[QLab] Qlab & how it handles gain & bits
*
has to do with adding gain. For example if you take a 16 bit file that has
levels up to 0 & add more gain, you're making the word longer & so
something has to give.
In the case of Metric Halo hardware, it's natively doing calculations at
either 24 bits or higher (up to 80 bits within plugins as I understand it)
OK. So what does this have to do with Qlab?
If I put a wave file into Qlab that has levels up to 0 that is a 16 bit
file & I go to the fader & add 12 db of gain, what happened to the word
length?
Maybe core audio handles everything & so there is some sort of bit
reduction going on automatically but I'd like to understand how best to
approach things & whether nor not this is even an issue.
I had a show the other night where the song was "Bolero" & as anyone who
has heard this song knows, it starts very soft & gets progressively louder
& louder until it peaks at the last few drum hits.
The student had boosted the gain in Qlab 12db above 0 to make the
beginning loud enough.
But then as the track got louder & louder we were forced to lower the
output of the Mac (running off headphone output) to keep from over driving
the console input. It still sounded distorted but we got thru that show.
Before the next evening, I attempted to build a fade group to counter the
built in rise in volume but it didn't sound very good.
So I just had the student put the gain back @ 0 & start the console fader
to +12 & manually pull it down to 0.
So maybe really, it's best to not add gain to a file that already has
peaks @ 0 & especially not to files that (insert any modern pop song) are
brick walled to begin with.
One of the songs in the dance show was "Blue Suede Shoes" performed by
Elvis. The track the teacher was using was distorted so I downloaded the
tune off Itunes. Distorted!
I checked out the 2007 Elvis CD & it was also distorted. If I'd had time,
I would of gone hunting for the LP which I would certainly hope is free
from distortion.
Maybe Qlab 3 could have a distortion removal tool:(
One more thing.
The cue list I inherited to fix had a mess of, AIF, WAVE, MP3, M4A, etc...
all mixed together. I know that Qlab has to do some sort of conversion in
order to handle some audio formats. Is there any sonic benefit to
converting files to either AIF or WAVE or is it just a matter of
processing & lag?
*
________________________________________________________
WHEN REPLYING, PLEASE QUOTE ONLY WHAT YOU NEED. Thanks!
Change your preferences or unsubscribe here:
http://lists.figure53.com/listinfo.cgi/qlab-figure53.com
Follow Figure 53 on Twitter here: http://twitter.com/Figure53
Christopher Ashworth
>
> If I put a wave file into QLab that has levels up to 0 that is a 16 bit
> file & I go to the fader & add 12 db of gain, what happened to the word
> length?
QLab performs format conversion as the first step in playing back audio.
Before samples are fed through any other step in the chain (i.e. volume changes, etc), they are converted from the file format to the output device format.
The audio leaving QLab is therefore 32 bit floating point, Linear PCM audio, with a sample rate matching whatever was set on the device.
Thus, if you have a 16 bit file, the first thing that will happen before any level changes are calculated is that the 16 bit data will be translated to a 32 bit representation.
> The cue list I inherited to fix had a mess of, AIF, WAVE, MP3, M4A, etc...
> all mixed together. I know that QLab has to do some sort of conversion in
> order to handle some audio formats. Is there any sonic benefit to
> converting files to either AIF or WAVE or is it just a matter of
> processing & lag?
I'm not sure how much the perceptive quality might differ if using different (non-Apple) code to do the conversion. It's an interesting question. The primary candidate for a possible difference would be in changing sample rates.
I suspect if a difference is audible it would only be when performing conversions between significantly different sample rates, such as up-sampling from a very under-sampled recording. But if that's true, at that point the problem is not really the conversion routine, but the fact that the source material doesn't have enough information.
-CA
Jeremy Lee
Even if QLab is using floating-point calculations internally, at the final stage it's getting turned into fixed point/ PCM data. Floating point audio can have some extreme dynamic range (thousands of dB between the noise floor and distortion). Fixed point is a fixed ceiling and floor (96dB for 16 bit audio). A CD is fixed, and so is the audio being sent to the sound card. When you convert a 16bit sound to 24bits, you get detail in the LOW end, not more headroom in the high end of the dynamic range.
QLab is not the place to make up for a sound system that is not loud enough.
--
Jeremy Lee
Sound Designer, NYC - USA 829
http://www.jjlee.com
On Jan 12, 2011, at 6:35 PM, * wrote:
> The student had boosted the gain in Qlab 12db above 0 to make the
> beginning loud enough.
>
> But then as the track got louder & louder we were forced to lower the
> output of the Mac (running off headphone output) to keep from over driving
> the console input. It still sounded distorted but we got thru that show.
________________________________________________________
*
> QLab is not the place to make up for a sound system that is not loud
> enough.
Re: A sound system that is not loud enough
The Bolero track being discussed starts @ -33 (when the dancers need to
know it's playing) & it ends at 0 on the last big peak. The school PA
system is adequate for what they need to do. It has no problem pumping out
the pop music (some of which is extreme rap). What it can't do is
magically boost the beginning of a track & then bring the gain down over 5
minutes smoothly. This must be done either manually or in Qlab & obviously
the PA isn't the problem.
Since Qlab is better at consistently performing a 5 minute volume
adjustment once the initial programming is done than multiple students who
don't know the music, that is what the goal was after I realized the
situation. The Bolero track was labeled "Dance Co II" or else I would of
known in advance what was going to happen & taken measures to correct it.
Re: Qlab & gain
Qlab allows for "adding" gain in multiple locations which all add up.
Specifically,
Workspace preference main audio tab = up to +12 on master & +12 per channel
Edit device tab = +12 on master & +12 per channel (with pro audio license)
cue levels tab = +12 on master & +12 per channel
trim tab = +12 on master & +12 per channel
No one that knows what they're doing is going to add gain at all these
places but it IS possible & certainly likely that someone will add some
gain at one of these places which is why I asked my first question. What
happens inside Qlab when you do? Chris answered this question generally
which is helpful.
Re: Qlab free
There are people using Qlab that know next to nothing about sound, qlab,
audio file formats, PA headroom, word length, clipping, sound design,
limiting, etc... & since those users are Figure 53s future clients, I
think it's important & relevant to be clear about things such as gain &
also how best to format files & what difference it actually makes when you
don't pay attention to these things. This is why I wrote referencing a
"student experience" so that maybe other non professionals can also learn
something useful.
Re: To adjust or not adjust volume in Qlab
For a dance concert that pulls music from classical, rock, pop & rap,
there will always be a need for "balancing" out the cues between each
other. Again this has nothing to do with the PA & isn't the PA's fault.
It's also not the choreographers fault for choosing more than one style &
dynamic of music. It's the sound persons job to deal with the audio they
are given & the PA they have as best they can.
Re: Bolero
This isn't the first time I've heard a choreographer ask for Bolero to be
boosted at the top & reduced at the end. Instead this is ALWAYS the case.
It is the nature of the recording & piece. The dancers can't hear the
beginning unless it's boosted & the end is too loud unless it's reduced
back to unity. Doing so defeats the natural dynamics of the piece but so
be it. It's a dance concert & not a live music concert.
*
Eric Lott
(mobile)
Jeremy Lee
--
Jeremy Lee
Sound Designer, NYC - USA 829
http://www.jjlee.com
On Jan 15, 2011, at 5:08 PM, * wrote:
> The school PA
> system is adequate for what they need to do. It has no problem pumping out
> the pop music (some of which is extreme rap). What it can't do is
> magically boost the beginning of a track & then bring the gain down over 5
> minutes smoothly.
________________________________________________________
Steve Lalonde
sam kusnetz
Jeremy Lee wrote:
> QLab is not the place to make up for a sound system that is not loud
> enough.
Ra Byn replied:
> The Bolero track being discussed starts @ -33 (when the dancers need to
> know it's playing) & it ends at 0 on the last big peak. The school PA
> system is adequate for what they need to do. It has no problem pumping out
> the pop music (some of which is extreme rap). What it can't do is
> magically boost the beginning of a track & then bring the gain down over 5
> minutes smoothly. This must be done either manually or in Qlab & obviously
> the PA isn't the problem.
if the PA is not performing as you want it to, then obviously the PA *is* the problem. of course it cannot "magically" do anything, but it's fully possible that it plays back loud pop or rap music acceptably but is still inappropriately configured. ultimately my position is that it is indeed inappropriately configured, because it is not doing what you want it to do. QED.
to get us where we want to go, i think a brief discussion of "0 db" is in order.
(i'm going to skip right over the various definitions of db since we're talking only about db as a measure of amplitude of a digital sound recording, ergo dbfs. for more info, google.)
if a digital recording contains peaks at 0 db, then adding gain will produce distortion. if you are inclined to disbelieve that, perhaps you are right now saying to yourself, "but i've done it all the time! i have a normalized track with peaks at 0, and i play it back, and it's too soft so i turn it up and it sounds fine. sam must be wrong." and to be sure, it does often look like that. but what you'll find if you look closely enough is that in those situations, there is at least one element in place that is attenuating the level such that when you pump it up somewhere else, you're not actually going above 0 db.
but let's look at the available gain stages:
1. the track itself,
2. the various level controls on the sound cue (levels and trim tabs),
3. the various level controls in the workspace preferences,
4. the settings of the hardware device (the simplest case is using the headphone jack on your mac; when you do this, the volume F keys on your keyboard become a grandmaster for all output from the mac),
5. the preamps/gain/trim and faders on the inputs of your mixer which are being fed from your audio hardware.
if you make a recording which is a sine wave at 0db in an aiff file, and you really and truly have *all* the level controls from steps 2 through 5 at unity (0db), *and* you have the level controls on your amplifier appropriately set to match the output voltage of your mixer (google a free program called gainset if you don't know what i mean by that) then i *promise* you, turning anything up will produce distortion. if you have big amps, it will be loud distortion. if you have small amps, then it will be quiet distortion.
so when the beginning of your track is a -33 and is too soft to be heard well in a noisy auditorium with a stage full of dancers, of course i understand that you need to turn it up. but there's just no getting around the impenetrable barrier of 0 dbfs. distortion cannot be avoided, and it has nothing to do with qlab, the mac os, core audio, your interface, or anything else. 0 is defined as "as loud as possible without clipping."
the only answer here is to set up the various gain stages in your sound system so that the final peak of the track, ostensibly the loudest moment in your show, is as loud as you want it from an artistic perspective without getting distortion. if your gains from steps 1 - 5 above are all truly at 0db and it's not loud enough, then you need bigger amps.
if your gains from steps 1 - 5 are all truly at 0db and you are getting distortion, then either your amps are not set correctly and the output from the mixer is clipping the inputs of the amps, or your amps are putting out too much power for your speakers. turn it OFF and get burlier speakers.
this may sound preachy and long winded, but i swear to god, i go through all of that every time i put a sound system together, and it's why i don't have gain problems. (well, not true. sometimes i have gain problems. but never with qlab, only with wireless mics.)
i was taught that use of the positive range of a fader was never a good idea. the logic is this: adding gain *always* adds noise. so if you're going to add gain, only do it inside the piece of equipment that is specifically designed to add gain: the amplifier. while it's technically true that the EQ level pots on your mixer control little tiny amplifiers, there's just no chance that those little tiny amplifiers are cleaner than your actual power amps (well, maybe on a cadac or a midas.) so don't do it.
ergo, i just don't use positive gain values in qlab. at all. if i bring something up to 0 and it's too quiet, then i re-engineer the rest of the sound system to allow for more level, and then adjust the rest of my cues down quieter.
if i were in your position, i would set the top of bolero at 0 db in qlab, and then adjust the mixer and the amplifiers until i liked how that sounded. then i would program in a fade cue that slowly brought the master level of the bolero cue down over the course of the song so that as the track got louder, it's level reduced.
if i got flak from anyone about the inevitable huge amount of time this would take to get just right, i would ask them how many hours the dancers rehearsed, and then tell them to leave me alone. ra byn, i know you often work with completely unreasonable people, so i do not envy you this conversation.
and this brings me to my next and final point, which dovetails with something you wrote:
> Re: Qlab free
> There are people using Qlab that know next to nothing about sound, qlab,
> audio file formats, PA headroom, word length, clipping, sound design,
> limiting, etc.
while what you say may be true, it is not good. tools like qlab (and sfx, and pro tools) make it very, very easy to do what was once very, very difficult to do. as the quality and complexity of our tools develops, the minimum amount of knowledge required to operate these tools gets lower and lower. that by itself is a good thing: more people are able to make the art that they want to make. that's wonderful. but by creating these hugely powerful tools, we have created a new problem: it's easier to get in over your head than it was before. with 1/4" open reels, you *had* to know a bunch of technical crap before you could get anything to happen at all.
what i'm saying is that we need to LEARN. i've been doing more projection lately, and every new step requires more knowledge. why can't i render a video with a transparent background? well, turns out only certain formats support transparency. what are the required steps to make use of the transparency? well, it's x, y, and z.
> Re: To adjust or not adjust volume in Qlab
> For a dance concert that pulls music from classical, rock, pop & rap,
> there will always be a need for "balancing" out the cues between each
> other. Again this has nothing to do with the PA & isn't the PA's fault.
> It's also not the choreographers fault for choosing more than one style &
> dynamic of music. It's the sound persons job to deal with the audio they
> are given & the PA they have as best they can.
i don't think anyone who knows anything about sound would argue otherwise.
> Re: Bolero
> This isn't the first time I've heard a choreographer ask for Bolero to be
> boosted at the top & reduced at the end. Instead this is ALWAYS the case.
let's be fair. i saw oregon ballet theater perform to bolero. it was being played by the symphony; no adjustment necessary. pieces like bolero remind us that the dynamic range of the systems we build is still humbled by the dynamic range of the human ear. and i believe that even that is dwarfed by the dynamic range of the imagination.
the most mundane and simplest answer to playing back a canned bolero is probably the best one: compress the damn track.
if i have made any mistakes in this post, please accept my apologies and please correct me. i am proud to say that i am at that level of knowledge wherein i know how little i know. also much of what i know was self taught or learned through deductive processes of trial and (noisy) error. also, i just noticed it's 3 am. so i maybe made some sleepy errors. if so, again, apologies.
kind regards
sam
Jeremy Lee
--
Jeremy Lee
Sound Designer, NYC - USA 829
http://www.jjlee.com
On Jan 16, 2011, at 2:57 AM, sam kusnetz wrote:
> ergo, i just don't use positive gain values in qlab. at all. if i bring something up to 0 and it's too quiet, then i re-engineer the rest of the sound system to allow for more level, and then adjust the rest of my cues down quieter.
________________________________________________________
*
the app.
This is NO indication that users should be running @ levels much lower
than that & never going past 0.
Assuming this is correct, Qlab should logically default to a lower master
level without any involvement from the user & not allow gain past 0. Even
better, throw away 20db behind the GUI so everyone can stay @ 0 (really
-20) & add 12 if they like & never know the difference but not suffer the
consequences.
I don't see how it can be universally applied without adding additional
noise in some if not all systems. This also ignores the matter of source
audio material for which you don't always have the time nor the tools to
resolve / control.
I'd like to hear what others have to say about this, especially Figure 53.
*
> On Sun, January 16, 2011 8:57 am, Jeremy Lee wrote:
> Yup. Nominal master QLab level of -20, and if I get above -10 on a cue,
> I start to question the system design.
&
Christopher Ashworth
>
> I'd like to hear what others have to say about this, especially Figure 53.
FWIW,
I think QLab should not change the volume of a file unless told to do so.
I think people work, and think, in different ways, using source materials of variable characteristics.
I think software should realize the thoughts of a human, rather than replace the thoughts of a human.
I think a tool can make a professional's job easier.
I think a tool can not make a professional.
-C
Andy Leviss
<ch...@figure53.com> wrote:
> I think a tool can make a professional's job easier.
>
> I think a tool can not make a professional.
Amen. These aren't things that are unique to QLab. These are
fundamental basics of *any* digital audio application, including
digital consoles, QLab, SFX, ProTools/Logic/any other DAW, etc. If you
try to add digital gain to audio that is already maxed out to digital
clip, you're going to have problems.
Heck, it's not even new to digital, if you tried to add analog gain
past the point of distortion, it's gonna eventually start to sound
nasty.
At what point do we transition from what an application/its
documentation needs to teach you to what is considered an essential
fundamental that is a reasonable assumption one would learn before
getting into the intricacies of a specific application/device?
Also, am I mistaken, or don't the level meters in QLab *tell* you when
the signal is in digital clip, so you can adjust if you're driving it
too hard? If you don't know that red=clip=bad, I'm not really
convinced it's QLab's fault. That's a fundamental basic that anybody
running a piece of sound equipment should know, before they learn
anything else. Green=go, yellow=slow, red=stop. If you can drive, you
can get a halfway passable gain structure going.
FWIW,
Andy
Jeremy Lee
It's easy enough to make up for poor source material with some quick editing. Don't expect your playback tool to fix bad audio.
Jeremy Lee
- A thumb is a terrible speller. Please forgive my trespasses.
On Jan 16, 2011, at 10:02 AM, "*" <ra...@rabyn.com> wrote:
> Except Qlab defaults to 0,0,0,0,0,0,0,0 & allows +12 @ each fader thru out
> the app.
>
> This is NO indication that users should be running @ levels much lower
> than that & never going past 0.
>
> Assuming this is correct, Qlab should logically default to a lower master
> level without any involvement from the user & not allow gain past 0. Even
> better, throw away 20db behind the GUI so everyone can stay @ 0 (really
> -20) & add 12 if they like & never know the difference but not suffer the
> consequences.
>
> I don't see how it can be universally applied without adding additional
> noise in some if not all systems. This also ignores the matter of source
> audio material for which you don't always have the time nor the tools to
> resolve / control.
>
> I'd like to hear what others have to say about this, especially Figure 53.
>
> *
>
*
How are the meters calibrated & at what level above 0 does red occur?
Is it possible that in the future, there could be a reset-able hold
function, an over indicator & some indication on the meters to show where
you're levels are at?
*
On Thu, January 13, 2011 6:11 am, Christopher Ashworth wrote:
> QLab performs format conversion as the first step in playing back audio.
>
> Before samples are fed through any other step in the chain (i.e. volume
> changes, etc), they are converted from the file format to the output
> device format.
>
> The audio leaving QLab is therefore 32 bit floating point, Linear PCM
> audio, with a sample rate matching whatever was set on the device.
>
> Thus, if you have a 16 bit file, the first thing that will happen before
> any level changes are calculated is that the 16 bit data will be
> translated to a 32 bit representation.
________________________________________________________
Christopher Ashworth
>
> How are the meters calibrated & at what level above 0 does red occur?
It's a gradient, so specific colors don't kick in at specific places, but it's yellow 'round about 0db and fades to red above that.
> Is it possible that in the future, there could be a reset-able hold
> function, an over indicator & some indication on the meters to show where
> you're levels are at?
I'm not sure a lot more can be fruitfully pushed into the space the current sliders occupy. I remember experimenting with a few representations, and finding the relatively bare "colored bar in the slider" approach to be the best way to provide some information without getting too cluttered.
Andy Leviss
<ch...@figure53.com> wrote:
> On Jan 16, 2011, at 4:11 PM, * wrote:
>>
>> How are the meters calibrated & at what level above 0 does red occur?
>
> It's a gradient, so specific colors don't kick in at specific places, but it's yellow 'round about 0db and fades to red above that.
In that case, can I humbly suggest that this is incorrect and
non-standard, and while the yellow point is open to debate, it should
hit red *just* before it hits digital 0, so that you have some warning
a hair before digital clip? So the quickest tap into red is ok but you
need to keep an eye on it, but steady red means you're likely in clip,
and should back it up.
--Andy
*
reaching 0db at the wave level & you add even .1 of gain in Qlab, you are
OVER.
To be clear, this is why I asked my first question. To find out if Qlab is
able to elongate the word & then dither back down @ the output without
becoming distorted.
So taking this information along with Jeremy & Sam's posts in to account,
I have to think that the responsible way of adjusting gain is to do some
metering in advance of programming in Qlab, figure out what is your
loudest cue & then adjust accordingly.
Since a person with a Mac, QLab free & no audio interface (other than the
headphone output) or an I/O without any hardware or software meters such
as the Presonus Firebox, has no "real" meters, if there is way to show an
OVER on the existing meter, I think that would be useful. I think it's
confusing for a meter to transition from green to yellow before reaching 0
& then stay yellow long after the audio has been compromised, I've always
presumed that there was still some room left until the red light came on
the Qlab meter came on. Obviously not.
Thanks to Chris, Jeremy, Sam & Andy for pushing thru this.
Best Regards,
*
On Sun, January 16, 2011 3:50 pm, Christopher Ashworth wrote:
> On Jan 16, 2011, at 4:11 PM, * wrote:
>> How are the meters calibrated & at what level above 0 does red occur?
> It's a gradient, so specific colors don't kick in at specific places,
but
> it's yellow 'round about 0db and fades to red above that.
>> Is it possible that in the future, there could be a reset-able hold
function, an over indicator & some indication on the meters to show where
>> you're levels are at?
sam kusnetz
> Except Qlab defaults to 0,0,0,0,0,0,0,0 & allows +12 @ each fader thru out the app.
you can set your own default sound levels in the audio section of the workspace preferences.
you can also set the maximum level that qlab permits. it defaults to +12 but you can make it 0 if you want.
the fact that qlab allows +12 speaks directly to one of your later points: you might not always have the tools or time available to correct the level of a track, so having +12 available can be handy.
> This is NO indication that users should be running @ levels much lower than that & never going past 0.
that is because it is not a qlab issue, it's an education issue. there's no sticker on your car that says don't run a red light either.
> Assuming this is correct, Qlab should logically default to a lower master
> level without any involvement from the user & not allow gain past 0.
no, qlab should not do this. tools should not be responsible for their misuse. I think this is especially important in a teaching environment. teaching a kid to run a show is fine, but teaching a kid how it all works is so much better.
> Even better, throw away 20db behind the GUI so everyone can stay @ 0 (really -20) & add 12 if they like & never know the difference but not suffer the consequences.
this is even worse! you want your tools to lie to you? why? just learn how everything works and then you won't need to be protected like this.
I don't want qlab to save me from myself, I want it to do exactly what it says it's doing. if there's a problem, I want to learn not just how to fix it, but how to prevent it from occurring again in the future. if the tool has invisible safeguards, then I'll never learn.
cheers
sam
Andy Leviss
> Via Spectra Foo & a MH 2882, I have verified that if a file has material
> reaching 0db at the wave level & you add even .1 of gain in Qlab, you are
> OVER.
This is as any digital gear should respond. That's what 0 dBFS is.
Full scale. Clip. I'm confused why you'd expect otherwise.
> To be clear, this is why I asked my first question. To find out if Qlab is
> able to elongate the word & then dither back down @ the output without
> becoming distorted.
How would this be possible? It wouldn't be getting any louder. 0 = 0 =
0. If it had float room above 0, and were able to go into a higher bit
depth internally to increase level, at the point of bringing it back
down to the depth of the output device, it would lower the level.
Unless I'm missing something, you're asking for magic.
Recording bit-depth increases low level resolution, not high. Floating
calculation bit depth increases headroom for *internal* calculations,
but going back to a lower level, it will only provide a gain in
accuracy, not a gain in level.
A slightly confused,
Andy
Andy Leviss
>
> Recording bit-depth increases low level resolution, not high. Floating
> calculation bit depth increases headroom for *internal* calculations,
> but going back to a lower level, it will only provide a gain in
> accuracy, not a gain in level.
Clarification...that last part should read:
"but going back to a lower DEPTH, it will only provide..."
Jeremy Lee
--
Jeremy Lee
Sound Designer, NYC - USA 829
http://www.jjlee.com
Dave Tosti-Lane
to the meters I would also like to see it. Even if it's just a grid
background with several percentage points.
Also useful might be a pre/post switch, or a dual meter which showed pre on
one side and post on the other. Those need not be as wide - the current
space would be plenty big enough, if it just split down the middle and
showed pre on one side and post on the other. To me, the meters are more
useful than the faders themselves, and the fader knobs need not be such
large rectangles. 9 times out of ten, I enter the levels in the numerical
boxes at the top.
Dave Tosti-Lane
On Sunday1/16,Sunday1/16:230 PM 2:30 PM, "Andy Leviss" <an...@ducksecho.com>
wrote:
> On Sun, Jan 16, 2011 at 4:50 PM, Christopher Ashworth
Andy Leviss
> Even
> better, throw away 20db behind the GUI so everyone can stay @ 0 (really
> -20) & add 12 if they like & never know the difference but not suffer the
> consequences.
I glossed over this before. Two issues here. One, what if I know I've
got the headroom, and *want* that boost? You're now artificially
limiting what I can do. There are situations where I may want to apply
gain beyond 0, for example if I know my file is peak normalized well
below 0 and want to correct for that without destructively editing it.
Second, as Sam said, it's easily argued that that's an education
issue. You want a digital meter to represent analog level. But it's
not an analog meter, nor an analog system. It's accepted practice that
0 represents 0 dBFS, not any sort of analog 0.
I will throw a Devil's Advocate curveball out there, that one of the
things that threw me at first when learning the Soundcraft Vi6 is that
its meters are labeled like analog meters. And it's actually not a bad
idea, *provided* it's made clear to the user. It's way better than
shifting standards that other digital console manufacturers use (the
worst being Yamaha, who uses -20 for *most* consoles, but -14 for the
DM2000 and 02R).
But that's not the way the industry as a whole does it, and is the
exception, not the rule. Is that something QLab should blaze a trail
on, in a way that requires explanation to the average user, or is it
something where QLab should follow the accepted standard, and leave
the trailblazing to high end digital consoles like the
Soundcraft/Studer lines?
--A
Christopher Ashworth
>
> In that case, can I humbly suggest that this is incorrect and
> digital 0
Sure thing. Thanks for the correction. Will fix this for the next update.
-C
Christopher Ashworth
> Much more interesting in the matrix is a way to label input/ output points...
And a text-field option for the crosspoint matrix, I'd wager.
-C
Andy Leviss
<ch...@figure53.com> wrote:
> On Jan 16, 2011, at 5:30 PM, Andy Leviss wrote:
>>
>> In that case, can I humbly suggest that this is incorrect and
>> non-standard, and [...] it should hit red *just* before it hits
>> digital 0
>
> Sure thing. Thanks for the correction. Will fix this for the next update.
[music]This is why Chris rocks, this is why Chris rocks{/music]
Theoretically, the expectation is that red = clip, so it should line
up at 0 on the dot. In actual implementation, most hardware/software
designers know that users skirt the line, and that occasionally
tapping into "red" is going to happen. So rather than having it be 0,
and immediately make nasty sounds, there's usually a small degree of
safety margin in there. But not the 12 dB *'s suggesting, something on
the line of 1-3 dB, depending on manufacturer. (And yes, I'm sure we
could debate exactly how much is best for days; really, it's a lot of
semantics, so long as the user is confident that a quick tap into red
is likely not actually hitting clip, but steady red is trouble.)
--A
*
I'd lean the other way. 0 is 0 & red is the indicator. More over = more
red...
It is this historic looseness of digital meters that has created the
confusion this thread is all about.
A meter that is neither calibrated or labeled is a toy. This is one reason
I've thought that unless you see red in QLab, there is something going on
behind the curtain.
Glad to know I am wrong & that it can be fixed.
Andy Wrote:
"Also, am I mistaken, or don't the level meters in QLab *tell* you when
the signal is in digital clip, so you can adjust if you're driving it
too hard? If you don't know that red=clip=bad, I'm not really
convinced it's QLab's fault. That's a fundamental basic that anybody
running a piece of sound equipment should know, before they learn
anything else. Green=go, yellow=slow, red=stop."
Exactly!
If I go to the settings tab & watch the Qlab meters, I want red to mean
"turn it down" & not "I wonder how much more it can handle."
A few samples over, fine. 1 to 3 db over before red is worthless IMO...
Stick to trailblazing Figure 53.
*
P.S. I never purposely suggested allowing for +12db before a meter goes
red. That wouldn't make any sense to anyone including me.
Knowing that Qlab meters currently don't turn red @ 0 is the key to most
of my posts & questions.
Again thanks to those who have been involved in the discussion:)
Andy Dolph
read red 1-3 DB BEFORE they go over - so an occasional flash of red
doesn't actually mean clipping.
Andy
On Sun, Jan 16, 2011 at 10:40 PM, * <ra...@rabyn.com> wrote:
> A few samples over, fine. 1 to 3 db over before red is worthless IMO...
>
Jeremy Lee
--
Jeremy Lee
Sound Designer, NYC - USA 829
http://www.jjlee.com
Sean Dougall
> To be clear, this is why I asked my first question. To find out if Qlab is
> able to elongate the word & then dither back down @ the output without
> becoming distorted.
Andy Leviss responded:
> How would this be possible? It wouldn't be getting any louder. 0 = 0 =
> 0. If it had float room above 0, and were able to go into a higher bit
> depth internally to increase level, at the point of bringing it back
> down to the depth of the output device, it would lower the level.
> Unless I'm missing something, you're asking for magic.
Come to think of it, what * is describing is akin to a limiter, which
means that AU support might, possibly, sort-of-kind-of, allow for this
in v3. All the audio data flying around within QLab is 32-bit floating
point, so there actually is headroom for the signal to go over 0dB and
then get pulled down by a limiter before output. Won't give you any
more output power than your amps have, of course, but it could be an
emergency measure if you're clipping slightly and you don't know why.
(Not that I condone this as a general strategy. You still have to be
just as careful with this setup, or you'll end up with "pumping"
artifacts, and your sound design will sound like an episode of Glee.)
Sean
Jeremy Lee
> All the audio data flying around within QLab is 32-bit floating
> point, so there actually is headroom for the signal to go over 0dB and
> then get pulled down by a limiter before output.
True- I do this all the time in DP by pulling down the master fader or adding a limiter on it's return to 24 bit linear.
> You still have to be
> just as careful with this setup, or you'll end up with "pumping"
> artifacts, and your sound design will sound like an episode of Glee.)
Only if you also know how to abuse AutoTune. Tried watching an episode, and just couldn't take it any more!
--
Jeremy Lee
Sound Designer, NYC - USA 829
http://www.jjlee.com