Repro-1 (out now)

[urosh]

So why didn’t 2 and 5 lead the poll?

Maybe because people actually prefer what they hear from more "expensive" models?
BTW, it's not difficult to figure out which model is which without listening at all, Urs gave one big clue early on.

[PAK]

Maybe because people actually prefer what they hear from more "expensive" models?
BTW, it's not difficult to figure out which model is which without listening at all, Urs gave one big clue early on.

So what do you feel you're getting for that expense then? What do your ears tell you it's giving you?

Perhaps the biggest clue given was the Pro 1 sweep posted. Which one do you feel copies that best and why?

[urosh]

So what do you feel you're getting for that expense then? What do your ears tell you it's giving you?

Well, for a start my personal choice (for practical use) is not most expensive/accurate model, but rather one of "mystran" models, probably one without corrective step.

Perhaps the biggest clue given was the Pro 1 sweep posted.

As far as telling which is which, critical clue about 2 and 5 was one simple little peace of info Urs gave. And you need pair of eyes, not pair of ears to figure out. But that is not the question, because this was not competition to figure out most expensive models, question was which of the filters "sound most analog".

[hakey]

So why didn’t 2 and 5 lead the poll?

I think it's that 2 and 5 exhibit muted responses to audiorate mod at high cutoff. It's as though the audiorate signal is low passed to avoid aliasing (though this is probably too crude an explanation to be what's really going on).

But perhaps that's how a Pro One behaves and 2 or 5 is the most accurate model?

[Urs]

it's like magic when someone finds a way to trick the equations like that.

Sir Isaac Newton 25 December 1642 – 20 March 1726/27

However, some of the stuff was already done by ancient folks, I think Babylonians.

I think what's happened in DSP is that people have waved the idea of doing it in favour of a delay element, and "what difference can a delay of 1/44100 second make anyway?". Well, it can make a filter explode for starters.

[beely]

Well, it can make a filter explode for starters.

Heh - you can go further than just modelling different tolerances between models - you can add in random component failures until your copy goes in for a "service" (for a modest charge! )...

[PAK]

Well, for a start my personal choice (for practical use) is not most expensive/accurate model, but rather one of "mystran" models, probably one without corrective step.

Ok. But what do you think the expense is giving you? I was hoping you’d have a go at describing what you hear?

As far as telling which is which, critical clue about 2 and 5 was one simple little peace of info Urs gave. And you need pair of eyes, not pair of ears to figure out. But that is not the question

Correct. The question was which one do you think best copies the filter sweep audio Urs posted, and why you think that?

because this was not competition to figure out most expensive models, question was which of the filters "sound most analog”.

Right, which is why I was asking you which you thought copies his own analogue synth best, since it's the main subject of the emulation after all.

If you don’t want to answer it’s fine. It’s just a shame so many get reluctant when asked to describe what they hear

[fmr]

So why didn’t 2 and 5 lead the poll?

I think it's that 2 and 5 exhibit muted responses to audiorate mod at high cutoff. It's as though the audiorate signal is low passed to avoid aliasing (though this is probably too crude an explanation to be what's really going on).

But perhaps that's how a Pro One behaves and 2 or 5 is the most accurate model?

I can't talk about the behaviour of the Pro-One since I never had one, but I have several analogue synths, and have worked with others I don't own too and, IMO that's the way ANY filter behaves, at least for what I know (I'm not claiming anything else than what is my experience, and, if I understand correctly, you are talking about filter too close). And certainly they do not behave like what 1 and 4 show when we use high resonance values, and that's why I was surprised to see those in the lead. For me, any filter that shows artifacts when resonating is always to be excluded, no matter what qualities it may have otherwise - I am looking at the whole picture, not just at some behaviour in some circumstances.

[PAK]

2 and 5 exhibit muted responses to audiorate mod at high cutoff. It's as though the audiorate signal is low passed to avoid aliasing (though this is probably too crude an explanation to be what's really going on).

Maybe, but it's the same sort of assumption I had.. sooo..

But perhaps that's how a Pro One behaves and 2 or 5 is the most accurate model?

In terms of artifacting, maybe. I would doubt it in terms of the lack of higher end. My own instinct is to want the more open quality of 1 and 4 married to the lack of artifacting in 2 and 5. 3 doesn't really do it, but I guess it's the closest to being in the middle, which is kinda why I'd maybe go with that if I was forced to pick one

[hakey]

IMO that's the way ANY filter behaves

But audiorate modulation of a resonating filter or, for that matter, any oscillator should be audible (as side bands).

At high cutoffs for filters 2 and 5, audiorate modulation has no discernible effect.

[Urs]

Well, it can make a filter explode for starters.

Heh - you can go further than just modelling different tolerances between models - you can add in random component failures until your copy goes in for a "service" (for a modest charge! )...

LOL, yes, this has become a running gag at u-he, possibly at many developers when it comes to defining the specs of a vintage hardware emulation. I would *love* to do something like it, say, where the user needs to press the "re-calibrate" button every once in a while. However, I'm afraid we would have to hire someone to deal with "bug reports" if we do so, and thus we don't

[urosh]

Ok. But what do you think the expense is giving you? I was hoping you’d have a go at describing what you hear?

In RePro topic on u-he subforum you have some. And then further more here, my user name on GS is recnsci. This is not all, but I think that long descriptions in vague terms (which is any describing timbre of sound) are boring for other folks.

[Urs]

IMO that's the way ANY filter behaves

But audiorate modulation of a resonating filter or, for that matter, any oscillator should be audible (as side bands).

At high cutoffs for filters 2 and 5, audiorate modulation has no discernible effect.

Yes, that's due to unit delays. They "lowpass" the filter's behaviour towards modulation/distortion for lack of a better word.

[Urs]

I'm slowly making my way through things. It'll all be explained, but maybe I have to keep the resolution of this experiment a bit shorter than I would like, otherwise people won't read it.

The riddle will be solved shortly.

[fmr]

I'm slowly making my way through things. It'll all be explained, but maybe I have to keep the resolution of this experiment a bit shorter than I would like, otherwise people won't read it.

The riddle will be solved shortly.

I'd suggest you post a shorter version, and create a PDF with the longer one, for people which may be interested to download. I am interested in the longer version, but I can understand some people won't