Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[bobkatz]

Today I was working and decided I needed to back off on the limiter threshold. So I reduced it 1 dB, thereby reducing the absolute level of the program material by 1 dB*. I then increased my monitor gain by exactly 1 dB to evaluate whether I liked the sound of less limiting better.

How can you accurately make an assessment such as this without having a calibrated monitor gain?  I can think of ways of cheating where you temporarily alter the ceiling or output of the limiter, but I mean, AUTHENTICALLY, turn up the monitor by the EXACT amount that you lowered the program. As we all know, loudness is the great fooler, and a small shift in loudness can make you think that something is better when it isn't. Only by having a monitor control in calibrated steps can you make these sorts of comparison decisions without fooling yourself, at least some of the time!

Refute that, DC  



*note, there are refinement issues here. The more squashed the program material, the less linear the relationship between apparent loudness difference and any small change in the limiter threshold. But if you are just occasionally peak limiting on short transients, the relationship will be pretty linear, e.g. if you drop the limiter threshold by 1 dB, the apparent loudness will go down by 1 dB.

[masterhse]

Not to be facetious Bob, but what if one wants to modify a threshold by .5 dB (or any other non-integer)?

One can adjust the master fader in a DAW for adjustments like this without a dedicated monitoring control (and then reset to 0 to avoid any additional processing).

[Level]

One of the key elements to consider is that EACH stage has its own set of dynamic irregularities and may not track true to a given dB level incrementation. For experiment sakes, loudspeakers do not absolutely track voltage swings. My buffer amplifiers tend to amplify dynamics within the top half of their output. Linear dynamics from circuit to circuit is something that has always been an area of notice. The true measurement will always be that of the output of the loudspeaker system in a given room. (considering their translation) Using the augmented source is the other true mensurement. One true acoustic/electrical and the other, electrical. Several circuits working in series with each other will have different dynamic attributes, for better or worse. To know how a circuit "hands off" to another circuit and to be aware of what is happening within all of the ranges of said circuits is an exercise that many do not consider. It should always be considered, studied and documented to keep aware of how the circuits work in conjunction, dynamically, with one another.

Calibration, experiment, document.

I have tools here where a 0.5dB increment can spell a much different swing at the loudspeaker output. Every circuit "does something" dynamically..even though they should be predictable, often this is not the case. Kudos to the ears...once again.

[barefoot]

bobkatz wrote on Wed, 20 October 2004 14:51

Today I was working and decided I needed to back off on the limiter threshold. So I reduced it 1 dB, thereby reducing the absolute level of the program material by 1 dB.............. How can you accurately make an assessment such as this without having a calibrated monitor gain?..........

I hear what you're saying, but do you find that you apply this sort of uniform level shift very often?   I would think if you applied any frequency dependent processing at all, the idea of a calibrated monitor adjustment would go straight out the window?  

I've debated this same question in my own product development.   I sometimes think stepped attenuators, for example, are far more useful for channel matching than they are as absolute level adjusters.   In many cases it seems that ganged, continuously adjustable, and perfectly matched (if there were such a thing)  pots would be more useful.  What do you think?  

Thanks!
Thomas

[Level]

Quote:

I sometimes think stepped attenuators, for example, are far more useful for cannel matching than they are as absolute level adjusters. In many cases it seems that perfectly matched, continuously adjustable pots would be more useful. What do you think?

Continuously variable attenuation/augmentation devices are very hard to match..where as multi-stepped attenuators with precision resistors (calibrated) is the better option. Obtaining resistors of 0.05% tolerance and using enough of them (110 steps) to get down to tenth or even hundredth of a dB may even yield other calibration problems within other circuits.

I use 63 step attenuators matched to within 0.01dB. Most other equipment does not track that fine. My console certainly doesn't...it is then...the ears again..

[barefoot]

Level wrote on Wed, 20 October 2004 15:31

Continuously variable attenuation/augmentation devices are very hard to match.....

You must have quoted me right before I updated my post with "and perfectly matched (if there were such a thing)"

Yeah, I realize that other than a digital attenuator,  such a  pot would be very difficult to realize.  Just thinking about it in principle.  

Thomas

[masterhse]

Quote:

Continuously variable attenuation/augmentation devices are very hard to match..where as multi-stepped attenuators with precision resistors (calibrated) is the better option. Obtaining resistors of 0.05% tolerance and using enough of them (110 steps) to get down to tenth or even hundredth of a dB may even yield other calibration problems within other circuits. I use 63 step attenuators matched to within 0.01dB. Most other equipment does not track that fine. My console certainly doesn't...it is then...the ears again..

Again it seems from the above that gain adjustments in the digital domain would be more accurate that fixed resistors.

[barefoot]

masterhse wrote on Wed, 20 October 2004 15:46

Again it seems from the above that gain adjustments in the digital domain would be more accurate that fixed resistors.

Careful, lest you spark another one of THOSE debates.

Thomas

[Level]

From my other text, the other circuits being used are far more (Insert...can be) inaccurate than a digital control and even stepped attenuators in analog. Some circuits can drift channel to channel on output as much as 1.5dB   even with a rock solid input. Certainly, we demand more accuracy but doing experiments with changing the known input X amount may give different results on the output between channels. I get plenty of mixing come here with the center off as much as 2dB when others say...the kick and vocal were dead centered. If it is centered here, it is centered to within 0.1dB through the chain. This comes from obsessive calibration. Circuits drift over time. It is not uncommon to see 1 dB drifts in a single piece of equipment (especially valve equipment) Add up several pieces chained together and inaccuracies channel to channel can get out of hand.

Just measure, listen and know what is happening. If you don't know..do a regiment of calibration weekly..eventually you can catch a naughty piece of equipment showing its ugly side. Switches are another area to check.

[dcollins]

Level wrote on Wed, 20 October 2004 15:23

absolutely My buffer amplifiers tend to amplify dynamics within the top half of their output.

Isn't that also called "distortion?"

DC

[Level]

No, it is called an hyperactive "intentional" gain stage. Many circuits compress anyway.

Only because of intentional impedance transfer, does it add gain, and gain equals dynamic enhancement in most cases....Like any well built line stage will within calibration.
Besides, I designed the goddamned thing 9 years ago. I know what it does.


DC, I use calibrated devices, you got an axe to grind? Seems like your replies have been augmented (to me personally) by negative response as of late.

No, it does not introduce harmonic or intermodulation distortion that is measurable.

So you take me as some sort of a joke? I certainly hope not.

You should try one someday. It allows you to hear the fine details.

If anything, it eliminates the distortions of the poorly designed line stages absorbing content, out there in use,  enmass.
I am not dropping names, out of respect as well.

It would be advisable to consider the same for me..if you will.

Can't we get along around here?

"distortion" No, when you consider how much typical loudspeakers, less than 4% eff. burn up in heat, I call it compensation. It is doing wonderful things considering most loudspeakers are far from linear dynamic over a wide bandwidth. How about your room correction. Measure it, see how much it is soaking out of your voltage swings, to ears. Hummm.

A. Take B&K lab instuments. Do a reference reading at reference FQ. Increase voltage by X2. Measure loudspeaker. See how far off at one meter it is from 3dB increase. Add line buffer, measure again. Calibrate your monitoring system. Is this something you have done or measured? Really? How often?

Come on now.

(Hint?)
(I hear the rumblings backstage, I was not born last week)

Some of this mistrust and lack of knowlege of physics makes me fucking ill.

[dcollins]

Level wrote on Wed, 20 October 2004 23:56

No, it is called an hyperactive "intentional" gain stage. Many circuits compress anyway.

So your line stage acts as a dynamic range _expander_ above a certain level? Not sure why this is desirable.

Quote:

Only because of intentional impedance transfer, does it add gain, and gain equals dynamic enhancement in most cases....Like any well built line stage will within calibration.

I'm confused here.  A line stage adds gain, and _maintains_ the dynamics -- not enhancing.

Quote:

DC, I use calibrated devices, you got an axe to grind? Seems like your replies have been augmented (to me personally) by negative response as of late.

No axe.  But when I have a question, I axe it.  Even had a pair of NS-1000's back in the day.  Beryllium mid/tweet wasn't it?

Quote:

No, it does not introduce harmonic or intermodulation distortion that is measurable.

Surely not Zero Distortion?  

Quote:

If anything, it eliminates the distortions of the poorly designed line stages absorbing content, out there in use,  enmass. I am not dropping names, out of respect as well.

Again, I'm confused.  What does "absorbing content" mean?  Noise?  distortion?  Magnitude errors?  Phase?

Quote:

A. Take B&K lab instuments. Do a reference reading at reference FQ. Increase voltage by X2. Measure loudspeaker. See how far off at one meter it is from 3dB increase. Add line buffer, measure again. Calibrate your monitoring system. Is this something you have done or measured? Really? How often?

Confused again.  Did adding a line stage make the speaker compress less?

DC

[masterhse]

Sorry, but I'd like to get back to BK's concept in original thread. While many compressors and limiters have gain makeup, has there been any work out there that has done this based on the concept of "loudness" (i.e. sones instead of voltage levels)? Granted it's probably something that you want to remove before creating the final master, but gives a better overall impression of what's going to happen to the final product once the consumer lowers the volume knob (probably not done in 1 dB increments).

[bblackwood]

Level wrote on Thu, 21 October 2004 01:56

dcollins wrote on Wed, 20 October 2004 22:05 Level wrote on Wed, 20 October 2004 15:23 absolutely My buffer amplifiers tend to amplify dynamics within the top half of their output. Isn't that also called "distortion?" No, it is called an hyperactive "intentional" gain stage. Many circuits compress anyway.

I dunno, anything that changes the signal is distortion where I'm from. Especially in the monitor path...

Why not just use a monitor path that doesn't audibly change the signal at all?

And Bill, chill. When you say something that seems completely contrary to others' experience, people are going to ask questions. That's what these forums are all about...

[bblackwood]

bobkatz wrote on Wed, 20 October 2004 16:51

Refute that, DC

Tom (masterhouse) made a good point above.

Also, iirc, Dave has always promoted the use of 1dB steps in the critical range. Could be wrong, but that's how I remember it.

[masterhse]

bblackwood wrote on Thu, 21 October 2004 10:04

bobkatz wrote on Wed, 20 October 2004 16:51 Refute that, DC   Tom (masterhouse) made a good point above. Also, iirc, Dave has always promoted the use of 1dB steps in the critical range. Could be wrong, but that's how I remember it.

Found this paper:

 http://www.sfxmachine.com/docs/LoudnessAndDynamicsMatching.p df

(assuming that they have permission to use)

[Phillip Graham]

Edit: No need for me to chime in on this.  If you don't have anything nice to say, keep yer' mouth shut  

[bobkatz]

masterhse wrote on Wed, 20 October 2004 18:07

Not to be facetious Bob, but what if one wants to modify a threshold by .5 dB (or any other non-integer)? One can adjust the master fader in a DAW for adjustments like this without a dedicated monitoring control (and then reset to 0 to avoid any additional processing).

I already knew these (valid) objections would come up. But no one can deny the strong case this makes for a repeatable, stepped (1 dB or less) monitor controller.

A small point: What if changing the level within the DAW affects the sound? Don't you want to have an objective monitor section that is independent from the issue of the level within the DAW?

BK

[dcollins]

masterhse wrote on Thu, 21 October 2004 07:19

http://www.sfxmachine.com/docs/LoudnessAndDynamicsMatching.p df

That's a great paper.  There's also one from the guys at Waves from a couple years ago.  I'll dig up the publication number if you want.

DC

[bobkatz]

As a reminder to those who brought up the valid objection that you cannot equate threshold changes in a compressor, for example, with exact changes in monitor level---yes, absolutely true, and I did mention that as an asterisk in my first post on this subject.

Nevertheless, I don't think this negates the need for or the concept of knowing exactly how far you have moved your monitor control. If only to say, "wow" I think my ears are getting fatigued---this doesn't sound loud to me, but look, my monitor pot is set 3 dB over where I usually keep it, and look at the meters!

BK

[masterhse]

dcollins wrote on Thu, 21 October 2004 17:04

masterhse wrote on Thu, 21 October 2004 07:19     http://www.sfxmachine.com/docs/LoudnessAndDynamicsMatching.p df That's a great paper.  There's also one from the guys at Waves from a couple years ago.  I'll dig up the publication number if you want. DC

If it's not a major PIA, sure!

[bobkatz]

masterhse wrote on Fri, 22 October 2004 17:48

dcollins wrote on Thu, 21 October 2004 17:04 masterhse wrote on Thu, 21 October 2004 07:19      http://www.sfxmachine.com/docs/LoudnessAndDynamicsMatching.p df That's a great paper.  There's also one from the guys at Waves from a couple years ago.  I'll dig up the publication number if you want. DC If it's not a major PIA, sure!

That is a good AES paper. One thing's for sure, it's a lot harder to level the songs of a very dynamic album than it is a very compressed album. And the more dynamic the song, the less the principles of the "averaging" idea in the PDF would apply, as if it starts soft yet averages loud, it still requires a lot of good ol' fashioned human intervention to make that song fit well after the previous loud one ends.

BK

[masterhse]

bobkatz wrote on Thu, 21 October 2004 12:02

I already knew these (valid) objections would come up. But no one can deny the strong case this makes for a repeatable, stepped (1 dB or less) monitor controller.

I believe that there is an even stronger case for automatic gain makeup in limiters and compressors that's based on the human perception of "loudness". Hearing the real effect of smearing and squashing dynamics as you lower the threshold may make a few engineers think twice about how far to take it, though as the article points out, this may be best done outside of real time. I really feel that the average consumer doesn't want volume per se, what they want is the "fullness" that you get from volume. It's also one on the major reasons IMHO that you see so many graphic EQs with smiley faces.

Until automatic gain makeup based on loudness shows up more ubiquitously, a monitor controller is definitely the best way to go, either via analog or digitally.