voxengo for reverb?

[bmanic]

I highly recommend the http://www.quikquak.com/ reverb thingy. It creates basically an "ir" reverb in realtime. Each 'ray' can be 'modulated' in realtime too which creates some interesting spaces. I think it's one of the best ambience effects out there right now. It's hard to make it sound like a lexicon though.

Cheers!
bManic

[Aleksey Vaneev]

Just for clarification, ER is an overly simplistic model. In real spaces, 'early reflections field' is what really matters, because hardly in practice any space creates distinct early echoes. In a real room, what is usually called ER is a kind of 'dense' reflection field - not a single spike. That is the reason why algorithmic reverbs can have problems with early reflections - just because using a set of single ER spikes (8, 16 or whatever number is) does not sound convincing to the ear most of the time.

On the other hand, early reflections is just estimation of how *close* the nearest walls are to the microphone. So, this does not make ER part a must or something that tells everything about reverb sound.

[Kiwburger]

And yet 99% of the time we are using convolution to model the famous algorithm reverbs. There is a huge difference between realistic reverb, and musically useful reverb.

[Krakatau]

And yet 99% of the time we are using convolution to model the famous algorithm reverbs. There is a huge difference between realistic reverb, and musically useful reverb.

As much as i experienced, the mistake with those so to say realistic reverbs is that it doesn't make much sense, on that ground of realism, to use them as an insert on a mixing table i guess

Though i did it actually only for small acoustic banks soundtakes, i think it makes much sense to use them in the mastering process especially, of course, when simulating a live situation for acoustic bands of any kind

[Krakatau]

Just for clarification, ER is an overly simplistic model. In real spaces, 'early reflections field' is what really matters, because hardly in practice any space creates distinct early echoes. In a real room, what is usually called ER is a kind of 'dense' reflection field - not a single spike. That is the reason why algorithmic reverbs can have problems with early reflections - just because using a set of single ER spikes (8, 16 or whatever number is) does not sound convincing to the ear most of the time.

On the other hand, early reflections is just estimation of how *close* the nearest walls are to the microphone. So, this does not make ER part a must or something that tells everything about reverb sound.

this is very interesting....

Thanks for our guidance ( at least mine ), Mr Vaneev !

[james0tucson]

I just spent the weekend in Sedona Arizona, a few hours drive from my home. Did some rock climbing in a certain canyon. Found a place that had the most excellent natural resonance, and played my flute.

I wish I knew how "IR" was done, because I'd love to have some kind of convolution model for places like *that*. Does it require some kind of location audio rig? Probably the kind of thing that's stupidly expensive for a backpacker...

[aMUSEd]

I would love to see somebody build a hybrid reverb that uses IR for ultra smooth reverb tails, and algorithms for the Early Reflections. I would love to see a reverb that can "sample" an impulse response, and find the highest peaks, and calculate the Early Reflection delays necessary to recreate them. The delays could then be user tweakable, but basically that would remove a lot of the tedium and tweaking algorithm designs.

I tend to think that to a certain extent, a diffuse reverb tail is fairly similar from one reverb to the next. The length and dampening obviously change - so maybe the sampling program could just detect the RT60 and slope of the filters and generate a smooth tail from a wavetable.

It's the ER's that give the character and sense of space. If we say assume that this reverb will have 16 early reflections, we could sample any impulse and look for the peaks on the Left and Right channels and create the parameters for all 16 delays, and i'm thinking this would be musically useful. The advantage? You could then tweak things to taste, and apply modulation to the delays or the tail. And it should be fairly economical on CPU too.

Doesn't the Wizzoo verb do something like this? - or at least the hybrid part.

[Krakatau]

As should prosoniq Rayverb do, as well as the last version of Altiverb as much as i can judge ?

[sascha]

Just for clarification, ER is an overly simplistic model. In real spaces, 'early reflections field' is what really matters, because hardly in practice any space creates distinct early echoes. In a real room, what is usually called ER is a kind of 'dense' reflection field - not a single spike. That is the reason why algorithmic reverbs can have problems with early reflections - just because using a set of single ER spikes (8, 16 or whatever number is) does not sound convincing to the ear most of the time.

I'd basically agree.
That's why Griesinger's proposal is to be *very* careful with ERs in artificial reverberation.

Most manufacturers love to 'wash' discrete echos with several allpasses in a schroeder-moorer fashion (without feedback). You can achieve sort of an ER field that way although it's not the real thing. Problem with the allpasses is that they destroy a sense of 'space'/dimensionality. A dedicated FDN/mesh structure (with very small FB and a larger amount of delay lines) would be a good starting point, but it eats already up a lot of CPU cycles because you also have to take frequency-dependent things like wall & air absorptions into account.

The 'ideal' reverb in terms of CPU cost vs. efficiency & sound would currently be based on IRs for the ER part and an algorithmical approach for the tails.

Many people think that it's the tails that are hard to implement nicely. ERs were easy. But that's just plain wrong. The smaller the room gets, the harder a reverb box has to struggle to sound good. You can easily build a reverb with monstrous never-ending tails, but it can fail easily with adding 'cheap' spatial jobs.

Alex, you seem to hold the banner for IR-based reverberation under all circumstances, maybe because you want the 'real' thing. If realism is the main goal, I would agree with you. But from an artistical point of view, algo reverbs are still a perfect thing for many jobs. Often, one needs that 'grainy', undense sound field for a signal to cut through in a busy mix. This is where IR verbs tend to fail, not because they are IR verbs, but because the user might not have access to a specific echoic pattern that just does the job without much hassle (remember, there's not much you can tweak...). Often, vocals can be fattened up by just stacking up a bunch of echoes in the background, much better than just wrapping a complex room around it (with combing, masking and that stuff included in multiples).
Okay, that's also taste, of course. I'm a rocker, classical or jazz people have another sonic viewpoint

Maybe there's some educational work for us developers to actually tell people that techniques are not only excellent by design. It's their application that decides over their excellence.
But I get constantly upset when it comes to the common thinking that convolution is the best and 'le dernier cri', so forgive me

[bmanic]

The smaller the room gets, the harder a reverb box has to struggle to sound good. You can easily build a reverb with monstrous never-ending tails, but it can fail easily with adding 'cheap' spatial jobs.

Note that this is also true for real rooms. The smaller they get, the harder it is to keep them from sounding like crappy toilets.

Cheers!
bManic

[sascha]

Of course. That's in the nature of physics. That's one of the reasons why an airport hangar feels more comfortable than an elevator cabin

And because of claustrophobia most of us tend to demo the large reverbs first whenever we check reverb gear. 'amaaazing...', 'huuuuge'...
Funny that most modern music asks for claustrophobic delay times instead

[Aleksey Vaneev]

I was not talking that IR is the best thing in reverb world... It's just best if you want 'exactly that ambience', and not just 'some decent ambience'.

[brok landers]

The smaller the room gets, the harder a reverb box has to struggle to sound good. You can easily build a reverb with monstrous never-ending tails, but it can fail easily with adding 'cheap' spatial jobs.

Note that this is also true for real rooms. The smaller they get, the harder it is to keep them from sounding like crappy toilets.

Cheers!
bManic

correcto mundo.
one has to get away from the labeling "real=good".
the situation in a mix or a song is mostly far away from having a realistic, natural ground. therefore one needs a reverb to get a certain "job" done, a certain "function" in the mix.
in my case i sometimes have best results with some reverbs that could hardly be called something like proffessional or good.
i.e. silverspikes roomsimulator (the free one) ... far away from sounding realistic in any way ...
solo'ed it simply sounds like crap and you tend to just ask yourself "why in hell would someone use this??"
but one i used in a mix i was simply astonished what i can do good for a too dense mix ...
it's only the impression at the end that counts, what it does to your mix ...
mostly it's not the "real thing" that one would like to achieve, but the "real impression under the given situation" ...
this makes a hughe difference ...