i thought it is best to record around -18dbfs

[snow lizard]

OK, now here's where *I* get confused; when one starts talking about bit depth as resolution.

<snip>

What am I missing there?

Not much, really.

What I've been trying to illustrate is based on what I've read that tries to say that printing higher levels in PCM audio result in finer dynamic resolution because of how many bits are operating in the selected 6 dB chunk range. Seems like it's been a common argument, but I'm not even sure that it's correct.


There's an article here that says the signal is mapped onto a set of evenly spaced discreet values. (What Tom said, basically) In short, I don't think there's a payoff in dynamics for printing hotter levels, even though that's always seemed like common advice.

If it were even true, I still don't think it's a good enough reason to abuse the gain stage going in. It just amounts to more of a misconception about how it actually works.


sl

 

[jrhager84]

so out of everybody's experience, peaking at ~-6dbfs is the most ideal?

 

[Farview]

so out of everybody's experience, peaking at ~-6dbfs is the most ideal?

NO! Having an average signal level around -18dbfs is ideal. (assuming that your converters are calibrated to -18dbfs = 0dbVU)

 

[jrhager84]

so out of everybody's experience, peaking at ~-6dbfs is the most ideal?

I said peaking, not average...

 

[SouthSIDE Glen]

What I've been trying to illustrate is based on what I've read that tries to say that printing higher levels in PCM audio result in finer dynamic resolution because of how many bits are operating in the selected 6 dB chunk range. Seems like it's been a common argument, but I'm not even sure that it's correct.

I think I have 99% of it now. It took me a whole morning of thinking it through (there's several hours of my life I'll never get back ), but I do get how the use of the bits can indeed be done without the dichotomy I thought I saw. Standard binary math, actually (as you demonstrated well in your last post. Thanks!) I was making a mental mistake as to how I was visualizing the 1bit = 6dB thang. I got it just like downtown now.

In short, I don't think there's a payoff in dynamics for printing hotter levels, even though that's always seemed like common advice.

I have to agree. The difference between 1/65,000th of a dB and 1/6millionth of a dB is meaningless to even the most golden ear. That would be like deciding whether to print a photograph with a resolution of one micron or one Angstrom; it doesn't matter because we'd need an electron microscope to see the difference anyway. Add to that all the other more practical reasons not to push it already listed in this thread, and resolution becomes a non-argument in my book.

Thanks a ton for the followup SL. You helped me slog though the mental gymnastics very nicely. I'd give you rep for it if I didn't already give you some yesterday for the same thread . Consider this an IOU until the BBS thinks I have spread it around enough to come back to you. Same for Tom, Jay, and everyone else who I have not already gotten and who has has helped in this admittedly very tough thread.

G.

 

[Farview]

I said peaking, not average...

Right! But it's all about the average levels. The peaks don't matter as long as you don't go over.

The optimal level will always be an average level, not a peak.

 

[SouthSIDE Glen]

I said peaking, not average...

JR,

Peaking somewhere near that number is not a bad rough ballpark figure for many situations, but it's really not a great way to look at the problem; it's really kind of asking the wrong question because asking about peaks ignores the nature of the audio itself.

If you're recording heavily sustanined and distorted electric guitar, there may be only a couple of dB difference between the average level of the signal and the peaks. In such a case, peaking at -6dBFS is probably pushing the signal a good 6-10dB on the hot side. On the other hand, if you're recording a signal with peaks that rise 18dB above the RMS level, peaking at -6dBRMS - while probably quite acceptable, might actually be a couple of dB low of the sweet spot.

Farview gave the right answer to the right question. Find out how your converter is calibrated, and send through the signal to average at that level, bringing it in lower than that if need be to avoid clipping the peaks.

G.

 

[jrhager84]

Why does this all have to be so complicated??? LOL

 

[SouthSIDE Glen]

Why does this all have to be so complicated??? LOL

That's why it's called "audio engineering" and not "audio playing in the sandbox"

Look, if while getting your feet sandy in this racket you want to use as a general rule of thumb the -6dBFS peak idea, it's not an awful place to start. It's certainly better than 0dBFS peak tracking.

But when you've been at it for a while and you get to the point where you feel your mixes can use improvement and you're not sure what steps to take next to refine your technique, keep this thread in mind .

G.

 

[flatfinger]

You would have to drop a 24-bit recording by 48 dB to reduce it to 16-bit resolution, so there's a lot of room---

That said........................USE EVERY DAM BIT!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

 

[jrhager84]

That's why it's called "audio engineering" and not "audio playing in the sandbox"

Look, if while getting your feet sandy in this racket you want to use as a general rule of thumb the -6dBFS peak idea, it's not an awful place to start. It's certainly better than 0dBFS peak tracking.

But when you've been at it for a while and you get to the point where you feel your mixes can use improvement and you're not sure what steps to take next to refine your technique, keep this thread in mind .

G.

Yeah, it was more of a statement of frustration more than anything...LOL

 

[snow lizard]

I think I have 99% of it now. It took me a whole morning of thinking it through (there's several hours of my life I'll never get back ), but I do get how the use of the bits can indeed be done without the dichotomy I thought I saw.

Me and my dichotomy...

The idea is counterintuitive and irrelevant at best and not true at worst. It just seems to be one of the only good arguments to push your levels too hard, and it still isn't good enough.

Thanks a ton for the followup SL. You helped me slog though the mental gymnastics very nicely. I'd give you rep for it if I didn't already give you some yesterday for the same thread . Consider this an IOU until the BBS thinks I have spread it around enough to come back to you. Same for Tom, Jay, and everyone else who I have not already gotten and who has has helped in this admittedly very tough thread.

G.

Well, it's ALL about the rep.

I've always liked this place for guys like yourself, Tom, John, Harvey, Ethan etc... that go out of their way to try to explain this stuff. Regardless of chicklets, that's where the real rep is.

Now in terms of straight 1 to 1 chicklets, I probably still owe you a couple hundred. Given my relative chicklet weight deficiency factor (the RCWDF), I don't think you need to worry about it for a while.


sl

 

[mrhotapples]

You would have to drop a 24-bit recording by 48 dB to reduce it to 16-bit resolution, so there's a lot of room---

That said........................USE EVERY DAM BIT!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

But human ears aren't sensitive enough to pick up on the dynamic range! The whole issue when it comes down to it is that people are using low-end equipment and pushing it past it's sweet spot causing all kinds of low level clipping, to get dynamics no human ear can hear and tracks that are overpowering in a mix that need to get turned down.

If noise being introduced through shitty equipment/environment weren't an issue, it'd still make no sense to track most kinds of music with higher RMS values because though the variables are there to utilize, they aren't there in the music.

A distorted guitar doesn't have much dynamic range. A heavy metal drum track doesn't either. It's all loud, louder, and loudest, not loudest, loud, soft, quiet, whisper yadda yadda! It doesn't matter that you can express millions of variations in volume when you're going to have no variation in volume.

I guess if you're recording a style of music where you have huge variations in volume going on...Say classical music, you know, where one section of a song can be 10db quieter than the next...That dynamic range you achieve tracking really hot (through clean pres and converters) might make more difference.

All this stuff aside, in 24bit it logically, to me, it makes sense to track an instrument as close as possible to where you need it to be in the mix so you have as much headroom as possible for making it louder if need be and using effects.

I'd do it completely different in 16bit though.

 

[SouthSIDE Glen]

I've always liked this place for guys like yourself, Tom, John, Harvey, Ethan etc... that go out of their way to try to explain this stuff.

I can't speak for the other old fogeys, but I'll let you in on a little personal secret...

...I sometimes think half of the reason I do it is to keep this stuff fresh in my mind, keep the gray cells exercised, and fight off any encroachment of Alzheimer's disease. I just do it in front of you guys because if I talked to myself, I would ALREADY be senile.



G.

 

[Tim Gillett]

JR,

Peaking somewhere near that number is not a bad rough ballpark figure for many situations, but it's really not a great way to look at the problem; it's really kind of asking the wrong question because asking about peaks ignores the nature of the audio itself.

If you're recording heavily sustanined and distorted electric guitar, there may be only a couple of dB difference between the average level of the signal and the peaks. In such a case, peaking at -6dBFS is probably pushing the signal a good 6-10dB on the hot side. On the other hand, if you're recording a signal with peaks that rise 18dB above the RMS level, peaking at -6dBRMS - while probably quite acceptable, might actually be a couple of dB low of the sweet spot.

Farview gave the right answer to the right question. Find out how your converter is calibrated, and send through the signal to average at that level, bringing it in lower than that if need be to avoid clipping the peaks.

G.

Glen I can understand why this sort of talk can sound very confusing to many. For clarity I'm trying to break it down to its essence.

There are basically two types of recording meters, fast reacting peak reading and averaging. For tracking, and assuming the gear is calibrated and gain staged, if the user keeps an eye on both but especially doesnt overcook the fast peak reading meter level, why isnt that good enough?
Tim

 

[Massive Master]

Because if you're recording something with little transient material (a synth pad for example), keeping an eye on the peak meters will blow the snot out of your front end. You'll be pushing the input maybe 15dB (or more) over what it was designed to handle.

And then to mix it, you're going to have to turn that (now distorted) signal *down* maybe 18 or 20dB to start your mix. So not only are you gaining nothing, you're losing a lot.

The audio industry got by just fine for decades without peak meters... IME, if I need to even *look* at the peak meters, I'm WAY too hot already.

 

[Tim Gillett]

The audio industry got by just fine for decades without peak meters... IME, if I need to even *look* at the peak meters, I'm WAY too hot already.

Well then dont dare "look" at the peak meters because you "might" be way too hot. What a laugh.

Tim

 

[Farview]

Well then dont dare "look" at the peak meters because you "might" be way too hot. What a laugh.

Tim

I think you misunderstood what he said.

If you use proper gain staging, you don't need peak meters.

 

[Tim Gillett]

Funny, I always thought adding a certain number of db's lead to a standard VU meter's sensitivity wasnt quite the same thing as gain staging but there I go. I learn something every day...

Tim

 

[Tim Gillett]

I've just read an article by Bob Katz which underlines what some have said here about some studio gear ( solid state amps and op amps amps) unable to deliver a totally clean sine wave with its peak at 0dbFS.

I had assumed that most solid state preamps, not being very taxing on their power supply , would not normally exhibit significant distortion until rail limits were reached. But it seems I was in error on that.
So to those who mentioned this, I apologize if I doubted your expertise here.

Still, I think that to generalize about this and make a general prohibition on the top say 6db of digital recording room simply because it may be a problem in a professional environment due to amplifier equipment limitations is I think unwise. Many users on this forum dont use pro levels at all. (I mostly dont in my home setup). The input to their converter may well be around prosumer levels. Quite possibly the pre's they are using dont struggle to drive sinewaves right up to 0dbFS (peak) cleanly. But sure, they may not either. It just depends.

For those who have an interface or SIAB where gain staging is largely if not entirely internal and fixed, it shouldnt be an issue as the manufacturer should have sorted that out. But I agree they may not have either.

Anyway, hope that clarifies things re the amp distortion issue.

Tim