Challenge to VST reverb developers

[birrbits]

I have an old Ensoniq DP-4 that does superb non-lin stuff. This is actually an excellent box all round. If you troll e-bay you see them, from time to time. *Very* programmable...a veritable tweaker's dream. Kilroy recommends it, he does.

I keep a rack computer with a bunch of older ISA cards including an Ensoniq Soundscape Elite. neat thing about this card is that it has the same DSP effects engine as the DP-4 (but only 1 processor so no chaining) and many of the same or similiar algorithms. Only other downside is that to use the effects they must be on the synth or a playing wav (no realtime) but hey it was 30 bucks & its effects are great. There is even an old software toolkit to program the crap out of the effects.

[kL]

"Convolution is like using a bulldozer to crack open a nut."
I love that one. Eloquently put.
It's true, perhaps, but that nut is well cracked after the 'dozer runs over it.
If it's tweakability you want, check out Pristine Space. It's a tweaker's dream, multi-channel, realtime, no latency, and sounds amazing. I'm surprised the reverb-wanting world hasn't beaten a path to its Russian door.
(But then, reverb has been somewhat out of favor in some circles since the '80's...)

[greendoor]

I remember reading about Lexicon, and how they chose to forget about modeling early reflections seperate from the reverb tail. I've had a lot of hardware and software reverbs, and I generally don't like the ones that have seperate early reflections. To me, a real room doesn't really seperate out early reflections - but then in a real room you have the luxery of near infinite delays. I think Lexicon just decided to use the technology they had (presumably simple delay lines and filters) and just make them sound good, regardless of the theory. If I want an early reflection sound, i'll use delays as a seperate effect.

I'm serious about suggesting prime numbers for basing the delay times. The problem with cheap reverbs is that they can model the worst characteristics of bad rooms - a build up of resonances, so that some notes can sound really bad. There are various models for perfect rooms for music, designed so that the harmonics of the nodes of the room are evenly distributed. It seems to me that if prime numbers were used, the delays would be evenly distributed - but I don't know, and maybe i'm missing something. Maybe they would be so evenly distributed there would be audible resonances ... which comes back to the idea of using modulation. I don't really care how it's done, as long as it sounds great and is very flexible. Maybe modulation using prime numbers Maybe if each delay had a subtle chorus effect, with varying lfo rates - 3s, 5s, 7, 11s, 13s etc

[greendoor]

KL - I know about Pristine Space, and I currently run Voxengo AF Impulse - the lite version that sounds as good but has zero tweakability. So I tend to use SIR, which is free and more tweakable. But the best impulses I have to run in these convolution processors are samples taken from Lexicon and TC reverbs ... which is the whole point of my little rant/motivational speech.

[Aleksey Vaneev]

greendoor, prime numbers for delay line lengths in a room reverb is a NO-NO thing. Otherwise it will sound like a metallic plate. Rooms and halls need another approach - initial delays should be finetuned for good early-reflection picture and the delay lengths chosen randomly, or according to user-controllable 'density' parameter. Delay line feedbacks should be chosen automatically according to the delay lengths and the RT60 value (overall reverb decay time).

[greendoor]

Thanks Aleksey - I've heard it is more art than science. Although I expect your impulse maker would be using some heavy duty maths based on the room models designed by users. Can you see algorithmn reverbs using real models of rooms being viable? I'm not knocking your excellent convolution products, but I think the earlier analogy to waving a corpse around does apply to static samples

[kL]

I think a preference for hardware reverbs is a preference for the artificial, an ear-of-the-beholder/fashion thing, something I was alluding to with the comment about the '80's and the popularity of reverb.
The first time I heard a high quality reverb "effect" it was a plate -- made of real metal -- in the first real studio I ever recorded in. So I still love the lovely, liquid sound of plates.
But they aren't real. And they aren't timeless. And neither are Lexicons or Quantecs. Not nearly so much so as an impulse of the Concertgebouw, for instance. Which isn't either, of course. (But I wouldn't compare that static sample to waving about a corpse: the walls of that grand room don't move, do they? Aren't both ceiling and floor pretty static?)

But a trip into the Heisenbergian crack wherein lives all pop music is not the subject of this thread. Reverb is; reverb, that easily overused effect that has destroyed many a mix and should be applied as carefully as mayonnaise -- or it's hideous, artifical counterpart, Miracle Whip.

Some people thought chorus was a wonderful, natural thing, too, back when Roland "invented" it. I say, play in time and play in tune.

(is there an emoticon for "tongue in cheek?")

[mandolarian]

I think the earlier analogy to waving a corpse around does apply to static samples

Not if you have enough samples!

We just need 64+ bulldozers, then the nuts will be really tasty and very life-like.
Imagine a room, one instrument, say a guitarist sitting down, not moving around much. And a stationary listener. That's one sample. The trouble with the single static sample for me is in the dynamics. The room is going to be 'excited' in different ways with various dynamic 'inputs'. We just need more layers of samples. And a few random overlays and more dozers. Like the fancy sampler, tho, as soon as we have more than one instrument in the virtual space, the illusion starts to display its fake, non-interactive 'life'.

I think convolution verbs are wonderful, but they have some limits. Reality, for one. Most real rooms sound bad. And much of the time it seems we want an 'idealized' verb. Better than real. Hence the success of Lexi, TC etc.

This holy verb chase has some history, I recall the days when those Eventide, Lexi hardware boxes were considered the 'corpse' units and for a real-life sound you needed an EMT 140 plate...

[mandolarian]

I say, play in time and play in tune.

But then the quantum fretless bass player turns to you and says, "In time, in tune. Pick one."

[Aleksey Vaneev]

greendoor, I think it is possible to model 8-12 early reflections based on simple room design and performer/mic position. It won't be as convincing as Impulse Modeler-generated impulse, but should work in most cases.

[Aleksey Vaneev]

I wanted to add that 'early reflections' is a very simplistic approach. In reality each disctinctive early reflection is usually surrounded by a trace of 'almost early' reflections, so that early reflections are smeared a lot.

[wsaidah]

Real Spaces is were its at when it comes to Convolution reverb for me! i don't like the hardware boxes much but there are some gems inside some of the packs! go with the real spaces learn how to eq and gate your verbs i use a flanger on all of my tracks most people are scared of that

[dystonia_ek]

The big thing missing in VST reverbs is excellent Inverse or Non-Linear tails. That requires full editability of the timing. I want a big slow reverse reverb tail that peaks right before the snare hit - that sort of thing.

If you have Reaktor or Session, try Ultimate Reverb from the User Library. Will do reverse tails from 0 to something like 30 seconds. UR and Spacemaster are about the best sounding software reverbs I've heard - only thing lacking is low end, to a degree.

[kritikon]

that easily overused effect that has destroyed many a mix and should be applied as carefully as mayonnaise

Well said with most of your post.........unfortunately I use mayonnaise somewhat as a replacement for butter or margarine. And then I add the extra mayonnaise for flavour as well. I use posh mayonnaise, mind you So although I totally agree that reverb should be used carefully...if I used it like mayonnaise, my mixes would sound like experimental Phil Spector. Perhaps we should use reverb more like Marmite

[WhiteNoise]

Most of what you mention is already being used in reverbs greendoor. Also, the freeverb model isn't necessarily the best model. There are other models out there as well (such as FDN's or waveguides). I think one problem is that reverb is still a very cpu intensive task. A fractional delay line that can be modulated and is bandlimited can start to take up some cpu time (especially when you need to have a bunch of them). Most hardware synths seem to have a dedicated CPU just for reverb (and probably have special instructions which make them faster).