What's new

0ms pre-delay on a separate tail reverb question

zigzag

New Member
I have a dry instrument with a convolution reverb as an insert to create early reflections. So the output of this track is combination of filtered/colored (for spatial position) direct sound and early reflections with 25 ms pre-delay, but no tail. Then, there's a send from this track to a tail reverb bus. The tail reverb has 0 ms pre-delay. Pretty standard setup, I think.

If we plot start time of each signal on a time axis, the dry signal starts at T0, ER at T25, but the tail starts at T0 for the dry signal and at T25 for ER. However, if the tail would be enabled directly on the insert reverb, the tail would start at T25+ for the direct signal and (I'm guessing) T50+ for ER (or is tail in this case calculated only from the direct signal?....I have no idea how is signal chain done inside the reverb plugin)

Is this correct? Is the difference audible to you?

Why is it OK to send a signal with direct sound and delayed ER to the tail reverb bus with 0 ms pre-delay? I'm guessing:
  1. It doesn't matter, as the direct sound to reverb tail signal ratio is so large it is barely/isn't audible.
  2. It doesn't matter, as it sound good, even if it's not physically correct.
  3. I'm doing it wrong. I should split direct signal and ER signal, and send direct signal delayed to the tail reverb.
  4. I'm completely wrong (please explain)
  5. Other (explain)
 

averystemmler

Active Member
Of course, the sound is what matters, but thinking theoretically it does depend on how your particular reverbs do things.

Using convolution, some reverbs will keep the tail at it's naturally delayed position when you disable ERs. Others will shift things forward. In my opinion, having the tail hit the dry signal before the early reflections kick in is a bit odd.

Your current approach also feeds the ERs through the tail. Some algorithmic reverbs do this, but in a real room or impulse response of it, the tail is really just these same reflections having reflected a few (hundred) more times, so that work has already been done. With convolution, it depends on which impulses you've chosen, but I think it has the potential to get weird when you put things in series.
 
OP
Z

zigzag

New Member
In my opinion, having the tail hit the dry signal before the early reflections kick in is a bit odd.
That's is what's bothering me with my current setup. But if I add pre-delay to the tail reverb, it will also add delay between ER and tail, not just between direct and tail. The only solutions seems to either split the direct and ER signal or enable tail on each inserted convolution reverb.

From what I gather, most people here that are adding ERs to dry instruments, send 100% wet signal to a single tail reverb bus. At least, that is how I understand it. I could be wrong or this is not a big issue.

I'm trying to create a sensible workflow for dry instruments and gain understanding of it's weak points (and how to work around them if needed).
 

Beat Kaufmann

Active Member
....

Why is it OK to send a signal with direct sound and delayed ER to the tail reverb bus with 0 ms pre-delay? I'm guessing:
  1. It doesn't matter, as the direct sound to reverb tail signal ratio is so large it is barely/isn't audible.
  2. It doesn't matter, as it sound good, even if it's not physically correct.
  3. I'm doing it wrong. I should split direct signal and ER signal, and send direct signal delayed to the tail reverb.
  4. I'm completely wrong (please explain)
  5. Other (explain)
As you suspect, it is physically correct, of course, if the TAIL only begins slowly in the course of the "ERs".
Since I don't know if "ER" and "Tail" create the same PlugIn and if the whole thing is even a manufacturer preset, it could be that a certain effect should be created.
If this is not the case, the "TAIL" should probably start after a time of 50ms to about 200ms.

By the way: If you only add the delayed "TAIL" to an instrument, it will remain in the foreground (acoustically) despite the reverb. Good for solo instruments.

All the best
Beat
 
OP
Z

zigzag

New Member
As you suspect, it is physically correct, of course, if the TAIL only begins slowly in the course of the "ERs".
Since I don't know if "ER" and "Tail" create the same PlugIn and if the whole thing is even a manufacturer preset, it could be that a certain effect should be created.
If this is not the case, the "TAIL" should probably start after a time of 50ms to about 200ms.

By the way: If you only add the delayed "TAIL" to an instrument, it will remain in the foreground (acoustically) despite the reverb. Good for solo instruments.

All the best
Beat
Thanks!

So, the most correct way would be separate signals for direct sound and ER? That would mean: sending ER to tail=delay of somewhere from 0 ms to ER's length; sending direct to tail = delay depending on the ER's pre-delay value and ER's length?

That complicates routing quite a bit and I am wondering, if the difference would be worth the effort.

Currently, I have Altiverb as an insert on each instrument group with "Positioner" turned on (for stage positioning of the direct sound) and "early" turned on, but "tail" turned off. For the tail I'm using one Valhalla Room with "Depth" set to 100% and "Early Send" set to 0 (so it is not creating additional ERs, but only the tail) on a separate bus.
 

Beat Kaufmann

Active Member
Hi zigzag
Maybe you're making this too hard for yourself. Treat anything you want for position from left to right from back to front with ER and Altiverb instances if you want. There would probably be more CPU friendly solutions... But it will certainly work that way.

But now add only one instance in the output channel, which adds the necessary tail to the whole mix (as insert effect -> 10-20% wet). Voilà, that's it. This works wonderfully and combines all instruments nicely in one and the same concert room ("glues" all instruments together). And it does not complicate the routing.
Good luck
Beat
 
OP
Z

zigzag

New Member
Hi zigzag
Maybe you're making this too hard for yourself. Treat anything you want for position from left to right from back to front with ER and Altiverb instances if you want. There would probably be more CPU friendly solutions... But it will certainly work that way.

But now add only one instance in the output channel, which adds the necessary tail to the whole mix (as insert effect -> 10-20% wet). Voilà, that's it. This works wonderfully and combines all instruments nicely in one and the same concert room ("glues" all instruments together). And it does not complicate the routing.
Good luck
Beat
The one instance in the output channel has zero pre-delay? That is essentially the same as I have it set up right now.

I have no intention of making an unnecessary complex routing just for the sake of it being 100% physically correct, but not creating any real improvement to the quality of the mix. I'm just trying to gain a deeper understanding of the technology and acoustics.
 

Beat Kaufmann

Active Member
The one instance in the output channel has zero pre-delay? That is essentially the same as I have it set up right now...
NO MY DEAR!
It shoud add the Tail to all those signals which are coming from the groups or single channels with their integrated ERs/positions. This Tail should have those 50 - 200ms pre-delay, so that it comes in after the ERs. :dancedance:
And also once more: This "Tail-Reverb" is an insert effect with probably around 10%-20% wet/dry ratio.


All the best
Beat

Maybe this link helps: https://www.beat-kaufmann.com/vitutorials/about-reverbs/index.php
 
Last edited:
OP
Z

zigzag

New Member
NO MY DEAR!
It shoud add the Tail to all those signals which are coming from the groups or single channels with their integrated ERs/positions. This Tail should have those 50 - 200ms pre-delay, so that it comes in after the ERs. :dancedance:
And also once more: This "Tail-Reverb" is an insert effect with probably around 10%-20% wet/dry ratio.


All the best
Beat

Maybe this link helps: https://www.beat-kaufmann.com/vitutorials/about-reverbs/index.php
Thanks for bearing with me :geek:

I still have a few questions about this:

1) With pre-delay on the tail, a necessary gap between direct and tail is created, so that the tail doesn't start before ER. However, doesn't this delay also create "unnecessary" gap between ER and tail?
I'm guessing that the tail pre-delay is so large because the tail needs to start when ER starts fading. Direct-to-ER delay is relatively small compared to ER's length, so it doesn't affect pre-delay of the tail much?

2) Any specific reason for the tail-reverb as an insert? Isn't send-reverb with 0 dB attenuation on the sends going into it, essentially the same as as an insert-reverb at about 50% wet/dry ratio? And 10-20% ratio is achieved by attenuating send volumes?
I am asking, as I like more control I have with the tail as a send. Instruments in the back can have bigger wet/dry ratio than instruments in the front, to create better sense of the depth. Or am I with this incorrectly doubling the depth effect that was already created by ER? With sends, I can also have different delays set up (for the tail) depending on instrument's depth.
 
Last edited:

Beat Kaufmann

Active Member
1) If you want to know how long your ERs will last, send a short pulse through the ER instance and record it. Then view the audio file in an audio editor and read the duration. Then you know exactly after what time the tail has to start - of course something before the ER signal has completely faded out.

2) The reverberation (Tail), for example, in a church is everywhere and always about the same. When a person speaks very close to you or very far away, their sound signal has more or less direct sound portion or it contains more or less ERs, so that you perceive a distance or none. But the reverb tail is always the same, no matter where a person speaks in the church. Even if you think that distant instruments have more tail, this is not the case. It only decreases the direct sound signal, the ERs tell you that the instrument sounds far away and you perceive the tail better because the direct signal and the ERs have become quieter because of the distance.
The just explained with an example (tail remains fixed)

This is now simulated in the mix as well. At the end you only have to "put" the generally valid reverberation tail over your ER mix. That's why this usually happens in the output channel for the whole mix and that's why no "send" is necessary here, but a fixed value for everything. You can set this fixed value according to your taste. 10%-20% has proven to be a good value in practice.
I hope that now clarifies everything.
All the best
Beat
 
OP
Z

zigzag

New Member
Thanks for this through explanation Beat!

One last thing regarding:
you perceive the tail better because the direct signal and the ERs have become quieter because of the distance.
If the tail-reverb is inserted at the end of the mix when the direct signal and the ERs are already quieter for instruments in the back, won't that cause the tail to be quieter than it should be? As the source was originally louder, than what the tail-reverb is receiving at the end.
 

Beat Kaufmann

Active Member
You're a hard nut to crack.
Sure, if something is loud, the tail is also louder.

Explained the other way around again:
In room X the tail always ends the same - whether an instrument plays near or far from me. The way the tail ends has to do with the construction of the room. Therefore: Tail = fix !!! 10-20% in the output channel with predelay, so that the tail starts after the ERs or a little before. The tail is the further arrival of reflections that are no longer perceived as individual reflections.

My mp3 example above shows: While the direct sound is reduced more and more compared to the ERs and both become quieter at the same time (because of the distance of the marimba), but the tail remains constant (15%), the marimba seems to sound further and further away.

------------------------
Whether an instrument sounds near or far, our brain perceives it mainly through the relationship "direct sound"/ERs. In your mix, this obviously happens in the audio channels or group channels. All these instruments in your mix are now missing only the decay in the room - the tail.
Add this to the output channel. So far you have controlled the distances and the positions of the instruments. With the tail you place the orchestra in a certain room (church, concert hall, forest,...), which fades out according to a certain pattern.
------------------------

Now you know what to do.
Of course you can ask more questions at any time.
You can also somehow set everything up with SEND as you see fit.
That would be a shame, because then I would have invested my time in vain.

------------------------
All my mixes are done with this "Tail in the Output-Channel". It's one of the best reverb concepts I've used over the years (I am working with samples since 2002).

Good luck
Beat
 

Nick Batzdorf

Moderator
Moderator
Technical aside: reverb tails are specified by RT60 - the time it takes for them to drop 60dB.

But if I add pre-delay to the tail reverb, it will also add delay between ER and tail
My words for the same answers above:

And this is a problem why?
 
OP
Z

zigzag

New Member
@Nick Batzdorf I was discussing this more on a theoretical level, trying to wrap my head around what is happening acoustically and how these plugins work. I've learned this isn't an issue.

@Beat Kaufmann Sorry about that :) You've got me to really rethink my current workflow. Those videos, in the link you posted, are terrific! Thanks again, you have been incredibly helpful!
 

Andrew Souter

Active Member
First let me say I'm a big fan of Beat's work, website, and tutorials. All very great stuff! And his VSL ochestral mockups that he provided stems of for educational mixing purposes are phenomenal!

Second let me point out for those who don't know, that Beat spends most of his professional carreer recording real orchestras and classical ensembles, and therefore has very good insights into what happens in real spaces with real mics etc.

Third let me add, that the method he suggests is a pretty good compromise. It works pretty well because different early reflections patterns and different direct-to-ER ratios are applied per track. That is ideal and is a big part of achieving the perfect solution. That conforms to suggestions we advocate also -- we just go one step further as explained below. I will just add some additional clarification below for those interested...

Explained the other way around again:
In room X the tail always ends the same - whether an instrument plays near or far from me. The way the tail ends has to do with the construction of the room. Therefore: Tail = fix !!! 10-20% in the output...

....

The reverberation (Tail), for example, in a church is everywhere and always about the same. When a person speaks very close to you or very far away, their sound signal has more or less direct sound portion or it contains more or less ERs, so that you perceive a distance or none. But the reverb tail is always the same, no matter where a person speaks in the church.
It's not quite that simple unfortunately... The energy of the diffuse late reflections i.e tail is usually somewhat consistent for any position in the room, yes. And the direct sound is NOT consistent in energy -- it is very much dependent on it's position relative to the microphones/listener, as Beat describes, yes. i.e. Distance. Correct.

But this effectively means the Direct to Reflected Ratio changes depending on Distance in an enclosed space. Direct (dry) energy falls with distance, reflected (wet) energy stays somewhat consistent. So let's say the reflected/wet energy is at -12dB, and direct energy at distance 0 is 0dB. We might have:

Distance 0: Direct = 0dB, Reflected = -12dB
Distance 1: Direct = -6dB, Reflected = -12dB
Distance 2: Direct = -12dB, Reflected = -12dB

In DAW/Reverb terms this means more distant sources would have a higher Wet/Dry Mix ratio. Most (good) reverbs use sin/cos mixing for wet dry ratios. So In the above example

Distance 0: = 15.6% wet
Distance 1: = 29.6% wet
Distance 2: = 50.0% wet

the math is:

mixRatio = (2.0 / pi) * atan(pow(10.0, 0.05 * (directDB - reflectedDB)))

(
btw you can use this formula for sends also if you ever wondered what send gain level = what effective wet/dry mix value

mixRatio = (2.0 / pi) * atan(pow(10.0, 0.05 * (-sendLevelDB)))
)

If you put one verb on the master with a fixed Wet/Dry ratio of 10-20%, this will NOT recreate that same situation. All mix elements will have the same relative Direct to Reflected ratio at least for the tails. This is not completely correct.

You can compensate for this to some extent by using extra ER's only direct on tracks, and the result might be reasonably convincing, yes. So Beat's method can acheive pretty good results, yes. But ideally the tail ratio should change also.

In the real world, there is not really a distinction between Early and Late Reflections; there is only a unified "Reflected Energy" which includes both. Early Reflections came about for psychoacoustic research and for convenience to save CPU resources in computer software that has to try to emulate the real world, such as Algorithmic Reverbs, or Acoustics Software used to design concert halls. Performing real physical modeling of a larger concert hall still requires a supercomputer; it's unrealistic. So we talk explicitly about discrete and specific Early Reflections bc they are mostly manageable. And they are arguably perceptually the most important. But even Early Reflections are rarely purely discrete/specular delays. They are usually blurred/smeared in time to some degree and in some fashion or another. i.e. they have diffusion. Generally we talk about Late Reflections as a statistical process which can basically be considered random noise bc it becomes so complex that discerning any kind of structure becomes exceedingly difficult.

The main point is there is not a clear distinction between early reflections and tail in many real spaces. And if you would like to model things as accurately as possible the "Early Reflections" and the "Tail" should ideally be treated in the same fashion. This means among other things:

1) The Direct to (total) Reflected Energy Ratio changes depending on depth/distance.

2) Pre-Delay should apply equally to both, and should be variable depending on depth/distance. The tail should ideally NOT share the same pre-delay for all instruments.


Other aspects also change according to distance.

Precedence Linked with Breeze 2 direct on track inserts in pairs is the only thing I know of that does this completely correctly... sorry if that sounds saleman-y... ;)
 
OP
Z

zigzag

New Member
First let me say I'm a big fan of Beat's work, website, and tutorials. All very great stuff! And his VSL ochestral mockups that he provided stems of for educational mixing purposes are phenomenal!

Second let me point out for those who don't know, that Beat spends most of his professional carreer recording real orchestras and classical ensembles, and therefore has very good insights into what happens in real spaces with real mics etc.

Third let me add, that the method he suggests is a pretty good compromise. It works pretty well because different early reflections patterns and different direct-to-ER ratios are applied per track. That is ideal and is a big part of achieving the perfect solution. That conforms to suggestions we advocate also -- we just go one step further as explained below. I will just add some additional clarification below for those interested...



It's not quite that simple unfortunately... The energy of the diffuse late reflections i.e tail is usually somewhat consistent for any position in the room, yes. And the direct sound is NOT consistent in energy -- it is very much dependent on it's position relative to the microphones/listener, as Beat describes, yes. i.e. Distance. Correct.

But this effectively means the Direct to Reflected Ratio changes depending on Distance in an enclosed space. Direct (dry) energy falls with distance, reflected (wet) energy stays somewhat consistent. So let's say the reflected/wet energy is at -12dB, and direct energy at distance 0 is 0dB. We might have:

Distance 0: Direct = 0dB, Reflected = -12dB
Distance 1: Direct = -6dB, Reflected = -12dB
Distance 2: Direct = -12dB, Reflected = -12dB

In DAW/Reverb terms this means more distant sources would have a higher Wet/Dry Mix ratio. Most (good) reverbs use sin/cos mixing for wet dry ratios. So In the above example

Distance 0: = 15.6% wet
Distance 1: = 29.6% wet
Distance 2: = 50.0% wet

the math is:

mixRatio = (2.0 / pi) * atan(pow(10.0, 0.05 * (directDB - reflectedDB)))

(
btw you can use this formula for sends also if you ever wondered what send gain level = what effective wet/dry mix value

mixRatio = (2.0 / pi) * atan(pow(10.0, 0.05 * (-sendLevelDB)))
)

If you put one verb on the master with a fixed Wet/Dry ratio of 10-20%, this will NOT recreate that same situation. All mix elements will have the same relative Direct to Reflected ratio at least for the tails. This is not completely correct.

You can compensate for this to some extent by using extra ER's only direct on tracks, and the result might be reasonably convincing, yes. So Beat's method can acheive pretty good results, yes. But ideally the tail ratio should change also.

In the real world, there is not really a distinction between Early and Late Reflections; there is only a unified "Reflected Energy" which includes both. Early Reflections came about for psychoacoustic research and for convenience to save CPU resources in computer software that has to try to emulate the real world, such as Algorithmic Reverbs, or Acoustics Software used to design concert halls. Performing real physical modeling of a larger concert hall still requires a supercomputer; it's unrealistic. So we talk explicitly about discrete and specific Early Reflections bc they are mostly manageable. And they are arguably perceptually the most important. But even Early Reflections are rarely purely discrete/specular delays. They are usually blurred/smeared in time to some degree and in some fashion or another. i.e. they have diffusion. Generally we talk about Late Reflections as a statistical process which can basically be considered random noise bc it becomes so complex that discerning any kind of structure becomes exceedingly difficult.

The main point is there is not a clear distinction between early reflections and tail in many real spaces. And if you would like to model things as accurately as possible the "Early Reflections" and the "Tail" should ideally be treated in the same fashion. This means among other things:

1) The Direct to (total) Reflected Energy Ratio changes depending on depth/distance.

2) Pre-Delay should apply equally to both, and should be variable depending on depth/distance. The tail should ideally NOT share the same pre-delay for all instruments.


Other aspects also change according to distance.

Precedence Linked with Breeze 2 direct on track inserts in pairs is the only thing I know of that does this completely correctly... sorry if that sounds saleman-y... ;)
Thank you so much for clarifying this! Now everything makes sense. Lots of good info in your post. That Wet/Dry formula will be useful for comparing or switching between tail-as-an-insert and tail-as-a-send setups. Link between Precedence and Breeze 2 also sounds interesting (I wasn't aware of that).

Beat certainly makes a strong argument for a simplified tail setup, with many examples that prove it can sound good. But, understanding proper physics of the reverb is also incredibly useful, if not a prerequisite. You have to know the rules, to break them effectively. :) IMO, both approaches have their merits.
 

Beat Kaufmann

Active Member
Hello zigzag
Andrew is the absolute pro when it comes to the physical laws of reverb.
However, I recommend that you first do a few tries with only one tail-over all. Anyway, it's not quite clear when exactly the early reflections stop and when the tail starts. By pushing the instruments further back with the early reflections, you automatically have to add more wet. So some of the different bet ratios are already covered by the ERs of your Altiverbs.

So that you don't have to work with different SENDs to the tail part, try to work with only one instance of tail in the output channel.

-------------------------------
Andrew mentioned that I record concerts. At the end I would like to use such an example.
My recordings are based on a main microphone setting and from a few to many supporting microphones. The main microphones "hear" certain instruments close (the ones positioned near) and other instruments further away. I also try to record as little room space as possible. You can actually compare the "dry recording" to your "adjusting the positions with the ERs".
But then the room (tail) that puts everything in the room is still missing. Whether you believe it or not, I do it with only one instance in the output channel.

Listen to the two examples DRY (without tail) and WET (with tail). Andrew will have fun, because for the tail part I use "BREEZE" for a long time.


Why only one TAIL instance? They are simply no more necessary.
As audio mixers we "sell" illusions. It is impossible to play a symphonic wind orchestra over room loudspeakers - but the illusion of it.

Beat

Here is the video of the examples above
 
OP
Z

zigzag

New Member
Hello zigzag
Andrew is the absolute pro when it comes to the physical laws of reverb.
However, I recommend that you first do a few tries with only one tail-over all. Anyway, it's not quite clear when exactly the early reflections stop and when the tail starts. By pushing the instruments further back with the early reflections, you automatically have to add more wet. So some of the different bet ratios are already covered by the ERs of your Altiverbs.

So that you don't have to work with different SENDs to the tail part, try to work with only one instance of tail in the output channel.

-------------------------------
Andrew mentioned that I record concerts. At the end I would like to use such an example.
My recordings are based on a main microphone setting and from a few to many supporting microphones. The main microphones "hear" certain instruments close (the ones positioned near) and other instruments further away. I also try to record as little room space as possible. You can actually compare the "dry recording" to your "adjusting the positions with the ERs".
But then the room (tail) that puts everything in the room is still missing. Whether you believe it or not, I do it with only one instance in the output channel.

Listen to the two examples DRY (without tail) and WET (with tail). Andrew will have fun, because for the tail part I use "BREEZE" for a long time.


Why only one TAIL instance? They are simply no more necessary.
As audio mixers we "sell" illusions. It is impossible to play a symphonic wind orchestra over room loudspeakers - but the illusion of it.

Beat

Here is the video of the examples above
Hello Beat
I have been testing one tail reverb and it's quite interesting what can be achieved with only a single tail reverb. I agree, what matters is the sonic quality of the end result and that is quite often an enhanced reality.
 
Top Bottom