UAD digital plugins vs KVR developers brightest minds!

[drinelli]

Goodnight where ever you are.

[kylen]

I wondered why they were the only ones to emulate the compressors so well. And it makes sense when you realize how the two brothers found their father's notebook with all the original schematics ( http://uaudio.com/company/index.html ). The secrets are in these docs.

Oh thanks Jeff! That was a fun & telling read - maybe we can get the actual notbooks under the Freedom of [audio] information act!

Maybe that's the whole thing right there - understanding the problems and how they were solved regardless of implementation, hardware or software. That's too over-simplified of course but I like the cold beer idea...so pop, ffssst!

[kylen]

Goodnight where ever you are.

Now it's off to the stars for you! Cool tips, Later...

[bmanic]

kylen, uploading file now.. check PM in 10 minutes.

- bManic

[kylen]

kylen, uploading file now.. check PM in 10 minutes.

- bManic

OK - will do - thanks!

[bmanic]

If somebody could explain in artistic,emotionalmanners
just HOW a compressor/limiter is supposed to sound
it would be lots easier to get good results.

I think the best way for a coder to really get under the hood of what a good hardware unit is, is to sit down on a mixing session with a good mix enginer (or mastering enginer) and let them explain everything during the mix.

When I hear a compressor I always immediately think of the produced sound in a mix context (or mastering context if that's what it's being marketed as), that is, I can already imagine how it will sound within the mix. What is usually of importance is how the transients and the 'body' of the sound come out of the box and how the overall sound is treated (added distortion, frequency loss etc).

What the hardware seems to gain over software is the 'resolution' of the compressor, that is, it's curve, the magic hand that controls the volume, has infinite 'levels'. There are no gaps in between two values which is why well designed hardware can do some very heavy gain reduction without sounding clumsy (I don't know how to describe this in words). Bad software compressors (like take the default cubase sx compressor) seem to jump from value to value.. like -30dB -20dB -5dB with no values in between whereas the hardware goes smoothly from -30 -> -5dB with infinite resolution or something.

What I'm trying to describe is NOT connected to the attack or release speed but rather the resolution of the action itself. Think of it as a spline envelope that is running at audio rate compared to one that is running at, say, 64 samples/second. One will be smooth and the other 'grainy/jumpy'.

Makes any sense at all? I'm no enginer nor coder so I might have it all totally wrong but this is how my ears hear it and why compressors tend to sound much better when they are analogue hardware.

Cheers!
bManic

[LiteOn]

Hi,

Wait the two reverbs in Uad1, is called Dreamverb, and what is the other...

RoomVerb (it's in every pack - DreamVerb is lacking from StudioPack). It's meant to be a room simulation - bad luck for me, because besides the extremes I really have difficulties in anticipating this reverb's sound perception with the room descriptions (wall material, room shapes etc.) provided. So, I'd experienced tweaking it to be rather tricky - and stopped using it completely.
Good luck, since I have the Plate140 (to my knowledge it's generating primarily "tails", but awesome ones. If used subtly, it blows up the sound by adding some sort of resonant "body"/addon without being percepted as a verb FX..). And: I have the AAR - just try the demo, tells its own tales...

...I'm just wondering - are those models so difficult, lock up, proprietary, patented - that no one else can figure out their sound or be given permission to produce something of that quality.
....

I think you're actually adressing different questions here. I have no idea about patents. But for the first approach to remodeling you need a plan, a copy of the structural design. If not available, you might very likely need a hardare unit for look up or even decomposition. Then one has to figure out, what components are making up the sonic fingerprint and what are just providing scaffold functions. Now (skipping TONS of additional tasks for reader's sake here..), you have to recompose the digital model - always considering things like performance etc.. After some (long) time you end up saying "Yeah, we've nailed it!". Well, at first that "heureka" is simply a claim. One has to proove it. How do you prove it? By A/B-comparisons. With a 1176 or a LA2A this might seem a breeze, but with a PlateReverb or a Fairchild compressor this may turn harder. And note, the Plate140 is modelled after THREE different real plate units!
So, I think UA's expertise is that they are able and competent to perform ALL those tasks.

cheers,
LiteOn

[kylen]

...How do you prove it? By A/B-comparisons. With a 1176 or a LA2A this might seem a breeze, but with a PlateReverb or a Fairchild compressor this may turn harder. And note, the Plate140 is modelled after THREE different real plate units!
So, I think UA's expertise is that they are able and competent to perform ALL those tasks.

cheers,
LiteOn

Yes - it looks like UA has assembled an expert team for sure. Very impressive folks, thanks! Who can compete with that? I like the idea of developers attending mastering sessions and renting quality equipment for A/B testing - and inviting a few sets of trained ears to listen to creations. Hmmm...dsp plugin developers are beginning to sound like musicians or producers - just a different branch of a similar market...

[Aleksey Vaneev]

Just wanted to say one thing - there will be time when UAD will have to offer something really exceptional to compete with native plug-ins. With multi-core compressors appearing (and we'll eventually see 4 or 8 core processors) it's hard to see how can UAD DSP boards survive in the long run. Only if they offer something exceptional (but here we have a problem of understanding of what exceptional really is).

[kilroy]

...(but here we have a problem of understanding of what exceptional really is).

Too true. One man's junk is another man's treasure.

[kylen]

Hi Aleksey - I was hoping you had something to say about all this! What is multi-core from a dsp point of view - I can google it I 'spose.

Also in terms of sounds and vibe I think you've been one of the developers here that has listened to lots of hardware, or so it would seem, with your impulses and gearmatch in CurveEQ for example. Any ideas about continuing this in the future for the purpose of adding some more organic analog type modes to some of your plugs - the modes you have seem to lend themselves to future extensions or different types of behavior.

[Aleksey Vaneev]

Kyle, I hate the word 'analog' when it comes to DSP development. People give too vast range of meanings to that word. I'll continue developing digital plug-ins for now. Of course, $5.000+ analog equipment will still be in demand for million dollar star producers. I'm not in that 'analog' market, and do not want to even touch it - it's a different demand, different mentality. And I guess a cheaper analog equipment is not that much better than the current level of digital FX/equipment.

If you have money, why not order UAD and try it yourself? I know it should sound quite good (of course, not miles better than what you already have), but from my developer's perspective that won't be a long reign. Anyway, 'don't let the tools of tomorrow hinder the music of today' (or something like that).

As for the multi-core (dual-core for now) processors - they already appear from both Intel and AMD. It's basically the same as having two processors working in parallel. At some point audio effects will be running on a couple of sub-processors while everything else runs on the others. It's mostly the same as having a separate DSP board - audio FX usually do not take too much of memory bandwidth (except convolution), so comparison is very close.

[kylen]

Thanks for the tips and info Aleksey! I might get a UAD down the road and most everyone is raving about them but I just can't help thinking - they don't need that board to do that - like someone else mentioned somewhere it's just a big printed circuit dongle that goes in the pci slot - I believe those same plugs could run on my P4 if they were ported. And that kind of led me to the idea for this thread here...as a plugin user I was wondering what developers felt about this analog modeling and development of that type of organic sounding dsp's - maybe sometimes analog based or partially modeled at least, maybe sometimes not - simply a pure digital model & design. But you're right sometimes analog and other colorful definitions just don't mean much as far as DSP development goes. It depends who is asked I suppose.

I was also trying to see what was on the horizon - get a feel for developers rushing new UAD competition in for the holiday season. So it remains a matter of taste I think and budget - I recently picked up a PSP and Sonalksis comp to keep my Voxengo arsenal company and think I'll stay with native plugins for now. Besides I have a Sonar5 upgrade to look forward to ( ).

[Silent Mind 2005]

I think the problem really is, or atleast used to be, developers of digital emulations of analog gear, mostly looked at the theory of the analog gear, and duplicated that, without considering how it actually worked in practise.

AFAIK, there is no fixed rules on how a compressor should sound. It's pretty much up to your ears. Which is what I did with mine.

True that - there are tons of transfer curves and oscilloscope pictures showing the performance of a piece of vintage gear. That surely doesn't give a clue as to the character of its' sound though, does it. Like how do you take a picture of a compressor sounding spongy or pumping (I guess pumping could be a short video - haha).

Modeling the components seemes like a good place to start which some folks have done in their opto and vari-mu models. I guess I'll know we're there when I have to replace the output driver on my VST compressor because it burned out!

Why not evolve these concepts, instead of trying to sound like some particular h/w.
I mean, I've used quite a bit of different equipment, and there's often been annoyances/things I thought could be better with most of them.
With ADL I tried to do a limiter that works in a similar way, nomatter what program material you feed it. For instance, it doesn't overshoot terribly with extreme peaks, it doesn't recover slow because of that either. I aims to distort as little as possible. (Ofcourse some analog limiter/compressors distorts in a nice way, but eventually you end up wanting to get rid of that.)
And many other things, and you'd be amazed of how many of them are on a psychoacoustic level, so small details that you'd probably even not know about them. (But they are there.)

Regards,
Silent Mind.

[Angus_FX]

Just wanted to say one thing - there will be time when UAD will have to offer something really exceptional to compete with native plug-ins. With multi-core compressors appearing (and we'll eventually see 4 or 8 core processors) it's hard to see how can UAD DSP boards survive in the long run.

Put a couple of Toshiba/IBM Cell chips on there? Don't think dual-core Pentiums will offer eight independent SIMD cores any time soon

Cheers,
Angus.