The No.1 Website for Pro Audio
 All  This Thread  Reviews  Gear Database  Gear for sale     Latest  Trending
Do I need SUM? For you guys who mix ITB and OTB...
Old 27th September 2005
  #31
Gear Maniac
 
kenkelly81's Avatar
 

Your right RKrizman about the numbers that are so big they can't fit into the 24 bit.
I was about to say I've never heard anything so absurd but then I remembered all the other summing threads I've read.

I don't know what people are thinking.

Lets see - even the simplest of old 386 and 486 computers and even pocket calculators can sum numbers.

And what is digital audio but numbers.

Now some computers will do it quicker and possibly more efficiently then others.
But give a modern computer any amount of numbers to sum and it will come up with same answer each and every time.
Given you don't exceed it theoretical limits.

This is a fact.

What your doing with digital audio and your summing bus is SUMMING NUMBERS!
I know this is an over simplified version of what is happening but this is the Cliffs Notes version.
And this is as much as my limited understanding reaches.

Given all that - I have an API 8200 running into the API 7800.
It has made a world of difference on my mixes. I can integrate all sorts of out board very easily into my setup with this and my HD3 and Apogee Rosetta 800's.

I won't be going back to mixing sole ITB anytime soon.
Unless of coarse a client ask for a totally ITB mix.
But given the sonic difference when I do a simple A/B of mix for anyone and they hands down pick the OTB summing. That's with doing the best mix possible ITB matching volumes and then simply running it out on 4 stereo pairs.

Enough about me what will work for you only you be able say for sure.

Just remember One Man's Rubbish Is Another Man's Gold.

YMMV

Oh, another thing to think about is that everyone always tries to defend there purchases/products/gear because it makes us feel better about the fact we spent al that money on said purchases/products/gear.
So consequently it is very hard to get an unbiased opinion out of pretty much anyone.
And so goes.

Time to
Old 27th September 2005
  #32
Gear Maniac
 
kenkelly81's Avatar
 

One thing I forgot to mention, once I read the TL white paper I don't know how long ago I start using much more conservative levels in PT and I think that made just as much of improvement in my mixes as OTB summing did if not more.

It really help to turn things down. At least it did for me.

Old 27th September 2005
  #33
Lives for gear
 

I believe also TC Electronic have some papers on intersample peaks.

But anyway, keeping low levels in the mix does only make sense with PT or fixed point daws, and not with floating point daws, as long as the output (the signal being fed the dac) doesn't clip or have intersample peaks.
Old 27th September 2005
  #34
Lives for gear
Quote:
Originally Posted by juicemaster1500
I believe also TC Electronic have some papers on intersample peaks.

But anyway, keeping low levels in the mix does only make sense with PT or fixed point daws, and not with floating point daws, as long as the output (the signal being fed the dac) doesn't clip or have intersample peaks.
that's the whole point of using oversampling meters. so you don't clip a typical cheapo DAC, which is where 99.9% of your stuff will be heard. for this reason, whether floating point or not, you need the meters. it's not what's happening in the box as much as what'll happen when you leave it!

least that's how i understood it, and it seems (whether imagined or not), everything sounds better, faster.
Old 27th September 2005
  #35
Lives for gear
 
gwailoh's Avatar
 

Quote:
Originally Posted by kenkelly81
Your right RKrizman about the numbers that are so big they can't fit into the 24 bit.
I was about to say I've never heard anything so absurd but then I remembered all the other summing threads I've read.

I don't know what people are thinking.

Lets see - even the simplest of old 386 and 486 computers and even pocket calculators can sum numbers.

And what is digital audio but numbers.

Now some computers will do it quicker and possibly more efficiently then others.
But give a modern computer any amount of numbers to sum and it will come up with same answer each and every time.
Given you don't exceed it theoretical limits.

This is a fact.
It is indeed a fact, but, it's not the issue. The issue is that there's a difference between sample values and reconstructed signal levels post-DAC. It has nothing at all to do with the mix bus or with the addition process inside the DAW!

Here are two paragraphs from the TL white paper which I think do a good job of summarizing the problem:

Quote:
The method used for computing the peak value inside the system however, is not particularly accurate. DAW and digital mixer manufacturers typically take the amplitude of the samples and use these as the basis for the peak meter. The problem with this approach is easily identified: the samples themselves do not represent the peak value of the waveform. The waveform is only complete after the reconstruction process. Until this process has been completed, the waveform is inaccurately represented by the samples. [...]

The consequence of the way in which DAW’s treat waveforms is that the meter inside the DAW or other digital mixers inevitably shows inaccurate information. It is virtually a mathematical certainty that the waveform will actually exceed the amplitude of the samples in any sampling system. The samples themselves only represent a waveform, It is important to understand that the amplitude of the waveform will invariably exceed the sample values.
Hopefully this can help to visualize what happens if you run too hot inside the DAW. A sample can be legal -- for instance, it can represent digital full scale. Suppose though there are several samples in a row which are at DFS. This is legal, but what does it mean to the DAC? Did the waveform actually exceeed DFS, so that these maxed-out samples represent digital clipping? The DAC can't know. Whatever the result is, it'll likely be different than the input which you recorded. Most likely there'll be "inharmonic distortion" produced by the DAC. This in turn contributes to the digital "grunginess", coldness, and stereo narrowing which some people experience when mixing ITB. It's also the reason why a digital limiter based on sample values rather than signal value could actually make the problem worse.

This is why it's important to utilize the fabulous digital "footroom" which 24-bit recording allows. Pulling those faders down inside the DAW allows you to avoid the unmetered distortion described above, without experiencing noise problems. And, it's why it's important to get yourself an oversampling meter plugin!
Old 27th September 2005
  #36
Gear Maniac
 
kenkelly81's Avatar
 

Yea what gwailoh said!

I'm not claiming to be an expert but gwailoh is right or the what paper is right.
It sounds so much better to me when a pull the levels down and record at a less HOT level.

The OTB summing just seems to make thing easier/quicker for me to get a mix done.

And I need all the help I can get.

I also apologize if I offended or sounded condescending in my previous remarks.
I was really tired.

KC
Old 27th September 2005
  #37
Lives for gear
 
gwailoh's Avatar
 

No worries!

One tip which Paul Frindle pointed out on PSW is that you need to lower the levels before they reach any plugins. Since plugins on PT are pre-fader, that means you should insert a trim plugin before any others, and knock your internal level down right there. I use -8db. As Raal is also doing, I'm then using an oversampled meter to check levels going into and out of each plugin. It's a little bit of extra work but this whole learning process about ITB level issues has made me a little paranoid.

--Mark
Old 27th September 2005
  #38
Lives for gear
Quote:
Originally Posted by gwailoh
I use -8db. As raal is also doing
actually, after buying the meter i've used -6dB as a starting point. seems to be enough. but -6 or -8dB really makes little difference with the top and bottom headroom afforded by most digital DAWs these days.

as soon as EQ or whatever plug is introduced, that -6 dissappears very quickly so it's very important to put meters after plugs to check, as gwailoh suggests.

it behooves anyone who is interested in this to read the TL paper. it's short, to the point, graphical and explains the whole deal in apples and oranges fashion. no brainer and priceless knowledge to have when summing anything digitally.

note: whether you add or subtract EQ, that -6dB 'headroom' will disappear very quickly.
Old 28th September 2005
  #39
Lives for gear
 

Quote:
Originally Posted by raal
that's the whole point of using oversampling meters. so you don't clip a typical cheapo DAC, which is where 99.9% of your stuff will be heard. for this reason, whether floating point or not, you need the meters. it's not what's happening in the box as much as what'll happen when you leave it!

least that's how i understood it, and it seems (whether imagined or not), everything sounds better, faster.
Yes this is corrrect, but in my previous post I was refering to the actual mixing process, feeding plugins etc.

Once again, it seems like a fixed point issue, this whole 'keeping levels low' business.
Old 28th September 2005
  #40
Gear Guru
 
u b k's Avatar
 

well i admit my understanding of floating point math is rudimentary; nevertheless, i proceed boldly!

my understanding is that floating point has the ability to deal with massive numbers because it can shift the placeholder and the remainder to accommodate what's needed. summing 16 or 24 24-bit words running near full scale results in a HUGE number. because of the huge difference in scale, the larger your 32-bit floating point sum, the more it needs to be scaled back in order to fit into a 24-bit fixed point word. more scaling back involves greater rounding error.

the net effect of this sounds to me like crowding and a loss of low-level detail. don't take my word for it, try it: run your mix hot, as hot as you can get without lighting the clip lights. bounce it. now stick a trim plug on every track and scale everything back 6db, and pull the master fader back as well. sum that mix, then level match the two. they don't sound the same, and the quieter one always sounds better to me.

again, my explanation may be off, but i stand by my conclusions.


gregoire
del ubik
Old 28th September 2005
  #41
Lives for gear
Quote:
Originally Posted by u b i k
don't take my word for it, try it: run your mix hot, as hot as you can get without lighting the clip lights. bounce it. now stick a trim plug on every track and scale everything back 6db, and pull the master fader back as well. sum that mix, then level match the two. they don't sound the same, and the quieter one always sounds better to me.
and if you put an oversampling meter after each plug to make sure you're not getting any intersample peaks, it'll sound even better.
just grab a dense mix with alot of plugs and place a meter after each plug, see what happens. even if you trimmed the original audio on each track by -6dB, you'll be surprised (i was).

link to TL white paper by nika:

http://www.tllabs.com/index.php?opti...d=20&Itemid=62
Old 28th September 2005
  #42
Gear Guru
 
u b k's Avatar
 

thanks for the links raal and gwailoh, those were great reads.

so, is it too much to hope for a freeware AU oversampling meter?

in the absence of that, is there a dbFS level that guarantees that no illegal waveforms will result before the next stage? iow, if a signal goes into a plug-in at (e.g.) -6dbFS and comes out the other side no hotter, am i safe?


gregoire
del ubik
Old 28th September 2005
  #43
Lives for gear
 
T_R_S's Avatar
Quote:
Originally Posted by raal
and if you put an oversampling meter after each plug to make sure you're not getting any intersample peaks, it'll sound even better.
just grab a dense mix with alot of plugs and place a meter after each plug, see what happens. even if you trimmed the original audio on each track by -6dB, you'll be surprised (i was).

link to TL white paper by nika:

http://www.tllabs.com/index.php?opti...d=20&Itemid=62
i can't download it like there plug-ins late and buggy.
Old 28th September 2005
  #44
Lives for gear
Quote:
Originally Posted by u b i k
so, is it too much to hope for a freeware AU oversampling meter?
think so dude. those things are just too cool.

Quote:
in the absence of that, is there a dbFS level that guarantees that no illegal waveforms will result before the next stage?
without a meter i'd go -8dB as gwailoh suggested, and keep all meters in the green (no yellow) as he also suggested to me. with the meters i find i can go to yellow and be fine, but it's an individidual thing. no set rules.

Quote:
iow, if a signal goes into a plug-in at (e.g.) -6dbFS and comes out the other side no hotter, am i safe?
problem with that (i think) is that by manipulating the plug (whether adding, subtracting EQ, compressing, whatever), the sample count goes way up, and all plugs are not created equal. i asked gwailoh the same question - is there a magic number where one is 'safe'? he basically said 'just get the friggin meter and try it'. i did that. once i tried it out, i understood his suggestion. best $100 i've spent on digital audio.
Old 28th September 2005
  #45
Lives for gear
 

Quote:
Originally Posted by u b i k
the net effect of this sounds to me like crowding and a loss of low-level detail. don't take my word for it, try it: run your mix hot, as hot as you can get without lighting the clip lights. bounce it. now stick a trim plug on every track and scale everything back 6db, and pull the master fader back as well. sum that mix, then level match the two. they don't sound the same, and the quieter one always sounds better to me.
If you use a plugin for this, the signal feeding dynamics plug-ins on buses or auxes or even the master fader will be lower, and you'll get the same effect as raising the threshold on the dynamics plugins = less compression.
Old 28th September 2005
  #46
Gear Addict
 

So to simplfy the conclusion, we should avoid intersample peak by recording and mixing in low levels. Right? Sounds good, in fact I mixed a song this way and it is one of the best in the box mix I did so far.

I mix in Pro Tools and bounce to disk my master then open a new session and import my 2 track master to do some pre-mastering (Phoenix, L3, etc).

But what I'm confused about is, if I do pre-mastering on my 2 track master to bring the overall level back up, wouldn't I be possibly introducing intersample peak to my 2 track master?
In other words, what would be the best way to bring up the level of my master (usually for client copies so this needs to be loud) without introducing intersample peak? Thanks for your help.
Old 28th September 2005
  #47
Lives for gear
Quote:
Originally Posted by juicemaster1500
If you use a plugin for this, the signal feeding dynamics plug-ins on buses or auxes or even the master fader will be lower, and you'll get the same effect as raising the threshold on the dynamics plugins = less compression.
yes, but once you're there you can still compress the crap out of the thing and still come out smelling like a rose, so not the same thing.
Old 28th September 2005
  #48
Lives for gear
 
goldphinga's Avatar
 

What??????

Someone explain this in simple terms.
Old 28th September 2005
  #49
Lives for gear
Quote:
Originally Posted by xqtion
So to simplfy the conclusion, we should avoid intersample peak by recording and mixing in low levels. Right?
WRONG! you record hot as always (with no overs), and you mix at lower levels, always monitoring with an oversampling meter. if you record hot (within reason), and simply reproduce the sound (no plugs or processing), it's all good - i tried it. it's when you manipulate the sound after the fact that it all goes to ****e. that's why the -6dB, or -8 or whatever. READ THE WHITE PAPER.

Quote:
But what I'm confused about is, if I dSounds good, in fact I mixed a song this way and it is one of the best in the box mix I did so far.o pre-mastering on my 2 track master to bring the overall level back up, wouldn't I be possibly introducing intersample peak to my 2 track master?
you betcha!

Quote:
In other words, what would be the best way to bring up the level of my master (usually for client copies so this needs to be loud) without introducing intersample peak? Thanks for your help.
for client copies use your ears, or use an oversampling meter. you'll end up printing somewhat hotter, but not at 0 or -.1, or -.2 or -.3 dB as we love to do.

if you want to get that last tenth of a dB edge and not blow it, get a pro to do it. and if you get a pro to do it, don't do any 'pre-mastering'... specially with plugs. analog compression on the stereo bus is a different matter. i don't call that 'premastering' - it's more like an effect, and can be an integral part of the track. my 2¢.
Old 28th September 2005
  #50
Lives for gear
Quote:
Originally Posted by goldphinga
What??????

Someone explain this in simple terms.
when you hit the plug at a lower level (because the meter didn't allow you to produce intersample peaks) you'll have to set the compressor to a lower threshold in order to get the same effect (because the input would be at a lower level than if you hadn't used the meter) -- but once you lower the threshold, you can squash til yer blue in the face, and not create intersample peaks -- if you're careful that the output isn't too hot for the next plug's input. hope this is somewhat helpful. if not, READ THE WHITE PAPER and you'll see the light, grasshopper.
Old 28th September 2005
  #51
Lives for gear
 

Quote:
Originally Posted by raal
yes, but once you're there you can still compress the crap out of the thing and still come out smelling like a rose, so not the same thing.
Of course you can

I was refering to the 'test' U b i k suggested, doing an a/b test the way he described will change how the respective plug-ins process audio.

Anyway, using floating point daws you can have intersample peaks of a hundred db's (but no dynamic plug, which relates to -0db), but it doesn't matter as long as you pull down the master fader before hitting the dac, or bouncing the file.
Old 28th September 2005
  #52
Lives for gear
Quote:
Originally Posted by juicemaster1500
Anyway, using floating point daws you can have intersample peaks of a hundred db's (but no dynamic plug, which relates to -0db), but it doesn't matter as long as you pull down the master fader before hitting the dac, or bouncing the file.
yes sir. AFAIK the same applies to a DAW like PT HD (not floating point). in their white paper, digi states it's pretty much impossible to 'overload' the stereo bus. don't know enough to affirm or refute this but the fact is, in any DAW, even if there are no 'overs' as viewed by the DAW's sample meters, you can very well be creating intersample peaks that may be able to be reproduced by high quality DA converters with little or no distortion; but as soon as these same peaks are reproduced by consumer DACs.... ooops!

so the inherent problem isn't what's happening in the DAW (floating point or otherwise) as much as what will happen when DACs try to reproduce the final result.

if you pull the master fader down in PT before hitting the DAC, same story. but if you were producing intersample peaks before getting to the master fader, you've got problems, even if no theoretical peaks are produced after lowering the master fader. it's getting late, don't know if that even made sense to me!

i believe what you're trying to say is that in a floating point DAW, it's impossible to produce intersample peaks before the master fader. well in a double precision math DAW (PT HD), it's definitely possible - i've seen it clearly.

the sure way of finding out though, is by using these meters. of that i have no doubt.
Old 28th September 2005
  #53
Lives for gear
 
gwailoh's Avatar
 

Quote:
Originally Posted by raal
so the inherent problem isn't what's happening in the DAW (floating point or otherwise) as much as what will happen when DACs try to reproduce the final result.
What he said.

The whole problem is that peak sample level and reconstructed signal level (reconstructed by the DAC somewhere downstream) aren't the same thing. DAW summing, which is just arithmetic, FP or otherwise, can produce perfectly legal sample values which will nevertheless cause reconstruction errors in the DAC. It's all about the DAC, not actually the DAW at all.

So what you're doing with those oversampling meters is simulating the reconstruction process which will take place in the DAC. This is still an approximation because you can't really know in advance precisely what a cheapo consumer DAC is gonna do. But the oversampled meters will get you into the ballpark; and they'll certainly make you aware of the issue! Like Raal said, it's quite an eye-opener to see what those plugins do to your levels.

I'm really glad this is turning out to be helpful to some of you. It's been quite the learning experience for me!

--Mark
Old 28th September 2005
  #54
Gear Guru
 
u b k's Avatar
 

okay, a few more things then.

first, yes, if you do what i suggested and trim any channels that are feeding comps, you have to lower the threshold on the comp to compensate. and after you do, the mix still will not sound the same.

second, my understanding is that many plugs, specifically eq's and comps, do a form of reconstruction on the word in order to apply their process to it. this is why it's important to avoid intersample peaks at the plug-in stage, even if the summed mix has no illegal values. am i getting that right?

if i'm not right about that, then i'm having trouble understanding how intersample peaks matter as long as the final output of the daw is low enough, because intersample peaks are only an issue for dac's.

thanks for your patience.


gregoire
del ubik
Old 28th September 2005
  #55
Gear Guru
 
u b k's Avatar
 

oh, and despite my best googling, it appears that there are currently no oversampling meters for we of the native ilk. tdm yes, audiounits/vst no.




gregoire
del ubik
Old 28th September 2005
  #56
Gear Maniac
 
kenkelly81's Avatar
 

Would Elemental Audio Systems - InspectorXL work?

http://www.elementalaudio.com/produc...rxl/index.html
EAS: InspectorXL Analysis Plug-In Suite
Old 29th September 2005
  #57
Lives for gear
Quote:
Originally Posted by u b i k
my understanding is that many plugs, specifically eq's and comps, do a form of reconstruction on the word in order to apply their process to it. this is why it's important to avoid intersample peaks at the plug-in stage, even if the summed mix has no illegal values. am i getting that right?
if you're not, i'm at a loss too.

Quote:
if i'm not right about that, then i'm having trouble understanding how intersample peaks matter as long as the final output of the daw is low enough, because intersample peaks are only an issue for dac's.

thanks for your patience.
and thanks for your description, which is how i understand it. so if this isn't correct, someone please chime in.

gwailoh?
Old 29th September 2005
  #58
Lives for gear
 

Quote:
Originally Posted by u b i k

second, my understanding is that many plugs, specifically eq's and comps, do a form of reconstruction on the word in order to apply their process to it. this is why it's important to avoid intersample peaks at the plug-in stage, even if the summed mix has no illegal values. am i getting that right?
Yes, many plug-ins do 'a form of reconstruction on the word in order to apply their process' to the signal. Usually plug-ins do this at 64 bits in floating point hosts. Some plug-ins are single precision 32 bit float. These plug-ins won't have any issues with intersample peaks.

BUT, there's a big but here, some plug-ins are fixed point (but designed to be used with floating point daws), and they won't tolerate any overs.. there's a couple of fixed point Waves plug-ins I know of (the linear phase eq and mb..) and everything for the TC Powercore platform.. I think you need to worry about intersample peaks with these plug-ins, if you use them.
Old 29th September 2005
  #59
Lives for gear
 

Quote:
Originally Posted by raal
in their white paper, digi states it's pretty much impossible to 'overload' the stereo bus.
That's because it has enough headroom, because the people that designed it know what they are doing. But still, it's easy to overload a lot of fixed point plug-ins (within the fixed point system), and that's when you need to be careful..
Old 29th September 2005
  #60
Lives for gear
thank you juicemaster.
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump