Repro-1 (out now)

[hakey]

Maybe they are "better behaved" because the developer couldn't drive them into serious territory without risking very "un-analogue" artifacts.

Which makes me suspect all the more that 1 is closest to the real thing. But I'm just guessing. I don't have a Pro One to compare with.

Of course, you could give us some audio from your Pro One- a filter sweep at self oscillation with audio rate mod would probably settle it.

[penguinfromdeep]

At extreme settings the nastiness of number 1 stands out from others. 2, 3 and 5 are smooth and well behaved and rather bland. 4 is somewhere between 1 and 2/3/5.

On the basis that VA filters are often better behaved than the real thing, I'll guess 1 or 4 is the most accurate model.

And 1 is so obviously different- if that's close to how u-he's Pro One behaves (it should be obvious from ABing against the real thing), then surely that has to be the model that makes it to the final version.

For me the filter 1 just doesn't have the low end fatness and depth of a real analog (among other things) .. But who knows, maybe it is the real deal after all! It is rather wild. I've never tried pro one myself (though have played multiple occasions with prophet 5). I own multiple other analogs though.

[do_androids_dream]

At extreme settings the nastiness of number 1 stands out from others. 2, 3 and 5 are smooth and well behaved and rather bland. 4 is somewhere between 1 and 2/3/5.

On the basis that VA filters are often better behaved than the real thing, I'll guess 1 or 4 is the most accurate model.

And 1 is so obviously different- if that's close to how u-he's Pro One behaves (it should be obvious from ABing against the real thing), then surely that has to be the model that makes it to the final version.

For me the filter 1 just doesn't have the low end fatness and depth of a real analog (among other things) .. But who knows, maybe it is the real deal after all! It is rather wild. I've never tried pro one myself (though have played multiple occasions with prophet 5). I own multiple other analogs though.

Do you mean in comparison to the other filters or to a real analog filter? Of the five, 1 (and possibly 2) are the only ones that convince me of the 'feel' of a pro one-like filter. It's fat, a bit dirty, and maintains the low end weight right down to almost completely closed. 3,4 and 5 don't do this - the low end isn't there as much at all and they just don't 'feel' like I'm tweaking an analog filter. This test does reassure me that it can be done but it seems that we might have come so far from analog being the norm that a lot of (perhaps younger) folks prefer this flatness of a lot of digital filters/softsynths in general. In a way kvr might not be the right place to draw solid conclusions from seeing as it is very much a software orientated place.

[Ingonator]

Filter #1 at higher Resonance gives Noise and artifacts that IMO should not happen with real analogs or high quality ZDF filters. Same is also true at Self oscillation (without oscillators) at high Cutoff frequencies.

Some people maybe see this as anlog behavior while for me those are digital atrifacts like in many older plugins and/or older filter models.

FWIW at the momemnt i got those hardware synths with real analog filters:
Waldorf Pulse 2, Korg DW-8000, Ensoniq ESQ-1
The ESQ-1 got a Curtis filter (CEM 3379) , the DW-8000 a Korg NJM2069 and the Pulse 2 got a discrete multimode filter design by Waldorf (no one chip designs like the other two).

[Vesius]

Hmm.

This morning I had a few thoughts. If i were a betting man I'd say 2 has a cheaper way of controlling the internal feedback caused by the resonance and that is why it loses bass at extreme settings.

3 and 5 both seem to control the max rate the cutoff can modulate at ... I suspect to help with aliasing, and that is why they have this tidy round quality - although maybe were into AA filters, or other here too.

1 and 4 again I think are testing different ways of controlling the resonance circuit. To me it sounds like 1 is the cheaper method letting through more non desireable characteristics.

So, it's a good thing I'm not a betting man, coz I'm probably wrong.

[penguinfromdeep]

Do you mean in comparison to the other filters or to a real analog filter? Of the five, 1 (and possibly 2) are the only ones that convince me of the 'feel' of a pro one-like filter. It's fat, a bit dirty, and maintains the low end weight right down to almost completely closed. 3,4 and 5 don't do this - the low end isn't there as much at all and they just don't 'feel' like I'm tweaking an analog filter.

Yes, in comparison to other real analog filters. Its funny I hear almost exactly the opposite as you. To me first one doesn't have a big low end weight at all. It was really obvious also when someone posted a clip with Repro (with filter #1) and real Pro-One (presumably). The repro example sounded just lacking in low end in that example compared to the real mccoy.

[gnagna]

So far it seems that 1-3-4-5 have similar percentages, that leaves a lot to say about which one is most unique.

P.S. I voted 2

[hakey]

For me the filter 1 just doesn't have the low end fatness and depth of a real analog

In most of the blind Analogue vs VA tests I've tried, the real analogue invariably turns out to have the thinner, more brittle sound.

[olikana]

penguin is right. n.1 loses the low frequencies which is my major gripe with it.
it's clear when making a ressy synth bass. basically when you switch to 1 it feels like the bottom end disappears.
even if my overall favourite filter is n.5 ( smoother , warmer etc) an edgy piercing filter like n.1 is actually what I'm more interested in as I don't have anything like that in my arsenal at the moment.....but if the low frequencies remain lacking I'll give it a miss.
n.2 mantains more of the bass with a bit less edgyness but feels like a compromise (neither fish nor fowl).
i'm after a filter like the CS01mkII (edgy but still bassy) to complete what I already have. i feel n.1 gets a similar higher spectrum but misses the lows..
and even my far recollection of a pro1 is that of better low end than that offered by n.1

[PAK]

Some CEM synths (can't speak to the Pro One specifically) are well known for noise (aka "fizz"). So, in that respect, version 1 leans towards that most. But, yeah, when you push the res it starts generating what sounds like digital artifacts. Model 4 steps things back, avoiding more obvious artifacts..

The other obviously different one is model 5. It almost takes the opposite approach, smoothing things out, which will help certain sounds, but might also sound lifeless at times. 2 and 3 are literally in the middle (though more towards 5 rather than 4), and are the most difficult to separate.

BTW People didn't seem to notice that Arturia's Matrix 12V includes a more interesting full resonance model. But it only ever triggers at 100% (it functions more like an on/off switch than something where you gradually introduce and maybe pull interesting sounds out just on the edges).

Based on their attempts, and Urs words, unless it's an elaborate ruse (sticking more noise on one and then down-sampling? ) it could be that the reason why devs have generally avoided modeling full res behaviours on more noisy (characterful?) synths is because of the possibility of said artifacts.. Would also maybe explain why Arturia's is a bit limited in the way they implemented it.

[Urs]

Some CEM synths (can't speak to the Pro One specifically) are well known for noise (aka "fizz"). So, in that respect, version 1 leans towards that most. But, yeah, when you push the res it starts generating what sounds like digital artifacts. Model 4 steps things back, avoiding more obvious artifacts..

I think the fizzy sound may come from the fact that these filters open well beyond 40kHz. I could actually send my Pro-One filter to 68kHz with all modulations.

BTW People didn't seem to notice that Arturia's Matrix 12V includes a more interesting full resonance model. But it only ever triggers at 100% (it functions more like an on/off switch than something where you gradually introduce and maybe pull interesting sounds out just on the edges).

Based on their attempts, and Urs words, unless it's an elaborate ruse (sticking more noise on one and then down-sampling? ) it could be that the reason why devs have generally avoided modeling full res behaviours on more noisy (characterful?) synths is because of the possibility of said artifacts.. Would also maybe explain why Arturia's is a bit limited in the way they implemented it.

I haven't checked the Arturia but I booted up my Matrix 12. It jumps to self oscillation when switching resonance from 62 to 63, so it only does that on the highest possible setting. It's a very different chip though.

[penguinfromdeep]

In most of the blind Analogue vs VA tests I've tried, the real analogue invariably turns out to have the thinner, more brittle sound.

Of course, depends also on the synth in question. But for me telltale sign of difference between a lot of analogs and their software counterparts is in the low end, not necessarily more bass but just more weight and depth. But yeah depends ... For example when this guy (forgot his name) made the awesome OB-X soundset for Diva and made comparison to his real OB-X, actually I thought the Diva sounded fatter in lows.

[hakey]

Having second thoughts. 1 is just too nasty. Who knows, maybe that's what a Pro One sounds like ...

4, then.

(But aren't we being asked the wrong question - surely what matters for an emulation of a Pro One is which gets closest to the behaviour of a Pro One, not which conforms most closely to the generic idea of analogueness?)

[hakey]

For example when this guy (forgot his name) made the awesome OB-X soundset for Diva and made comparison to his real OB-X, actually I thought the Diva sounded fatter in lows.

I thought the same about the OP-X vs OB-X comparisons.

[bcm70]

For me it's 1 or 4 - 1 sounds a little wilder to me, but it also produces pops and crackles when sweeping the filter with the right settings. No idea if the original synth does the same thing.

5 is the worst, followed by 2 and then 3.

My methodology was to setup a looping arp pattern with only Osc A audible and to use Osc B at max frequency, octave, PW, the LFO set to normal and the modulation set to max. I then set the resonance to max and swept the cutoff whilst changing between filter models.

Filter models 1 and 4 both produced filter sweeps that sounded very much like an 80s video game. I think Filter 1 sounded slightly better except it would pop and crackle as the cutoff got to around 4 or 5 on the dial every time. 4 made pretty much the same sort of sound, but with no accompanying pops and crackles. It's not a computer issue as even with two RePros running along with a Diva and a stock kit from Live I'm only at 30% on the CPU meter.

Filter 5 lost all the bottom end very quickly - before I hit 1 on the cutoff.

Filter 2 was slightly better than 5 but still very thin sounding by around 1 on the cutoff.

Filter 3 sits between 2 and 5. It starts making the same sort of sound as 1 and 4 but it drops off very quickly.

I've voted 4, but would probably prefer 1 without the artifacts.

Testing complete - now time to play with it! It's great sounding as it is. Can't wait for it to be released.