Macker, I would be very happy to see if a theory behind the interlinking Volume and Expression curves will emerge!

Bon appétit!

Paolo

Going to the first couple posts from last march again....  I want to say something more..

@Another User said:

Generally, CC7 should be the final volume leaving the instrument.

as I said earlier..that is the same as if you were adjusting the volume slider on a hardware synth.  Its the final and last gain coming out of the VST instrument.

So first thing, the above statement should be verified to be correct from VSL, but that would be my assumption.

What happens inside the instrument is known only by VSL, but we can assume some of the following gain staging unless told otherwise by VSL:

• First gain stage is the instrument programming including the amplitude of the waves as recorded and saved as sample data
  
  
• Second gain stage involves any programming incurred inside the instrument affecting gain
  
  
• Third gain stage is application of velocity or VelocityXF (really part of the above 2nd stage)
  
  
• 4th gain stage is any changes to the output level applied by CC11 expression (which might also be part of the above stage2)
  
  
• 5th gain stage is the final volume coming out of the VST instrument, which is analogous to the volume slider on a hardware synth and controlled by CC7
  
  
• The instrument output feeds to an instrument channel in your DAW where there is another gain stage, There may be further gain stages if the signal goes through any other BUS, etc.  These gain stages are able to boost the signal rather then merely reduce.  However, you want to avoid doing that unless all of the earlier gain stages are already at full unity gain and there is no other way.

There are numerous tutorials on the internet about gain staging and best practices which should probably not be ignored when using gain staging, in order to ensure that you don't raise the noise floor.  This is part of what I have been referring to as preserving the dynamic range.

I don't entirely understand what PaoloT is wanting to determine in terms of percentages between cc7 and cc11, which are but two of the various gain stages involved, and they are most likely right next to each other in the signal chain as well.  But the interesting question I think he is trying to understand is what is the difference between changing one vs changing the other.

You can probably think of both of them as being full unity gain when set to 127, and any smaller number is a gain reduction at that stage.  (VSL would need to confirm that for this instrument).   Assuming they are linear, to keep things simple for now, I am going to call these two faders by generic names for now....F1 and F2

If F1 = 110, that is 87% of full.  So then we assuming the signal is reduced to 87%, and if F2=127, then it doesn't  reduce gain...final output is 87% of full.

if we do the reverse, F1=127 and F2=110.  Same result.  First fader doesn't reduce, second fader reduces to 87%.

If we set them both to 110, I would expect the result to be 87% of 87%. (which is the same as overall 76% by the way.

Note that in that case, the first fader F1 is shaving off 13%, the second fader F2, is only shaving off 11% when set to the same value of 110.  This is because it is is dealing with an already reduced signal.

All of the above is assuming linear response rather then logarithmic, just to keep things simple.

what do you want to do with this information PaoloT?

The above is basic gain staging and makes assumptions which we don't know the answer unless VSL answers this thread about specific instrument programming.

In particular interest to me is what happens with CC11?  if it is a simple gain control, just prior to leaving the instrument and right before CC7's gain stage...then CC7 and CC11 are really just working together as one effective gain stage...which also most likely only provides gain reduction in general, no boosting.  And the exact amount of gain reduction is probably along the lines of what I said earlier.  

However, its always possible VSL does something different in terms of how these two controls manipulate gain, only they can answer that.  In particular its always possible that CC11 is somehow changing gain much earlier in the instrument's internal signal chain in some way.  VelocityXF, for example, actually cross fades the samples while also applying gain reduction according to the value.  So its quite a bit more than a simple gain reduction.  So the question for VSL is, what exactly does CC11 do.  

Another question is about the Synchron player's own built in mixer...is either CC11 or CC7 applying gain reduction prior to the Synchron internal mixer (which by the way is also a gain stage, part of the instrument programming).

So these are all questions we don't really know the full info could effect the answer.

Paolo, I've been measuring what Synchron Player's Main Volume and Expression faders do, but haven't finished that yet - in every measurement it can take ten minutes or more until the meter readout is settled to 2 decimal places. (I'm using BlueCat's DP Pro Meter plugin, which does lots of very useful stuff, including RMS level averaging in real time over however long I want.)

A few highlights of my findings so far:-

   Both faders have the same logarithmic transfer function, as one would expect.

   -1.0 dB is at MIDI value 120; (all dB values here are relative to 0 dB attenuation at MIDI val 127.

   -2.5 dB at 110;

   -4.1 dB at 100;

   -6 dB at 90

   -8 dB at 80;

   -10.3 dB at 70

   Going all the way down to -83 dB at 1, then complete cutoff at 0.

If both faders are set to 90, the combined result is of course -12 dB. All combinations (excluding zeros) are simply the sum of the two dB attenuation values as yielded at the two MIDI values. There is no easy way of avoiding having to know what dB attenuation is yielded at what MIDI values on these two faders.

Sorry to say your heroically bold percentage idea can't be applied to any other combination except the one you've derived by listening or by meter readings. And as you mentioned, the simple but somewhat fussy mathematical expressions that can derive MIDI value from required attenuation dB, and attenutation dB from MIDI value, just aren't the sort of thing that mix engineers generally would care to memorise or even look up, let alone use in the heat of the moment while mixing.

What I want at least for myself is a complete and accurate chart of attenuation dB versus MIDI values - and I hope it may also be useful for you and many others. That's going to take a while because doing these measurements takes forever and certainly ain't fun. And I must say I'm a wee bit miffed that VSL haven't seen fit to provide any such chart, table, map or graph in the Synchron documentation.

Also I want to produce a simple drawing that shows several key dB values on one side of a fader and MIDI values on the other, linked by horizontal lines - as a sort of graphical template I'd like to see (or at least imagine) overlaid on these two faders in the Synchron Player.

The particular use case you outlined at the start of this thread is a critical one; it highlights a problem we all have to face if, like you, we want some default headroom on the Expression fader and yet also want to preserve the Synchron Player's preset overall output level. As I've quoted above, 6 dB down is at MIDI val 90. I'm worried that in some presets it might be impossible to raise the CC7 fader by 6 dB to compensate for this (not unusual) 6 dB of default headroom on the Expression fader.

One workaround might be to set the required Expression headroom of 6dB in all Synchron Players in use, but leaving the Volume faders untouched - then simply make up for this 6 dB loss on all the Synchron Players, at some common stage downstream.

I'll post the full results of my measurements probably in a few days.

Macker, your dynamics table would be very useful. Reasoning in terms of dB would make a lot more sense. The resulting math would however be, I fear, out of my immediate reach, and I will end up doing all my adjustments by ear!

Paolo

To add confusion to my already confused mind: when doing volume change on many channels in a mixer, trying to preserve the same balance between them, you move the faders all of the same amount.

So, does this mean that changing the Volume control of the same amount in all the instruments is the right way to do it?

Paolo

It all depends on what you mean by "amount".

There are two main cases, each having different consequences.

   1. Strictly speaking, if the faders at first are each at different settings, and then they are all moved by the same amount of distance (in mm, inches, or whatever), the balance will be altered. But if at first the faders are all at exactly the same setting, then moving them all by the same amount of distance - any possible distance - will preserve the previous balance. In the case of MIDI CC faders calibrated in number values from 0 to 127, these values can also be regarded as if units of physical distance of fader movement.

   2. Now if the "amount" by which all the faders are moved is dB, then the balance will always be preserved, no matter what their settings were before the move. In this case, if the previous settings of each fader were all substantially different, then moving them all by the same amount of dB means that each fader will be moved by a different amount of physical distance - which also means by a different amount of MIDI CC values.

However, in actual practice in Case 1, especially when mixing for orchestral concert performance (i.e no heavy compression or limiting), if the initial settings of the faders before the overall move aren't very different, and if the overall move is quite a small distance, then you may well find that the balance is not altered significantly by the move. This is because the logarithmic curve of these volume faders is not hugely different from a straight line when considered within a small segment of the curve. Thus for example, if initial fader settings are all within about 2 or 3 dB of each other, and then you move all faders, say, from MIDI CC 115 to 95, you'll probably get away with it.

Don't worry, Paolo, decibels are alien to natural human thought and can trip up even the finest minds now and then - I've seen it happen enough times. From my own experience I'd say it's best to shape your own intuition-friendly mental model of how decibels work (of course ensuring that it's valid for your purposes), then lock it down deep and hard in your mind, use it often, but only let it come back out into the light of day rarely if at all, lol.

Macker, very clear, thank you! (I said 'very clear' indeed?)

Paolo

This has nothing to do with logarithmic scaling, so humor me and keep that out of discussion here for a minute.

I think you are trying to find a way to use VSL's CC7 default values since that appears to be the way VSL is level-balancing Synchron instruments against each other.  Yes so far?

Defaults

So the defaults are to have CC11 at Max....  and then you can adjust CC11 from 0-127 which will essentially attenuate the signal even further as you desire over the range set by CC7.    Note that in a way CC7 provides the total dynamic range...and CC11 adjusts the level linearly within that range.   When CC1 and CC11 are set to max 127 you can think of that as the max possible volume of the instrument, which VSL is level balancing against other instruments and their max levels...by using CC7.

Breathing Room

Now I think you are saying that's all fine and good, but you want some breathing room because if you use that default setup and you start drawing in a bunch of expression curves to your composition and then suddenly you find out that its not quite loud enough even when both CC1 and CC11 are set to max.  And you need to be able to bump it up a little more.  Fair enough.   Also level balanced sample orchestras are probably never perfectly balanced.  So you need some room to fudge it.  Fair enough.

That means you didn't probably have the max volume level set right to begin with via CC7.  So assuming that you could bump up CC7 a little bit to get the max level where you want it...but then all the curves you already drew for CC11 would all be too loud.  So I don't think that's a good solution really, you probably should leave CC7 alone once you start recording tracks and programming CC11 curves.

So you suggest that you could have CC11=110 set as the home base "max" position, and that way if you get halfway into your composing you realize that you need more max volume then you originally thought, then you can just pump up cc11 past 110...take it to "Eleven" so to speak...  

Reduced level

So assuming you set CC11=110 and generally treat that as your max instrument level....and only rarely go to Eleven...I don't think there is anything else you need to do to compensate anything.  Just be consistent with all your tracks, always use the same approach.  Me personally I still prefer generally to use CC1 (velocity XF) for each instrument dynamics anyway...and I would generally leave CC11 set to its home base position (whether that is 110 or 127), and only rarely use CC11 to change dynamics slightly louder then max perhaps.  I don't see the point of using CC11 instead of velocity XF.  But no matter, some other sample libraries don't provide velocity XF, so that would be the case when you would use CC11 and just always treat the max as 110 except for rare situations where you have to go to Eleven as mentioned above.

In the above, yes the overall instrument output will be lower then factory, since you'll be using CC11=110 as the max in order to have room to go to Eleven.  That is an attenuated signal by a factor of 13%.  That doesn't really matter as long as all your instruments are following the same rule, then their output will all by attenuated by 13% and the level-balancing will be preserved...just everyone at a lower level.

MirPro

As a side note, MIRPro does essentially exactly this.  When you use MirPro instrument presets, they are all dropped way down in level in order to preserve level balancing against some other really loud instruments such as Timpani.  It means that other instruments output from MirPro and very low levels, way way way below -18db, which many people including myself have struggled with a little bit, but it doesn't matter, its all relative.  

Trim Plugin

If you want you can simply add a trim plugin to the master bus to bring them all back up again evenly by the same amount.  How much gain to add would be 13%...but of course gain plugins often use use DB units to specify the gain,, so you'd have to figure out how many DB's of gain to add, one setting for the entire orchestra.  One master bus with just one trim plugin using some amount of gain that brings them all back up, relative to each other, by a factor of 13%.  

That's all you really need to do.

Adjusting CC7 back up

I think it would be very difficult to try to adjust all the CC7 values up from the factory setting in order to compensate for the 13% reduction you are making with CC11=110 because of the math I mentioned where its a percentage of a percentage.  I am not sure you could simply raise CC7 by 17 (which is 13% of 127) and have the level balancing preserved.   It needs to be bumped up by 13% from the current cc7 range.

If you want to try the math, you need to use percentages, maybe you were already suggesting this but I could not understand what you wrote, but simply take the following math, assuming VSL's cc7 val and changing CC11 from 127 to 110.

@Another User said:

new_cc7_val =  oldCC7val / (110 / 127)

Then change CC7 to the above and use 110 for CC11 and it should be the same output.

Also, some of the instruments may not have room to go up enough to compensate, before hitting the ceiling of 127 after running this calculation.

Therefore its going to be easier to compensate that 13% signal loss from CC11 by simply adding trim plugins after Synchron in the plugin chain, you can either add one trim plugin on every track, and make sure they are all set the same...or just send them all to some common bus and put one trim plugin there that raises the level ~13%.

I am essentially doing exactly that with MirPro now when I have ensembles and stuff that will never have loud sounds playing....

I think involving DB calculations just complicates the issue because of the logarithmic nature of DB as a measurement.  You just need 13% of gain added back. whatever DB that turns out to be will be different depending on the loudness of the instruments.

Dewdman, it's not just a matter of getting louder. The Expression control can be used when you don't need a change in timbre, therefore prefer to leave the VelXF control unchanged, and change the relative volume with Expression. Sometimes, you want both controls to act together, to give more emphasis to the change.

Having Expression already at maximum would prevent any change up. I don't know how others do in this case, but I find much easier to start with some headroom.

Your handy formula to calculate the volume compensation is very useful. Funny how it give a similar result to the one I found, showing how little the difference can be with these small change. So, maybe it's not all that difficult to preserve the Natural balance set by VSL for their presets.

Paolo

@Another User said:

Having Expression already at maximum would prevent any change up. I don't know how others do in this case, but I find much easier to start with some headroom. Your handy formula to calculate the volume compensation is very useful. Funny how it give a similar result to the one I found, showing how little the difference can be with these small change. So, maybe it's not all that difficult to preserve the Natural balance set by VSL for their presets. Paolo

Andreas, I'm not exactly sure what you meant in you post above by "not identical" and "results in slightly different curves". Am I to understand that the individual Master Volume fader and the individual Expression fader each have a unique 'transfer function' or 'calibration' - call it what you will - in terms of attenuation dB versus CCvalue? I think that's what you've asserted but would you kindly clarify?

I promise you I'm not asking merely pedantically; I do have a real and practical need to know.

[UPDATE]

Well I've just done many more test measurements. And now my question is different. Now the issue is quantitative rather than qualitative.

Yes I have now discovered what a few spot checks didn't reveal previously: there are indeed numerous small differences in calibration between the two faders. I was caught napping by this unexpected circumstance.

And yet strangely, many CCvals give exactly the same attenuation dB level in both faders - within my meter's resolution of 0.01dB at least. The differences seem to be just peppered around - I haven't yet come up with any plausible purpose..

I've found that only 4 of these differences come anywhere near to being possibly significant (0,58, 0.61 0.70, and 0.76 dB, all at below -30dB of attenuation). The other differences are less than 0.4 dB, many of these being so small that they're impossible for me to distinguish from experimental error. I don't imagine that any of them are likely to have any noticeable impact on gain-staging and mixing in actual practice. It's said that most people can't distinguish level changes of less than about 1 dB. All of these differences are less than 1 dB; all but a few being far less.

So my question, Andreas, is changed.

Why is there this strange peppering of insignificantly small 'calibration' or 'transfer function' differences between these two main faders? And why your caution? Why do we need to be mindful of which way around they are connected? Cannot their respective attenuation dB levels be simply summed in order to know their combined attenuation?

I don't actually mind the presence of these small differences because I don't see how they can significantly affect any of our usual, practical level-adjustment activities. But is there perhaps some other factor, maybe some hidden conditional attenuation mechanism operating between these 2 faders that prevents their combined use from being straightforward and virtually 'ambidextrous'? 

[Deleted due to trolling]

(moved to other thread)

@PaoloT

I just made some measurements because I'm curious...in case this is of any help here is what I got.

I used Synchronized Dim Brass Trumpets XF

The default from factory is CC7=90, CC1=127, CC11=127. Max peak = -7.32 dBFS, RMS=20.84 dBFS

Moving CC7 to full max, I get Max Peak = -1.29 dBFS, RMS=-15.10 dBFS

So the first observation is that when all controls are turned all the way up, Synchronized Dim Brass Trumpet ALL XF gets very very close to clipping. Dangerously close actually, but it only clipped for me if and when I played a chord, which its not designed to be used that way anyway. So the point is, it looks to me that VSL endeavored to keep the signal coming out of Synchron in this instrument as close to 0 dBFS as they felt was possible.

It also shows that their default setting for cc7=90, which lowers peak to -7.32 dBFS, which presumably is how much they feel Trumpets need to be lower to balance realistically with other Synchron instruments.

Trumpet is a fairly loud instrument in the orch anyway, and at -7 dBFS its not at all a problematic level in terms of headroom IMHO.

Change CC11=110 and we get Max Peak = -9.90 dBFS, RMS=-23.34 dBFS

Now in my view that is still totally usable, but its also a little bit low, because of the CC11 home base point set at 87% of full unity.

In order to use a trim plugin to bring all the instruments back up together...I believe a corresponding gain plugin bump would be: +2.5db, which resulted in following measurements: Max Peak= -7.36 dBFS, RMS=-20.85 dBFS

That is close enough for government work, as we say....

I do want to point out one more thing though, if you bump the level back up using either a trim plugin or CC7, then later end up taking CC11 to "Eleven" (CC11=127), then you hit MaxPeak = -4.9 dBFS and RMS=-18.3 dBFS.

That's still not clipping range, so may be ok...but if working with something like percussion or something that has capability to get really loud, there is always a remote possibility of clipping that instrument if and when you push it to "Eleven" with the CC's and the Trim plugin adding +2.5db on top of that. So watch out for that. For example, with this instrument, if I move both CC7 and CC11 to max (while sending through the trim plugin at +2.5db), it definitely clipped.

One advantage of using CC7 to compensate, rather than a trim plugin is that you wouldn't have to worry about clipping like that (as long as you don't play chords in this case). Because VSL gain staged this instrument to peak out at -1.29db when all CC's are at full max and playing monophonic lines.

The disadvantage of compensating with CC7 is that some instruments may not have enough room at the top to fully compensate them, depending on what VSL's default CC7 setting was. Also, you have to make this adjustment in every single instrument preset as custom presets (which you're doing anyway when you set CC11 to 110).. whereas with the trim plugin you can send them all through a bus and compensate them all with a single trim plugin.

Hope that makes sense.

All of the above makes me ponder my MirPro setup.. MirPro presets impose quite a bit of attenuation also for the sake of level balancing. Actually significantly more than the Synchron CC7 approach. I will have to do some similar testing both inside and out of VePro to see what results I come to. But I think in the case of using MirPro, I would set CC7=127 for all instruments and then set CC11=110, as @PaoloT is doing, for the wiggle room, but with out the CC7 compensation (since I will use MirPro's natural volume slider for that.

its not clear to me yet whether MirPro presets are calibrated already to take into account that Synchron instruments have CC7 set lower or not and then perhaps adjusts them a little more, but what I can say is that by default the MirPro attenuation with Natural volume is rather significant and usually feels like too much, so I will almost certainly be using some kind of trim plugin to bring it back up, but it also means that if and when I go to "Eleven" with CC11, there is more possibility of clipping something if I'm not careful. So that will be something to watch out for.

UPDATE: I measured the results of running this default Synchronized Trumpets through MirPro, and yes its calibrated to depend on the factory default of CC7=90 in this case. When running this Synchron instrument through MirPro instead of using the IR's in the instrument, with CC7=90, CC11=127, and MirPro using Dim Trumpet instrument preset as well as Mirx mode for character....the resulting level measured very very close to the above scenarios without MirPro...so conclusion is: they have set these MirPro preset's natural volume level to expect the level coming from the VSL CC7 defaults, as far as I can see, at least with this instrument. That's a good thing...but it means basically I could pontentially change all my Synchron presets to use CC11=110 as the home base, and either compensate with cc7 bumped up a little bit for each instrument or a trim plugin with +2.5db on the whole bus...and they should all work with the MirPro presets and the default Natural Volume levels. and those Synchron presets would still workout without MirPro in that case.

The only thing I think I don't like about VSL's approach of using CC7 for the level-balancing, is that a CC7 fader is often exposed in numerous places on most DAW midi tracks, making it all too easy to accidentally bump this fader and lose the carefully planned orchestra balance.

I think I would rather reprogram my presets and swap those around, while setting the 110 thing, so that CC7 is set to 110 and CC11 gets whatever value that VSL put into CC7.

That way, the CC11 value will represent the orchestra balancing...(instead of expression) and CC7 would be used for expression...which makes it very convenient to access from the midi fader in most DAW's without upsetting the intended orchestra balance.

So in summary... This is what I will do:

1. Reprogram Synchron presets to assign VSL's cc7 value to the CC11 slider
   
   
2. set cc7 to 110 as home position
   
   
3. add trim plugin to my main orch bus with +2.5db of gain
   
   
4. use either with or without MirPro in this configuration
   
   
5. Use CC7 for expression above and beyond CC1 XF

@Dewdman42 said:

All of the above makes me ponder my MirPro setup..  MirPro presets impose quite a bit of attenuation also for the sake of level balancing.  Actually significantly more than the Synchron CC7 approach.  I will have to do some similar testing both inside and out of VePro to see what results I come to.  But I think in the case of using MirPro, I would set CC7=127 for all instruments and then set CC11=110, as @PaoloT is doing, for the wiggle room, but with out the CC7 compensation (since I will use MirPro's natural volume slider for that.

I asked Dietz about this a while back.  Specifically: what the CC7 should be set to when using something (either VI or SYN-zed with the SYN-zed impulse turned off) in Mir Pro.  He said that as long as CC7 is set to the same value across the board for all instruments, it shouldn't matter *what* it's set to as the natural volume values (which are just adjustments in gain, not CC7 values) will then all be based off the same CC7 "foundation".  He did mention that Herb tweaked the natural volume settings with CC7 set to 100 (as he recalled), and I decided just to go with that since CC11 increases in granularity the lower CC7 is from its max value.

What I mean by that is that Expression has iterations from 0-127, but within the volume range defined by CC7.  If CC7 is set to 127, Expression and CC7 are basically the same thing.  But if CC7 is set to, say, 100, then Expression gives you 128 iterations *within* the 0-100 window defined by CC7.  So, each step or iteration of CC11 then results in a finer volume change, which I enjoy as "smoothing" on top of VelXF (CC1 in my case).

@Another User said:

and I decided just to go with that since CC11 increases in granularity the lower CC7 is from its max value.

What I mean by that is that Expression has iterations from 0-127, but within the volume range defined by CC7.  If CC7 is set to 127, Expression and CC7 are basically the same thing.  But if CC7 is set to, say, 100, then Expression gives you 128 iterations *within* the 0-100 window defined by CC7.  So, each step or iteration of CC11 then results in a finer volume change, which I enjoy as "smoothing" on top of VelXF (CC1 in my case).

I understand you perfectly here.  The interesting thing here is that the desired level of the instrument needs to be pretty much where it needs to end up for that increase granularity to hold..  If you were to later add some audio gain to the signal (after reducing the gain with lower CC7 setting), then the increments would simply be expanded into the same granularity as before... if you see what I mean..  But yea if lowering the CC7 instead of lowering audio, gets you the correct final desired level for that instrument, then that will result in finer granularity while using CC11

@Dewdman42 said:

I understand you perfectly here.  The interesting thing here is that the desired level of the instrument needs to be pretty much where it needs to end up for that increase granularity to hold..  If you were to later add some audio gain to the signal (after reducing the gain with lower CC7 setting), then the increments would simply be expanded into the same granularity as before... if you see what I mean.

I do see what you mean, never considered that.  My approach (thus far) has been to set all the CC7 and Gain levels (natural volume +/- my own preferences, etc.) and then let VelXF and CC11 do everything regarding volume and dynamics.  I'm not sure if this is the best way, but so far it's working for me for anything needing to sound more-or-less like orchestral whatever.  However, given what you said, I can see how that "fixed" balance could be thrown off via something like stereo bus gain automation after the fact.  One solution (if needed) that comes to mind would be to render/freeze the audio before automating gain - that way, the balance is retained because gain automation is raising the level of a simple audio file, not one of three (four?) different sliders...