192,118 users have contributed to 42,832 threads and 257,540 posts.
In the past 24 hours, we have 13 new thread(s), 40 new post(s) and 200 new user(s).
Paolo, so, we're back at the original problem - we haven't provided you with any real help. Shame on us, lol.
Actually, it's a very interesting discussion, with some math that is useful to learn. But a clear answer (at least, not one I can grab) doesn't seem to be there, yet.
An empirical solution has the advantage of living in our main domain – the music/sound dimension. This one, we can easily understand!
Paolo
Actually, it's a very interesting discussion, with some math that is useful to learn. But a clear answer (at least, not one I can grab) doesn't seem to be there, yet.
oh wait...I thought we figured that out already. What is the open question exactly?
So I did a simple test.
A couple points. Its hard to be precise with this test because we have to rely on a Synchron instrument as the test sound, which is applying unknown humanization to the sound which may or may not produce consistent results. By playing at least a dozen notes at the same velocity and keeping track of the max peak during that time..might be ok enough.
I agree with Macker that the amount of gain needed to bring the level of step#3 back to the level of Step#1 required is around 2.5db
which is not the same value as indicated by the 20*log10 formula, which comes up with -1.25db (note - the fact that its pretty much exactly 2x of the calculated value is interesting also)
So there is definitely something more to the story, but at the moment, I'm not sure what, I think its probably that none of us, including myself are calculating reference level into it properly while doing that math.
VSL has already stated very clearly that the CC controls are a simple percentage of amplitude. The formula I mentioned absolutely *IS* the correct formula for converting amplitude percentage to dBFS. But it doesn't match the measured results, I agree.
I updated the spreadsheet I posted earlier to include a 2x fudge factor to make up for the discrepancy between what Synchron is outputting and the standard Amp->db calculation. This is essentially the same as Macker's 40*log10() formula. I don't know why this is necessary right now, but it does seem to be necessary.
I also added a line for @PauloT to calculate a CC7 makeup value if you choose to increase CC7 in order to make up for setting CC11 to something lower to represent "unity".
https://docs.google.com/spreadsheets/d/19uQltWICdycednQbQRA0jNHSl75uzhIW7rsyi9OM0Ig/edit?usp=sharing
@PaoloT I think you solved your original problem by one of two ways, either make up the gain with CC7, using the calculation I gave you (or you can use the spreadsheet above if you want). Or...use external gain plugin to make up the gain...which should be 2.5db to make up for cc11=110.
You can put a separate gain after every Synchron instrument, or you can send them all together through a bus and then just need a single gain plugin at +2.5db to bring them all back up as if you didn't lower CC11.
However, note that with the gain plugin approach, you will want to be on the lookout for clipping mainly on the master bus whenever you raise CC11 over 110.
Was there something else unresolved?
The good news is, following an attempted insurrection by the crude forces of dogmatism, modern science has reasserted its long-term possession of certain territory in the field of audio engineering,
The bad news is that the attempted insurrection happened at all, and here in VSL's forum.
In the Mac's dictionary, "dogmatism" is defined as "the tendency to lay down principles as undeniably true, without consideration of evidence ..."
Over the past 5 centuries, modern science has shown the value that empiricism brings to the progress of human existence. Do we want to turn the clock back to pre-modern times, when often absurd dogmas ruled the roost pretty much across the board? I don't think so. Do we appreciate attempts to drag us back there? I don't think so.
So I'll just say this. Let's be careful who we listen to and believe, and let's make sure that the long-established value and reputation of empiricism in the disciplines of mathematics, science and engineering are protected and upheld against any who try to cast them aside imperiously as if mere trash. Also there are further offences to watch out for and guard against, such as bush-league sophistry and smearing tactics, as used since time immemorial to pave the way for, defend and promote dogmatism.
Over and out.
oh wait...I thought we figured that out already. What is the open question exactly?
It's still the one I asked in the first post. Many interesting ideas and solutions in this thread, but not yet one looking as effective as working with a vu-meter!
Paolo
A practical answer to myself to get the same default 'natural' balance of the presets when lowering Expression from 127 to 110
I compared the original VSL presets and my custom ones, whose Master Volume was recalculated by multiplying the original value by 115.3%. It works! Probably just by chance, but the level are matching for all the presets.
Paolo
A practical answer to myself to get the same default 'natural' balance of the presets when lowering Expression from 127 to 110I compared the original VSL presets and my custom ones, whose Master Volume was recalculated by multiplying the original value by 115.3%. It works! Probably just by chance, but the level are matching for all the presets. Paolo And the Vu meter approach has been mentioned several times in the thread as an additional solution. So it seems you are in fact resolved. What other solution could there be? What is left unresolved? Using approximately 115% to compensate cc11=110 is right in line with the calculations I gave you earlier ( now on the spreadsheet ) . The cc11=110 is 86.61% of full. If you take 1/.8661 you get 1.15 which is 115%. Your Vu measurements confirm this. You can change cc11 to any other value instead of using 110 for home and use the calculation 1 / ( cc11 / 127 ) to obtain the needed compensation to use for cc7 fader as a percentage. But when in doubt, using meters to get it where you want is also perfectly fine way to get it without trusting math
right in line with the calculations I gave you earlier
Not that I pretend any merit, but it's the formula I proposed in the fifth post of this thread. I only needed confirmation with the most accurate measurement I could do.
I don't really get this idea of "it's resolved, so what's that you still want?". The process is more interesting than the pure result.
Paolo
Paolo, the error exposed in the spreadsheet has some far-reaching consequences. You may or may not understand the theory behind this, but at the very least you do need to be aware that the erroneous thinking behind that error in the spreadsheet put the discussion on the wrong footing right from the start.
If you'll forgive the techie tedium, I'll go through it one more time.
The formula 20*log(CC ratio) originally (and wrongly) used for the spreadsheet is applicable ONLY IF the CC ratio (perhaps expressed as a percentage) determines audio level changes produced by a device according to what's called a linear law (also known as a straight-line law). That is to say, where a linear law applies to a specific CC-controllable device, any given % change in CC value on the device's input, will cause the same % change in audio level at the device's output. But in the case of volume controls, such as Sy Player's Master Volume and Expression faders, we know from my measurements that they don't work according to a linear law. And that's no surprise at all, because in fact volume controls are always physically manufactured (and now digitally modelled in software) to have what's called a logarithmic law. That's been the general rule as far back as the earliest days of electronics, and it's to match the fact that our ears (and certain other senses also) respond to stimuli in a physiological way that happens to be pretty close to mathematical logarithmic functions, as first described by Ernst Weber and Gustav Fechner in 1860.
The 20*log(ratio) formula is used to express voltage ratios as decibels. The standard definition of the decibel specifies power ratio, not voltage ratio, and dB are defined as 10*log(power ratio). However, as long as impedance remains a constant, it can be shown that power is directly proportional to voltage squared. And in the magical world of logarithms, multiplying the log of a number by 2 is the same as squaring that number before converting the result to a logarithm. So we multiply the power-decibel formula by 2, and arrive at 20*log(voltage ratio). And in practice the result of the 20*log(voltage ratio) formula is sometimes expressed as dBV. Why voltage? Because voltage (in a well designed audio system) is directly proportional to the resulting sound pressure (units of which are called pascals, defined as Newtons per square metre), which is what our ears respond to. And because of this direct proportionality between voltage levels inside the sound system and the resulting sound pressures eventually produced in the acoustic domain, dB SPL (Sound Pressure Level) are used in the same way as dBV are used in the electronic domain, i.e. using 20*log(sound pressure ratio) to express audio level changes. In practice, dB SPL is always referenced to the threshold of human hearing, conventionally defined as 20 micropascals. And today, inside digital sound systems, instead of audio voltages we have numbers that are directly proportional to the voltage eventually produced by the digital-to-analogue convertor in the system's output stage. So the 20*log(ratio) formula still applies in digital sound systems, just as it did in the days of analogue mixing boards. Moreover, dBFS ("FS" = full scale) is merely the most convenient way to express an absolute level inside the sound system; in fact a ratio is still involved, but in this case the level of interest is referenced to the Full Scale level inherent in the system - regardless of whether the system is analogue or digital. It is customary to express attenuation or gain amounts as simply dB, not as dBFS, because for attenuation and gain, usually it is the ratio which is of interest, not the absolute level.
Now let's consider further the ratio part of 20*log(ratio). If we move the Master Volume slider from full scale (127) down to half scale (64), obviously we can call that a ratio of 0.5 or 50% in relation to the full scale setting on the device's input. But does that result in a 50% change in sound level on the output of this volume control device? No, it most certainly does not, as is verifiable by watching SyPlayer's output level meter, and from my graph and table of input/output measurements. If the sound level dropped from 100% to 50% we would expect to see a fall of 6dB in the output meter reading. But in reality the sound level falls by 12dB. My 40*log(ratio) formula is simply a reasonably valid mathematical model of these faders, according to my measurements. The factor of 2 above the usual 20*log(ratio) expression is due to the inherent scaling of the logarithmic law of the fader devices. Calling it a "fudge factor" is merely admitting inability to understand the underlying mathematics, as well as an attempt to deflect attention from the fact that the original formula incorrectly assumed a linear law.
From what I've read in the thread - some of which I see has been subsequently edited - you've been advised all along to treat the two volume faders as if they were linear-law devices. I have added my advice that it may be possible to get away with this incorrect assumption just so long as very small changes of ratio are involved. But more generally, I don't see that there's any quick, easy or convenient way of working around the mathematical ramifications of the logarithmic law of these two faders, regardless of whether we express the output audio level changes as dB or as simple ratios. That said, logarithms and decibels owe their existence to their mathematical convenience. For example in working with audio, we can just add or subtract decibels, pretty much in the same quick and easy way as when, say, we do cash transactions in the grocery shop. As far as I know, nobody's invented or discovered a way of making things even more convenient than using those simple additions and subtractions.
My conclusion is that the goal of ignoring the log-law of the fader devices and getting away from decibels, is, sorry to say, just pie in the sky.
Well this dreadfully long lecture probably hasn't served to simplify or clean up the awful mess this thread got into. But I had to put it out there in the hope that, with any luck, nobody else will be led up the garden path by those very wrong technical assumptions and principles posted in this thread.
To @Dewdman42,
You answered me by claiming I'm wrong. And you tried to back up your claim with some apparently technical points that don't make technical sense. Oh please, I'm no noob. All I spoke about was some testing I'd done, the fact that your spreadsheet gave answers way different from my test results. and that my test results agreed with Macker's numbers. Eventually you changed your spreadsheet, saying that a "fudge factor" seemed to be needed, instead of admitting you got it very wrong. And not a word of apology from you to anyone.
What's up with that?
After some years, and hundred of preset made, I've changed my mind.
The volume of VSL libraries is quite low, so I would prefer to get as much volume as possible along the chain. Therefore, I would prefer to keep Expression at the maximum level, and be free to increase the Volume of the various instruments of the same amount, if needed.
To use Expression to push-up intensity I will simply lower it before the crescendo, and start from there. All other instruments will be adapted to the level of the one needing this help.
Paolo
Hey! I've already changed my mind again!
There should be enough volume left even if lowering the Expression value and increasing the Volume one.
A bit less than if keeping the Expression value at the maximum. So, I've just a doubt on what I should do. Expression at 127 would for sure make the preset defaults simpler.
Paolo
Another reason to leave CC11 to 127 by default is that increasing its value when you have set the correct balance may cause overload. You have maybe calibrated everything so that the loudest passages go near the maximum, and in the end you decide to use that CC11 overhead. Immediate overload!
Also, I think it is good hygiene not to push loudness when needing an instrument to emerge. On the contrary, it should be better to make space around it, therefore making the others sound a bit softer.
Paolo
