Another secret feature?

[Rich Walsh]

I found this by accident – when did this form start working?

setLevel cue "1" row 0 column 0 db "-inf"

Works in 4.7 (and 5); for years I’ve been using a manual variable to match the workspace "Min Volume Limit” – but this apparently undocumented feature seems to be able to pull that setting into AppleScript directly!

(According to the SDEF, db should be a real number.)

Thanks.

Rich

[micpool]

I've always used -100 which achieves the same thing (as long as your min volume  is set higher than this.)

[Rich Walsh]

I’ve always used -120 as that was the minimum possible at some point in QLab’s history, but this form seems to pull the actual value from the settings which feels much neater. It’s there in the OSC API as far back as v3, but I never thought to try it in AS.

Rich

[micpool]

Both methods pull the level correctly

setLevel cue "1" row 0 column 0 db -120

   getLevel cue "1" row 0 column 0

returns the minimum level set in settings.

[Rich Walsh]

Yes, but only if the minimum level is not less than -120 – which, although unlikely, is perfectly feasible. Being able to use “-inf” instead of a relatively arbitrary number of the coder’s choosing seems like a step forward: what if someone were trying to eke every dB out of a 32-bit file and thought their noise floor should be -192? Even -144 isn’t crazy with 24-bit files*.

(The default of -60 is odd though: I’ve never been in a theatre yet with only 72dB of headroom; I generally find -85 or below is the point at which a fade out actually fades into the noise.)

Rich

* You could argue 12 - (23 x 6) = -126 should be the minimum, if +12 is the maximum and you want to account for the sign bit correctly

[Chris Ashworth]

By coincidence we were just noticing that the old v4 minimum of -120 didn’t get passed along for v5, and were discussing what a proper minimum ought to be. 

Glad to be reading everyone’s thoughts on this, as presumably a totally unbounded lower range is not helpful per se.

Also, regarding the default -60 number, I believe that came from the v1 days, and was not extremely rigorous. My memory is that I was trying to pick a reasonable default, sat there listening with my headphones on, and picked something that achieved silence on my headphones without having such an extended range that fade-outs spent a lot of time effectively in silence. 

I hesitate to tweak that value, though as to my memory it’s been rare for anyone to mention it and I am unsure what the consequences of changing it might be 

-C

--
Contact support anytime: sup...@figure53.com
Follow QLab on Twitter: https://twitter.com/QLabApp
User Group Code of Conduct: https://qlab.app/code-of-conduct/
---
You received this message because you are subscribed to the Google Groups "QLab" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qlab+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/58578D1F-1950-4C23-AE58-81FE401F11FA%40mac.com.

[micpool]

I generally  use -70dB for minimum level

The 12dB above 0 are only provided as extra gain for quieter  signals. A 0dB FS peak  will distort in QLab quite soon after the main fader hits 0 (if the cue output sliders  and matrix  are at 0),  so -60 often means just that 

But, unless you are playing back a tone, or correlated noise, your average level may well not be  above -12dB, so adding 12 to the minimum level to determine effective dynamic range is probably reasonable.

So with -70dB minimum, my effective dynamic range might be 82dB  which, in an NR25 empty space,  equating  to a noise floor of 30dBA, and allowing another  15dB for cooling fans and minimum audience occupancy  noise etc,  would  allow an SPL  of  around 122dB depending how much amplification was provided following QLabs outputs, before the -70dB minimum would need to be lowered further.

The likely  result of using very low values for the minimum is that your fade times will have to be reduced considerably to allow for the fact that, on fades of most material, a significant portion of your fade will be beneath the practical thresholds of audibility. It should also be borne in mind how rare  a fade to pure silence is in most shows. Often sounds will be fading into other sounds, dialogue will have started, audience may be reacting etc.

Some of the maths and physics above might not  be strictly  scientifically accurate, but anecdotally  I have never used a minimum below -70dB, and have never really heard the end of a fade.

Interested though in other users that have felt the need to set their minimums lower, or have even noticed the default of -60 seems intuitively quite high. (Although not dissimilar to the end point, before shorting to ground, of a lot of analog faders)

Mic

[Sam Kusnetz]

As usual, I find the analysis and reasoning from both Rich and Mic to be thorough, logical, and immensely helpful.

I have myself never dipped below -87 dB as the minimum, and the time I did that was in an environment in which I was connected to a very high gain sound system and was not given access to much of the equipment downstream of my Mac.

I was very uneasy auditioning each new cue in that venue. I think I set my new cue defaults to something like -30.

sk

—

Sam Kusnetz (he/him) | Figure 53

s...@figure53.com

--
Contact support anytime: sup...@figure53.com
Follow QLab on Twitter: https://twitter.com/QLabApp
User Group Code of Conduct: https://qlab.app/code-of-conduct/
---
You received this message because you are subscribed to the Google Groups "QLab" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qlab+uns...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/26fd9d0f-a7da-44ce-8b18-fac2e38a5b38n%40googlegroups.com.

[micpool]

On Tuesday, November 22, 2022 at 3:51:01 PM UTC micpool wrote:.

The likely  result of using very low values for the minimum is that your fade times will have to be reduced considerably to allow for the fact that, on fades of most material, a significant portion of your fade will be beneath the practical thresholds of audibility.

That should have course  been fade times will have to be EXTENDED considerably..

Mic

 

[micpool]

I though it was worth attempting a definitive reference recording of a test to determine the  differences between QLab Fades, using different musical styles, -120dB and -60dB Minimum Levels (set in settings audio) , and Linear and S curves

You can download the test recording in 96/32 96/24 and 48/24 versions here:

https://we.tl/t-gD0kerNBE0

It's a 300MB download. (I have also attached an AAC compressed version to my next post, so you can decide whether it's worth downloading the uncompressed versions)

The recordings contain the following tests:

Ref Music Set Loud Listening Level

Jazz Trio 4824 Linear -120
Jazz Trio 4824 Linear -60
Jazz Trio 4824 S -120
Jazz Trio 4824 S -60

World Music 4824 Linear -120
World Music 4824 Linear -60
World Music 4824 S -120
World Music 4824 S -60

Piano 4824 Linear -120
Piano 4824 Linear -60
Piano 4824 S -120
Piano 4824 S -60

The main takeaway from the test is that the difference between a -120dB Minimum Audio Level and -60dB is not really as dramatic as you might expect. 

With -60 Linear,  the end of a fade  sounds pretty much like it might if operated well on a Penny &Giles Fader on a good analogue console.  

With  -120 Linear  there is a subtle difference to the final cut off,  but this is pretty much a matter of preference. With -60 you can make subtle changes to the fade length to make the fade finish at the end of a phrase or note, but with -120 this is far more difficult as you can't really  determine where the sound will appear to finish and this will vary far more depending on background noise levels or distance from the speaker. 

I have attached an audio example to this post which illustrates this difference rather well. This is the last 4 seconds of a 10 sec fade first at -120 and then at -60 minimum levels

Here are the waveforms for the example, with the amplitude greatly exaggerated

With S curves the difference is more marked. 

With  -60dB Min the fade is perceived to last pretty much the fade duration set,,

With  -120dB,  when the fade gets to the flatter portion at the end of the S curve, you are pretty  much in inaudible territory, so the fade duration appears shorter and again the end of the fade is musically  indeterminate.

MIc

[micpool]

On Wednesday, November 23, 2022 at 4:16:00 PM UTC micpool wrote:

(I have also attached an AAC compressed version to my next post, so you can decide whether it's worth downloading the uncompressed versions)

Here it is

 

[Maik Waschfeld]

It even works as expected, if 

setLevel cue „1" row 0 column 0 db -1000

getLevel cue „1" row 0 column 0

With kindest regards…
…Maik Waschfeld

(sent from my MBPro16,2)
also at <mailto:Maik.Wa...@Staatstheater-Stuttgart.de>

--
Contact support anytime: sup...@figure53.com
Follow QLab on Twitter: https://twitter.com/QLabApp
User Group Code of Conduct: https://qlab.app/code-of-conduct/
---
You received this message because you are subscribed to the Google Groups "QLab" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qlab+uns...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/b462fe5f-2151-4351-8803-6c2023e631c4n%40googlegroups.com.

Mit freundlichen Grüßen…
…Maik Waschfeld

(gesendet von meinem MBPro16,2)

[micpool]

On Tuesday, November 22, 2022 at 4:05:35 PM UTC micpool attempted to correct  himself :

The likely  result of using very low values for the minimum  means fade times will have to be EXTENDED considerably..

I was a bit confused when I did the tests that the levels at the midpoint of a fade to -120 min and a fade to -60 minimum were only a couple of dB different and that even the last few seconds of a 10s fade were not dramatically different.

I then realised that I had, in the 16 years I have been using QLab that I had never  used  the Fader Curve Audio Domain setting in anything other than it's default mode of slider.

In fact I had  completely forgotten that other audio domains existed in fades!. (Does anyone actually have their audio domain set to anything other than slider in their standard show template settings?)

Here's the explanation of the domains from the Manual:

The Audio Domain pop-up menu lets you choose the scale that QLab uses to fade audio levels. This pop-up is only relevant to fading audio levels; it has no effect on any other parameters, and no effect on audio levels when they are not being actively faded. There are three options for the audio domain:

• Slider domain. The slider domain emulates the design of physical sound consoles, maximizing the useful range of audible levels and making a straight fade sound as smooth and natural as possible.
• Decibel domain. The decibel domain uses a logarithmic scale.
• Linear domain. The linear domain uses a linear scale.

Here's a graphical illustration of the effect of the Fader Curve Audio Domain settings.

If you were using  Decibel domain fades, with a -120dB minimum audio level,  you would literally have to double your fade lengths to make them sound similar to  fades with a  -60dB minimum audio level.

So, in the absence of tons of posts complaining that 10 second fades sound like 5s, I can only conclude that almost all users are using fades in the slider domain, and therefore any audible differences between -60 and -120dB minimum audio levels in settings, are  extremely subtle!

I also conclude that the defaults Chris set when designing the fade cue and workspace settings , (S curve, Slider domain, -60dB minimum audio level) are a really good starting point  for most users.

Mic

[Rich Walsh]

I think you need to define what “as expected” means…

I don’t think it’s surprising that if you set a level below the “Min Volume Limit” then the level that is actually set is the “Min Volume Limit”, so when you query it you get the “Min Volume Limit”.

The situation I was excited to finally be able to address properly is if you want to set a level of “-inf” without knowing what the “Min Volume Limit” is and find picking an arbitrary number that you assume is larger than the end user has chosen for “Min Volume Limit” somewhat unsatisfactory.

A more fascinating can of worms to explore is what happens if you do set the “Min Volume Limit” to -1000 and then start typing in the Inspector. I played with this in 4.7 & 5.0.8 (sorry, not gone to 5.0.10 yet). I wasn’t able to nudge the slider in such a way to get a definitive lowest value before the text field blanks out, but saw numbers around -200 pop up before vanishing. If you type in -999 it goes blank – but if you type in -950 first, you can type in -999!? If you type in -900 you get -897.1… Probably best to stay out of those woods!

Rich

[Rich Walsh]

I did a load of work on fade shapes years ago and have subsequently included pictures of them in my templates, possibly since v2. My default curve in v4 is handles-to-the-corners in the Decibel Audio Domain as that looks like a Pro Tools equal-power S-curve (if about 75% shorter in duration).

I haven’t put a lot of mental effort into gain structure since first using QLab – 13 years ago. I was very used to how ugly the tail ends of SFX 5.6 fades were, so I adopted Gareth Fry’s approach of effectively pre-heating all fades to the noise floor – which translates in QLab to playing pink noise out of your loudest output, fading it down until you can’t hear it and then setting that level as the “Min Volume Limit”. I remember hearing the fade snap out if I didn’t change from the default of -60. If I want a fade to snap out on a button, then I’ll program a button rather than have all my fades truncate like this.

For me, the comparison with analogue faders isn’t useful as:

1. When I’m mixing, I naturally shape the fade in response to the action so a computerised fade of fixed shape and duration is already a different beast
2. Most of my critical listening to fades has been in Pro Tools, so that’s my point of reference and the language to which I translate QLab to quickly get the results I’m looking for

In terms of gain structure, the systems I first worked with were configured around digital consoles and the headroom for vocals and live bands. Slotting into that digitally I found that a master level of -20 is a good starting point for most content I generate, so maybe with a “Min Volume Limit” of -90 I am using the 70dB of dynamic range you outlined earlier anyway?

My quick logic – which I haven’t had audible cause to revisit – was also along the lines of:

• I have an E12-D capable of creating an SPL of 133dB at 1m, so call that 124dB at 8m – the closest seat
• Theatre is NR30, so I need 124-30 = 94dB of dynamic range
• Treating 0 on a QLab fader as the top of that range (reserving the positive gain for boosting the odd quiet piece of content) translates to a “Min Volume Limit” of -94, roughly what I found by experiment at the time

I didn’t consider summing or what was fading to what as the dynamic range for one sound at 0dBFS is the same as that for two incoherent ones at -3dBFS, and so on. If I play my loudest sound as “full” through the system I’ll hit the end stop of the loudspeaker; at the other end my quietest sound at -90 is just at the noise floor of the theatre. Fades will do what I want without having to worry about what else is playing, talking or fanning at that precise time – I just have to remember that a 10s fade out is in reality inaudible at 7.5s (as a result of the fade curve NOT the final end level), but it does have the shape I like.

As I say, it’s not something I’ve felt the need to put any more effort into since adopting this approach. Next time I’m with a system I will see if I can still hear the fades truncate at the default “Min Volume Limit” of -60.

Rich

On 24 Nov 2022, at 01:27, micpool <m...@micpool.com> wrote:

On Tuesday, November 22, 2022 at 4:05:35 PM UTC micpool attempted to correct  himself :

The likely  result of using very low values for the minimum  means fade times will have to be EXTENDED considerably..

I was a bit confused when I did the tests that the levels at the midpoint of a fade to -120 min and a fade to -60 minimum were only a couple of dB different and that even the last few seconds of a 10s fade were not dramatically different.

I then realised that I had, in the 16 years I have been using QLab that I had never  used  the Fader Curve Audio Domain setting in anything other than it's default mode of slider.

In fact I had  completely forgotten that other audio domains existed in fades!. (Does anyone actually have their audio domain set to anything other than slider in their standard show template settings?)

Here's the explanation of the domains from the Manual:

The Audio Domain pop-up menu lets you choose the scale that QLab uses to fade audio levels. This pop-up is only relevant to fading audio levels; it has no effect on any other parameters, and no effect on audio levels when they are not being actively faded. There are three options for the audio domain:

• Slider domain. The slider domain emulates the design of physical sound consoles, maximizing the useful range of audible levels and making a straight fade sound as smooth and natural as possible.
• Decibel domain. The decibel domain uses a logarithmic scale.
• Linear domain. The linear domain uses a linear scale.

Here's a graphical illustration of the effect of the Fader Curve Audio Domain settings.

<SliderDecibelLinear.png>

[micpool]

Hi Rich

Several users have told me that their default is Decibel Domain with a corner curve, as in your description (and probably based on your templates and posts), involving the handles being placed in the top corners. This all seems to go back to QLab 2 (or possibly even QLab 1) where the only fader domain was digital, and the corner curve method was adopted by some users as the best sounding fade.

As we have gone so far down the rabbit hole, I thought it worth completing my fade curves plot to include the QLab 2 curves, both default and corner curves, the  Decibel  domain corner curve in QLab 5, and the QLab5 S Curves in Slider Domain, all curves at -120 and -60dB minimum level settings:

 

This is a 10 second fade of a 1KHz Sine Wave.  The levels at the midpoint (5s) are shown in red and also on the graphic meter display. . The rough point where the fade reaches -60dB is indicated by the purple arrow on each waveform, and the right hand side waveforms show a y magnification of the ends of the fades, from -50dB to the minimum audio level setting.

The first thing I noticed is, that with the exception of the QLab 5 Default Decibel Domain Curve at -120  and the default QLab2 curve at -120,  all the curves reach or pass through  -60 within the last second or so.

The main variations are the speed of cut off, and the level at the midpoint of the fade.

The thing I found most interesting was the closeness of the Default QLab 5 settings, (slider domain/S Curve/ -60 minimum) to the digital domain/corner curve/-120dB plots of QLab 5 (identical to  QLab 2). They both have an overall S shape, are close enough to make little difference  at the midpoint, and have a nice end shape.

The adjustment of the minimum audio level in settings  seems to have the least effect in Slider Domain The midpoint level is with a few dB  and the effective audible fade time is within 10 percent of the duration set. The faders sound good, and variations of the  minimum level setting  mainly just affect the sound of the very end of the fade, in a subtle way that pretty much comes down to personal preference.

On this basis I personally think that the current defaults in QLab 5 (Slider domain, S curve, -60dB minimum setting) are a good general purpose starting point for most users, and can be modified with quite small changes to the minimum level setting, or the curve.

(The manual has other specific recommendations about switching to linear domain for crossfades between cues or crossfades between space loudspeaker systems, and I haven't looked at these)

Mic