thru hoops to hide their own fallibility. Classy

[Vortifex]

..

[ImNotDedYet]

Good news about CPU:

I just had a look into the filter code. It still uses a call to a pretty expensive math function, like once per stage, with 4 stages, 8 times oversampling and an average of 5 iterations... like 160 times per sample...?

That's an overhead which will be gone in a few days. I expect the filter code to fly from then. Half the CPU of the beta might very well be a possibility.

Hmm this sounds very fishy to me. I thought you coded the filter yourself and if you did surely you would have noticed this earlier.....hmm.

So it's either:

1. You left the expensive math function call in deliberately just so you could say half way through beta 'Oh I'm going to cut the CPU hit in half on release.'

or...

2. Somebody else has been coding the filter. Maybe some Russian maths genius that you have been outsoursing development to, and he has come up with a better solution just now.

Either way, well done!

edit: actually both 1 AND 2 could be true. Now that would be clever!

You've obviously never written software, particularly a large software project where it's quite easy to miss something like what Urs has described.

But in terms of the Repro-1. Very well done u-he team!

[gassle]

Good news about CPU:

I just had a look into the filter code. It still uses a call to a pretty expensive math function, like once per stage, with 4 stages, 8 times oversampling and an average of 5 iterations... like 160 times per sample...?

That's an overhead which will be gone in a few days. I expect the filter code to fly from then. Half the CPU of the beta might very well be a possibility.

Hmm this sounds very fishy to me. I thought you coded the filter yourself and if you did surely you would have noticed this earlier.....hmm.

So it's either:

1. You left the expensive math function call in deliberately just so you could say half way through beta 'Oh I'm going to cut the CPU hit in half on release.'

or...

2. Somebody else has been coding the filter. Maybe some Russian maths genius that you have been outsoursing development to, and he has come up with a better solution just now.

Either way, well done!

edit: actually both 1 AND 2 could be true. Now that would be clever!

Oh man... the internet people.

[e@rs]

to hear about possible halving of the CPU usage. Go, Urs!

[Ingonator]

While re-programming Prophet 5 presets from Arturia Prophet V3 i found another difference of the P5 vs Pro One besides the poyphony, unison and different Cutoff range.

The Prophet 5 in the "Wheel-Mod" section has a knob to morph between using the LFO and the Noise source for odulation comparable to the Minimoog that has a comparable knob (and using the pink noise is quite useful for modulation there).
In Repro-1 and the real Pro One there does not seem to be a way to use a noise source or S&H for modulation.

[Urs]

While re-programming Prophet 5 presets from Arturia Prophet V3 i found another difference of the P5 vs Pro One besides the poyphony, unison and different Cutoff range.

The Prophet 5 in the "Wheel-Mod" section has a knob to morph between using the LFO and the Noise source for odulation comparable to the Minimoog that has a comparable knob (and using the pink noise is quite useful for modulation there).
In Repro-1 and the real Pro One there does not seem to be a way to use a noise source or S&H for modulation.

Yep.

The biggest difference between Pro-One and Prophet-5 is the volume and polarity of the waveforms. You simply can't recreate any Prophet-5 patch on a Pro-One once any Osc has more than one waveform going, or Osc Sync between two different ones (take that with grain of salt).

Likewise I think the gains and polarity of the LFO might differ too. Haven't measured it yet. Any patch using LFO might be impossible to recreate among both.

Also... PWM might be positive on the P-5 while it's oddly inverted on Pro-One.

Filter range is the same btw. - its just the knob on the real Prophet-5 which is rastered in, hmmm, 64? steps. You need to add KeyTracking or Envelope to fully open Cutoff, but the it does open to 40+ kHz as well.

- U

[cake builder]

I can run 15 repro1s before I run into trouble.

How is this possible? I have an i7 6700k and can only run about 6 before my CPU is maxed out.

Edit: Hmm, it seems FL Studio isn't displaying the CPU load correctly. With 4 Repros open it says I have over 60% CPU load, but in Windows Resource Monitor it says 13%. So I don't know what's going on there.

FL studio has some issue with CPU usage. In FL , the default patch on diva uses 40% on my i7-4790k, but in reaper, it uses 3%.