snow lizard
Dedicated Slacker
I didn't think dB's were linear to begin with so I would guess that it's pretty bad.
I'm just guessing.
sl
I didn't think dB's were linear to begin with so I would guess that it's pretty bad.
I'm just guessing.
sl
snow lizard
Dedicated Slacker
Wow...
I wonder if the same is true if you're going in the other direction. i.e. turning the signal down. It seems typical that if you're recording a lot of tracks to be mixed together, they're going to sum so you're going to have to end up turning them down anyway. Regardless of printing at 0VU or whatever, it makes me wonder what other kinds of errors you're going to get later in the process. From your tests, it sounds like it's the converters doing the mangling but I wonder if there also might be something going on in the DAW.
sl
SouthSIDE Glen
independentrecording.net
Pipeline, thanks a bunch for the excellent and detailed reply. Very informative - not to mention very timely, as (to some people's dismay) I am just finishing up on a web app re introductory metering and gain sturcture, and I'll add a bit of a blurb nodding to this information.
I wanted to highlight the above quote as being a really excellent point that underlines the whole discussion, IMHO.
Thanks again...oh, and congratulations on having a new TV series named after your software
.
True. What I was referring to was the break in the normally linear (though offset) 1:1 relationship between dBu and dBFS (i.e. adding one dBu means adding one dBFS), as well as what I assume (wrongly?) to be a probably non-linear response in the converter circuitry that causes the effect that pipeline described (the "Pipeline Effect"?snow lizard said:
)Oh, and P.S.: I love when I see Nika Aldrich referenced in this board. It's both gratifying and weird to see it. Why? Because he happened to be my sales rep at Sweetwater back in the mid-late 90s before he broke out on his own. Most of the outboard gear I have now I actually bought through him. He was one of the most helpful sales reps I have ever dealt with in any field I have dealt with, he bent over backwards to help me out, and it's really gratifying to see his success above and beyond his old days of hawking gear on the phone.
G.
Last edited:
snow lizard
Dedicated Slacker
"The Pipeline Effect"...
An idea like that is probably good for an album, a tour...
Maybe a few groupies...
Getting back to Gain Staging 101 briefly here, I'm surprised the article was talking in peak levels rather than RMS. If a preamp is equipped with a VU meter, it's going to have slow ballistics, so you're not going to pick up most of the transients. Digital meters are fast. Targeting -6 peak dBFS as a generalization could be dangerous given the differences in transient content of, say, a clean acoustic guitar vs. a distorted electric guitar vs. a drum kit vs. a sine wave, et cetera. I could see turning the preamp down a couple of notches to keep peak levels in check, but have a hard time going the other way.
sl
An idea like that is probably good for an album, a tour...
Maybe a few groupies...
Getting back to Gain Staging 101 briefly here, I'm surprised the article was talking in peak levels rather than RMS. If a preamp is equipped with a VU meter, it's going to have slow ballistics, so you're not going to pick up most of the transients. Digital meters are fast. Targeting -6 peak dBFS as a generalization could be dangerous given the differences in transient content of, say, a clean acoustic guitar vs. a distorted electric guitar vs. a drum kit vs. a sine wave, et cetera. I could see turning the preamp down a couple of notches to keep peak levels in check, but have a hard time going the other way.
sl
T
Tim Gillett
Banned
Glen,You still are implying that, even with a safety factor, boosting digital gain to get the most bit usage without clipping is the way to go. We have been trying to explain why that is not the case.
Maybe once your rep strength moves off of 666, you'll see the differences in the devil in the details
G.
So I'm not just implying it now. I'm still implying it, am I? Then why did I say in my initial post (the part you edited out of your quote) "theoretically anyway, for minimum converter noise"?
Why did you edit out my caveat which limits what I say and qualifies it, and then criticize me for allegedly not having said it?
I was speaking where converter noise is the limiting factor. Not the analog preamp noise or indeed anything which appears earlier in the signal chain. What do we know about this guy's equipment and which part of it is the weak link, if any?
OF COURSE boosting gain only to add system noise to the track will limit the potential dynamic range of the recording.
Interesting that you imply I said to turn up converter gain. I said nothing about where in the chain to increase gain. It could have been by getting the talent to come closer to the mic! (Interestingly that's how Les Paul recounts how he cut preamp noise in some of his early recordings)
But even if gain is increased inappropriately so as to reduce overall headroom (and I regard it as inappropriate, like everybody else who understands the issue) it's only a reduction in POTENTIAL. If the recording did not clip, nothing was lost. If converter noise does not intrude above noise in the preceding stages, AND the track does not clip, then that's all we can ask for, in terms of having got that track in the can, optimally, using that particular gear.
Once again, headroom is only a potential thing. Unused headroom for a given track does not result in a cleaner recording, in any way, shape or form. It's great to have it up your sleeve, and who wouldnt want it, especially in the unpredictable situation of live location recording, but that's all it is. That's why I described it in terms of a safety margin, rather than actual improved fidelity on any given track.
Thinking about what I said, yes I could have said it better. Better would be, "get your track above system noise, but dont clip". But the practical result will be exactly the same as for a person who tracked as high as they could and yet didnt clip. In audio fidelity terms, not a shred of difference. Or to reuse your farmyard metaphor, the two tracks will walk, quack and smell the same. Any subsequent need to digitally change gain in no way affects the basic track's fidelity, even if it's one more little chore to perform.
"track high as you can but dont clip" could be read in three different ways.
1. just as it's written
2. TRACK HIGH AS YOU CAN but dont clip.
3. track high as you can BUT DONT CLIP"
I meant it in the totally neutral first sense, not giving more weight to one or the other. So long as both criteria are satisfied it's good advice.
To quote from a Protools reference guide:
7. Adjust the output level of your sound source (instrument, mixer or preamp)
Monitor the track's meter levels in Protools to ensure that you get the highest possible signal without clipping. My italics.
Unless you want to start a tangential discussion about possible amplitude or other distortion in the last one db before digital clipping occurs in the DAW. But what would anybody in their right mind be doing nudging anywhere near that close to clipping in a live tracking session anyway? Any distortion a pre has just before it clips will probably be nothing compared to the horrible crunch when it actually clips.
The unpredictability of live voice or instrument recording means just one unexpected note can suddenly go WAY over the previously highest level. Again, "dont clip" means "dont clip". In practice it is wise to use quite a safety margin, even given a rehearsal. But then I think that was covered by "dont clip".
But in any case, the normal meaning of headroom is headroom that is clean, undistorted and totally useable. That's the meaning I use when I use the term. Do you want to give it a different meaning? You are free I suppose to define headroom however you like.
Ideally one should send a test tone through the gear from start to finish to check that the headroom is really headroom right up to the point of digital clipping, or at least so that the metering you use for tracking is telling you the truth, or perhaps a little conservative.
Again, I'm happy with the Protools manual advice. "highest possible signal without clipping" If you have a problem with that, so be it.
As for your parting rep points comment, really, what can one say, Glen?
Tim
pipelineaudio
Well-known member
There's a pretty important reason that PT would say that though, which is an issue most here don't have to deal with.
AFAIK on the TDM PT systems, not sure if this was changed in HD, fx plugins, bus sends, and some other routings HAD TO be reduced to 24 bits, which IIRC was done by dither...AFAIK the internal bit depth was 48bit fixed in the app, which meant a 144dB drop in dynamic range when adding a plugin. This means constantly dealing with being as close to dither noise as your ADC, and worse adding multiple instances of dither every time it needs to go back to 24 bits.
Dont want to dither? then deal with truncation, or quantization noise.
So you absolutely must in that app, keep your signal levels high at all times.
(I could be very very wrong about the above and would appreciate it if someone corrected me)
Most native apps simply dont have this problem, operating at 32bit float or 64 bit float, and staying there, thru busses, routing and plugins, until the final output where you would dither ONCE, therefore for most of us, theres no reason to be pushing your analog side to its limits.
M
mrhotapples
New member
I record everything at -24dbfs unless it's pudding and then I don't pay attention and use all the gain I have. Pudding's quiet and it sounds good when it clips in digital.
S
struberg
charhizma mb
A loud mix only depends on how you pump up the volume in the mixdown and now how hot it is recoreded (unless you record to tape which has a natural kind of compression when it gets into saturation).
Imagine the following:
-6db means half signal voltage
-12db means 1/4 signal voltage
-18db means 1/8 signal voltage
On the digital side one may argue that this makes not much difference for a 24 bit system, but this is not entirely true. The least significant bit (LSB) of an 16bit system is not equal to the LSB of an 24 bit system. Its more that the additional 8 bits are 'the commas behind the dot' and not an additional headroom. So the LSB (or most likely the 3 least significant bits) more or less mostly carry noise and 'aural information'.
You also have to look at the analogue part of the recording before it gets A/D converted:
1/8 voltage therefore increases the noise to 8x if you only look at the analogue part.
snow lizard
Dedicated Slacker
Imagine the following:
0 dB VU is 1 volt at 1 kHz.
-18 dBFS is exactly the same level.
-12 dBFS is 2 times that level.
Take it one step further and imagine that you're producing a square wave recording because of the "Loudness War" stuff that keeps pushing RMS levels of CDs up, and the recording will have a final overall dynamic range of around 6 dB. Your LSB will be well over 60 dB lower than that, at least, and many half decent stereo systems would be in good shape to have more than 60 dB range. Don't even think about boom boxes, iPods and computer speakers. The LSB's of 16 bit systems are beyond that range. The LSB's of 24 bit systems are beyond studio monitoring systems and audiophile gear with way more range.
sl
0 dB VU is 1 volt at 1 kHz.
-18 dBFS is exactly the same level.
-12 dBFS is 2 times that level.
Take it one step further and imagine that you're producing a square wave recording because of the "Loudness War" stuff that keeps pushing RMS levels of CDs up, and the recording will have a final overall dynamic range of around 6 dB. Your LSB will be well over 60 dB lower than that, at least, and many half decent stereo systems would be in good shape to have more than 60 dB range. Don't even think about boom boxes, iPods and computer speakers. The LSB's of 16 bit systems are beyond that range. The LSB's of 24 bit systems are beyond studio monitoring systems and audiophile gear with way more range.
sl
Last edited:
SouthSIDE Glen
independentrecording.net
Tim,
We apparently are talking entirely different languages now, since I had already answered that question in the last post. This is really getting tiring.
What you still apparently don't understand, Tim, is that there is an A in ADC. Every converter includes an analog input stage. It's the analog level at that gain point (either at the ADC input gain or, if one is not available, at the output gain of the preceeding device) that we're referring to. Just because one uses the term "preamp" doesn't automatically mean a seperate microphone preamp.
This is the problem in our lack of communication on this topic, Tim; you're in a little over your head and you're just not understanding what's being said. That's fine, you are far from alone there. But geez, that's no reason to get hot under the collar.
Even that is a misunderstanding of what I was trying to explain. I think I should give up, because it's just not getting through the translation barrier, you just don't understand.
Now you're just contradicting yourself even more, Tim. First you're getting on our cases for bringing up anything upstream of the D part of the ADC, because the OP and the thread were specifically referring to the digital levels only, and now you're getting on my case for "assuming" that you're talking about digital gain levels.
Oh and BTW, yet another point that was apparently lost in translation: I wasn't even assuming that you were boosting converter input gain; it sounded mre to me like you were talking levels downstream of the converter more than anything else.
One last time, I was referring to boosting the digital gain, which serves only to raise the converted noise floor. No it noes not add noise, but it does make the existing noise louder.
Better would be to remove the word "clip" altogether, as it referrs to the digital side of the equation, and the idea of boosting the post-converter digital gain is what's bogus. Best would be to remove the "hot as you can" reference altogether and instead state that one should simply follow good gain structure procedure. "As hot as you can" and "as hot as you should" are two different things.
No it's not.
The entire professional audio engineering community has a problem with it. C'mon, Tim, you've been around this forum long enough to have seen the conversation several times over. The PT manual isn't the only one to say that; there are a *lot* of manuals out there for various hardware and software that says much the same thing.
And every one of them is giving the kind of advice that only makes sense from the perspective of that particular piece of 'ware existing in isolation. But when considered as part of a real signal chain in the real world, it is simply not appropriate advice. Go look up what any name brand engineer who has ever spoken on the subject has to say about it. Go ahead; we'll wait. You won't be able to come up with a single respectable one that says what those manuals say on the topic is actually good real life advice.
One could say a few things, Tim. First, that it was a joke that you are apparently still wound too tight to get. Loosen up, bud
. Second, that you did an amazing job of artifically doubling your total points in a matter of a couple of hours by trading meaningless rep points like Chinese carbon credits down in the cave.At any rate, Tim, I'm through with you on this topic. We are only taking past each other at this point, which serves only to burn bandwidth. I give up.
Exactly my position, too, sl. Looking at it from the peak perspective is looking at it ass-backwards, IMHO. of course one does not want to clip their peaks, but to select a specific level for peak values totally ignores the actual crest factor of the program material.snow lizard said:
G.
X
xstatic
New member
This is almost getting funny. As far as I am concerned, "track high as you can but dont clip" is NOT good advice but is actually BAD advice. If you want to be technical here Tim, there IS a difference between distortion and actualy clipping. Your advice to track as high as possible without clipping actually invites and promotes overdriving the front end which is inviting and promoting tracks which are less than optimal.
"Unused headroom for a given track does not result in a cleaner recording, in any way, shape or form" is also a false statement. It is a tricky one, but false none the less. This statement is true if you consider ONLY the converter and have this statement ONLY refer to the converter. If however you are looking at the big picture, like everyone but you seems to be, than once again, approaching the converters limit means levels far above optimal on the preamp. Levels far above optimal on the preamp are contaminated with all sorts of distortions and irregularities. By looking at the BIG picture, the whole signal chain, the opposite of your statement is true. On the other side of the coin though, some of the characteristics of a preamp operating past its unity point may be sonically desirable, but I have NEVER heard a preamp which gets cleaner as it approaches the end of its range. I have also never heard one that stays the same.
"Unused headroom for a given track does not result in a cleaner recording, in any way, shape or form" is also a false statement. It is a tricky one, but false none the less. This statement is true if you consider ONLY the converter and have this statement ONLY refer to the converter. If however you are looking at the big picture, like everyone but you seems to be, than once again, approaching the converters limit means levels far above optimal on the preamp. Levels far above optimal on the preamp are contaminated with all sorts of distortions and irregularities. By looking at the BIG picture, the whole signal chain, the opposite of your statement is true. On the other side of the coin though, some of the characteristics of a preamp operating past its unity point may be sonically desirable, but I have NEVER heard a preamp which gets cleaner as it approaches the end of its range. I have also never heard one that stays the same.
M
mrhotapples
New member
I just know that with my Firepod, when I track snare drums and let them excede an average -12db in, they start to slightly distort. I can push guitars to -5db average. Bass starts to distort hardcore if I approach -10db. I've run 1khz signals at varying volumes from different sources into my Firepod and established that -20 to -24db is my range of error. I try and shoot for that range when recording anything. My recordings don't end up quiet or hissy.
To each his own in the end. I'd never use an SM58 to record a person singing or screaming, infact, I hate them. I know plenty of people who think otherwise. I prefer cheap mics to some expensive ones. It's preference in technique. If you achieve good results tracking hot as fuck on lowend stuff, great.
To each his own in the end. I'd never use an SM58 to record a person singing or screaming, infact, I hate them. I know plenty of people who think otherwise. I prefer cheap mics to some expensive ones. It's preference in technique. If you achieve good results tracking hot as fuck on lowend stuff, great.
NL5
Unpossible!
SouthSIDE Glen
independentrecording.net
Let's turn things around for a minute, maybe it'll open up some fresh perspectives.
For those who advocate the "digitally record as hot as you can without/before clipping" side (not just Tim, though you're welcome to respond if you wish, but anybody advocating that position), here's your chance to defend and explain it using a real-live type of situation that's probably encountered a thousand times a day in real life by members and guests of this board:
Lets say that you are recording a high-distortion, high-sustain, heavy-on-the-power-chords, wall-of-electric-guitar track to a computer hard disc for later mixing. You already have the tone, the mic placement, the mic pre, and any EQ or compression (if needed, or none if not needed) dialed in and sounding just how you want it to sound. You don't really want to alter the sound any more or add any more noise than necessary, you just need to get it into the computer and onto disc at what you consider to be the right tracking level. Because of the heavy sustain and distortion, the signal has a crest factor (the range from RMS to peak level) of about 6 decibels.
To get there you need to run into one of the the line in ports of a typical 8-channel-analog-to-FireWire interface, through the device driver for that interface on the computer and to the multitrack recorder/editor of your choice, which records it as a mono track (no plugs) to the hard drive.
Given that extremely common situation, please describe your gain structure setup at these four gain stages, why you set up the gain structure that way, and why such a structure at each stage yields the best theoretical results over any other methods discussed here:
1. The line going into the interface (i.e. the output from the previous upstream device.
2. The input gain (if adjustable) on the interface. If not adjustable, how that may or may not affect what you do in #1.
3. The digital output or driver level control of the digital signal coming out of the interface and into the computer.
4. The track recording level fader in the multitrack software.
This is your chance to stop defending individual points and posts, and actually explain how and why the concept of "digitally record as hot as you can without clipping" actually works best in a real-world setup.
G.
For those who advocate the "digitally record as hot as you can without/before clipping" side (not just Tim, though you're welcome to respond if you wish, but anybody advocating that position), here's your chance to defend and explain it using a real-live type of situation that's probably encountered a thousand times a day in real life by members and guests of this board:
Lets say that you are recording a high-distortion, high-sustain, heavy-on-the-power-chords, wall-of-electric-guitar track to a computer hard disc for later mixing. You already have the tone, the mic placement, the mic pre, and any EQ or compression (if needed, or none if not needed) dialed in and sounding just how you want it to sound. You don't really want to alter the sound any more or add any more noise than necessary, you just need to get it into the computer and onto disc at what you consider to be the right tracking level. Because of the heavy sustain and distortion, the signal has a crest factor (the range from RMS to peak level) of about 6 decibels.
To get there you need to run into one of the the line in ports of a typical 8-channel-analog-to-FireWire interface, through the device driver for that interface on the computer and to the multitrack recorder/editor of your choice, which records it as a mono track (no plugs) to the hard drive.
Given that extremely common situation, please describe your gain structure setup at these four gain stages, why you set up the gain structure that way, and why such a structure at each stage yields the best theoretical results over any other methods discussed here:
1. The line going into the interface (i.e. the output from the previous upstream device.
2. The input gain (if adjustable) on the interface. If not adjustable, how that may or may not affect what you do in #1.
3. The digital output or driver level control of the digital signal coming out of the interface and into the computer.
4. The track recording level fader in the multitrack software.
This is your chance to stop defending individual points and posts, and actually explain how and why the concept of "digitally record as hot as you can without clipping" actually works best in a real-world setup.
G.
Massive Master
www.massivemastering.com
Reading through, it seems that he either (A) doesn't understand or (B) doesn't want to understand the whole concept.
masteringhouse
www.masteringhouse.com
Isn't it odd though that while we MEs preach about not recording hot, keeping peaks below 0 dBFS, and not clipping, we make CDs that are hot, peak as close to 0dBFS as possible, and clip converters for certain types of music?
snow lizard
Dedicated Slacker
Absolutely!
sl
sl
SouthSIDE Glen
independentrecording.net
I think it's 70% (A) and 30% (B), and that it's just not him, but a whole generation of home recordists.
The irony that I find in this whole topic is that it's a fairly regularly heard complaint on this board and those like it that the "pros" keep the important "secrets" to themselves. One of the most important not so secret "secrets" that most of the more expereinced folks here constantly drill is the importance of getting the tracking right. A major part of getting the tracking right is understanding the importance of good gain structure in the tracking chain to get the best performance out of whatever gear one uses. Yet when the topic *does* come up and is discussed, those that complain about not being let in on the "secrets" don't get when one is being handed to them for free on a silver modem.
The good news is that the Volume Wars are not going to last forever; they are simply the polyester leisure suit and big hair of audio production fashon. The bad news is I have no idea of what production version of narrow ties and hair mousse is going to replace them next.masteringhouse said:
G.
T
Tim Gillett
Banned
Let me get this straight. So for these home recordists, even though they might be using a PT TDM , or any one of "a lot" of others for which "ensure that you get the highest level without clipping" is (in your words) perfectly good advice, it's actually wrong advice...
Why is it wrong advice? Because it would be wrong advice if they were using some sort of other equipment, like the equipment you are using, or other "pro's" are using.
Wow, Glen. Just wow.
Tim
I
Ironklad Audio
New member
1. set the gain on the preamp until it's 3-4db below clipping on the highest peaks. maybe a couple db hotter if i want to add more of the preamp's particular sonic characteristics.
2. let the converters do the rest. if i'm using an outboard pre, i set the gain on my board to the line-in "U" mark.
3. watch the input meters on my screen hover around -16, jumping to around -10 on the peaks - not once ever touching a digital fader.
i really don't see why gain staging seems to be such a difficult concept...or why people go around and around in circles arguing/discussing it