Orchestral reverb spacing

[ManFromGlass]

D’oh!
Of course. Got it now.
But my stubborn brain is saying in theory that xylophone at the back of the hall is very close to the back wall of the hall so virtually no delay for the reflections but you are talking about the reflections reaching us in the audience.

I need to play with your system a bit as I’ve never successfully figured out how to get bright instruments with sharp attacks sound like real instruments at the back of a real hall.

[Wonks]

You could play with this calculator to work out how much to cut with EQ at different frequencies for different distances.

http://www.sengpielaudio.com/calculator-air.htm

I'd decide how far away from the instrument I wanted to be, then I'd plot out a graph for different octaves and then EQ for the relative attenuations at that distance. I'd take the temperature as between 22-23°C and around 55%rH.

Where it becomes more difficult is the attenuation of the reverb frequencies if you are using a single reverb. You could try adding a different EQ into each of the delay paths, but in reality, the reverb sound is taking many paths of many different lengths, so will have a real mix of HF attenuations. Hopefully, a good reverb algorithm should take that into account.

[The Elf]

Still a bit of confusion I think...

But my stubborn brain is saying in theory that xylophone at the back of the hall is very close to the back wall of the hall so virtually no delay for the reflections

Yes, that's it! So from the audience perspective we will hear the reflections at practically the same time as the source - so a very short (if any) pre-delay will help to create this illusion.

Anything at the front of the orchestra will exhibit a much longer pre-delay, and likely produce a weaker overall reverb (though I often fudge this to generate power and excitement in a mix!) - it will also be brighter and louder.

[Dr Huge Longjohns]

Alternatively, get a reverb like Altiverb that lets you place the sounds on a stage or in a room exactly where you want them. You can get a similar but a lot less sophisticated effect with non-convolution verbs like Waves Truverb which has a 'distance' parameter and Eventide's SP16 which I just bought which has a similar control. (It's fantastic.)

[The Elf]

Alternatively, get a reverb like Altiverb...

But you need a different instance for every placement. I find that the overlaps get phasey and cluttered. I prefer one reverb for a convincing single space.

[ManFromGlass]

It gets even more complicated for me. I usually need to bounce out up to 8 stems (in Logic) in one pass. That means a reverb on each stem. In this case potentially 3-4 delays sending to each reverb (and associated stem). Each of the 8 reverbs would be identical.

Doable once I get my head around it. For most music I write I've liked the idea of every thing living in the same "space" if I am going for more organic (as opposed to a fantasy anything goes environment). Thanks for this concept!

[Martin Walker]

I typically end up with bigger values than those, but I do it by ear without reading the actual values. The 'front' of my orchestra often ends up in the 80/90ms region.

Ah yes, the impoverished student always listening from the cheap seats right on the back row eh Elf?

Martin

[Gone To Lunch]

I have abandoned using the DP in-house delays and instead use 5 instances of 2CAether reverb, which allows separate early and late reflections, ER & LR, with individual delay controls. Thus 1-4 are ER only with delays of 8, 16, 24, and 32, all feeding 5 which is LR with no delay, all instances of N7 Golden Hall, which I think is a Bricasti clone ? Much much better, the L-R positioning of the instruments is clear and the sound is altogether bigger and brighter. Until I start playing with the EQ tomorrow.

[The Elf]

I'm not a fan of using multiple reverbs when realism is the aim, but if it works for you, then great!

[Martin Walker]

I'm not a fan of using multiple reverbs when realism is the aim, but if it works for you, then great!

But surely Gone To Lunch is actually in his description using one reverb (and Aether is my go-to choice as well), but inserting its early reflections at different times from the different delay taps, before sending all of them into the late reflections of that same reverb.

To me that would imply a carefully thought out 'single' reverb with more creative opportunities.

I'm going to try this myself too

Martin

[The Elf]

Depends what processing is going on it the reverb. I long ago tried this with Altiverb and it phased horribly!

[Gone To Lunch]

But surely Gone To Lunch is actually in his description using one reverb (and Aether is my go-to choice as well), but inserting its early reflections at different times from the different delay taps, before sending all of them into the late reflections of that same reverb.

In the 1-4 ER instances I vary the predelay setting, 8,16,24 & 32, and have no predelay in the LR instance.

However there is a minor downside. With just the all-in-one single reverb approach it is very easy in Aether to scoot through all the various pre-sets. So one needs to do that first to pick the best candidate for the multi-approach, so as to avoid having to change settings x 5.

[The Elf]

Ah... I see what you're meaning. I'd misunderstood.

Worth a try, I suppose. It offers the potential to change ER patterns, but is that an advantage? Dunno! Worth a try, though.

[Gone To Lunch]

It offers the potential to change ER patterns, but is that an advantage? Dunno! Worth a try, though.

Yes it does, because Aether is so editable, but the only thing I actually vary is the predelay because I am trying to fake different distances in the same space obviously.

[Gone To Lunch]

But surely Gone To Lunch is actually in his description using one reverb (and Aether is my go-to choice as well), but inserting its early reflections at different times from the different delay taps, before sending all of them into the late reflections of that same reverb.

In the 1-4 ER instances I vary the predelay setting, 8,16,24 & 32, and have no predelay in the LR instance.

The angels, like the devils, are indeed in the details.

Delays of 10,20,30 & 40 sound noticeably 'better' as in cleaner/clearer, brighter.

[The Elf]

I'm struggling to see, assuming the same ER patterns are chosen, how this method differs from using a few simple delays.

[Gone To Lunch]

I'm struggling to see, assuming the same ER patterns are chosen, how this method differs from using a few simple delays.

I don't know for sure. I am hoping Uncle Hugh will be along shortly.

But as a guess, could it be that since Aether creates the reverberation in realtime, the four differently delayed ER algorithms create four different reverberations which are then combined via the shared LR algorithm, not least because the four ER algorithms are supplied by different sound sources ?

Eg, in my track :

F = front - strings
FM = front midddle - winds
BM = back middle - horns
B = back - percussion

In contrast to the four discrete delays going into the one ER + LR combined.

Thus in the four ER + one LR set up, there is more variation in source, and therefore in output, which I then optimistically perceive as 'better' and 'more realistic' ?

[Gone To Lunch]

But surely Gone To Lunch is actually in his description using one reverb (and Aether is my go-to choice as well), but inserting its early reflections at different times from the different delay taps, before sending all of them into the late reflections of that same reverb.

I have now gone back to using just one single reverb instance, but changing the amount sent from each instrument. Because I found that changing the frequency profiles available in Aether actually gave better results. In other words, I hadn't yet understood what this function is in Aether, but when I explored it in more detail, it is very powerful, and works much better than doing EQ on the reverb aux output.

The down side is the ultra-tiny rev send controls in DP, too small to be seen with the naked eye, and no useful numerical display.

Another up side is that it is much easier to trawl through all the presets at the beginning of the process.

[Moroccomoose]

I Have a question about how pushing the sound further back affects the panning.

Using my Rompler (Proteus 2000), I have 3 stereo pairs for outputs. So I have these set up with delays prior to sending to the reverb for near middle and far distance.

I am now in the process off arranging the orchestra using the MIDI panning and the front to back outputs to place the instruments in the pit.

As an element is pushed further back from the listening position, (not) using the delay, then sending to reverb, do I also need to pan out further. eg 1st violins near the front go to the longest delay with a paning of say L48 and the 1st violins towards the middle go to the next shorter delay with panning to L63. Or do I give both front and middle instruments the same panning.

I get if it sounds good, its good etc and that this is probably a second order effect to the overall sound, but I model physics in my day job and wondered how might be the most accurate way to model the orchestra panning when the instrument is given depth.

Stu.

[The Elf]

Massive generalisation warning...

Source signals further away will exhibit less stereo 'width'. Sources closer to us (imagine your head inside a piano!) will exhibit more stereo 'width'. Their reverb is agnostic.

So to make something seem more distant I would typically use less overt panning and narrow (if not simply mono) source material.

[Hugh Robjohns]

I guess it depends on what orchestral layout you're trying to model....

Really, I understand you want it to 'be right' but there is no 'right' because every orchestra is different and their seating plans change with different venues and different repertoires.

So just fiddle with the panning until it sounds right to you... but as the Elf says, the image width tends to decrease the further away things are...

H

[Moroccomoose]

Thanks, that settles the argument in my mind! I was torn between visualising panning as an angle in the sonic field, or as a distance to the right or left from the listening position. By visualising as an angle, no need to push things further L or R as they radiate away from the listening position.

Right - now to set up all the articulations

Cheers!

[Peterlkarl]

I always set up 3-4 reverbs in DP, though not usually to simulate an orchestral performance environment. If I'm understanding what you're trying to, you could just set up 4 stereo Aux sends and returns, each with similar Proverb impulse responses employed, but varying pre-delay settings, (and maybe progressively longer decay times). Have a ball!

[The Elf]

I always set up 3-4 reverbs in DP, though not usually to simulate an orchestral performance environment. If I'm understanding what you're trying to, you could just set up 4 stereo Aux sends and returns, each with similar Proverb impulse responses employed, but varying pre-delay settings, (and maybe progressively longer decay times). Have a ball!

As I explained above, one reverb produces a less cluttered, less phasey and more convicing result that simply adding the same reverb multiple times.

Try it yourself and see.

[Zukan]

Good technique Elf. I like.