Old 09-01-2014, 12:16 AM   #1
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 372
Default The -18 dBFS rule -- myth or reality?

(this is a continuation thread from this thread starting at post #34 -- didn't want to be OT any more over there. Please read that thread for background before posting here.)

Quote:
Originally Posted by karbomusic
I had to raise the preamp ~10dB higher than necessary to just so I could get the raw guitar signal to -9 RMS. Why do I need to do any of that when recording @24 bit more than allows the gear sending the signal into the DAW to run within its prescribed range? And when some VSTs expect it; not for quality reasons but because of my statement below concerning incoming signals and where they naturally fall?
As i said in my other posts, i see the merit of the "levels convenience" issue: for some (many? most?) the way the levels work out through their signal chain, into and out of the digital world, etc, works out better for them at -18. I record into the computer and mix ITB, and for me recording hot is easier to see on the meters, works more conveniently with the plugins i use (primarily iZotope Alloy 2 these days, but others as well), and so forth. Signals peaking at -18 dBFS RMS are slightly harder to handle everywhere in my workflow... I end up boosting them immediately anyway to make them more useful. Others' mileage varies, clearly, to each their own, etc etc.

In terms of gain-staging, I would ideally send a hot signal into my interface to get as close as is comfortable to 0 dBFS without risk of clipping, if possible to do so without amplifying anything. In reality i don't even worry that much about making it hot, since the added noise I get coming in at -18 RMS is inaudible under normal conditions, but I certainly wouldn't make a point to record at -18.

In terms of the gain on the pre-amp, I see your point: in theory, no reason to boost the signal at the pre-amp (and add more noise) just to get closer to 0 dBFS. But for whatever reason of design, the two interfaces I've tested (both PreSonus) seem to add a non-variable amount of noise somewhere in the internal chain, the end result being that if I boost the preamp knob to send a hotter signal to the internal ADC, I end up with a signal with less noise than sending at a level that results in -18. Not a big deal at all, but that's what happens. Maybe my interface sucks, no idea.

But the noise issue is beside the point, which is that there is a notion floating around that for most interfaces recording -18 dBFS RMS peak will make the recorded material itself sound better, and not just because of convenient levels or because it's a handy rule of thumb to avoid digital clipping, and that specific assertion is what I'm skeptical about. Your post in the other thread referred to "those poor converters you are trashing on the way in" and said "-18 db is +18 outside the box and near the max most converters/pres can reliably handle"... maybe I misunderstood and you were only referring to clipping, but it sounded like you expect perceptible differences in audio quality. And even if you don't, there are plenty out there who do and who post all time about how recording at -18 made their audio sound so much better, more punchy, more clear, changed their lives completely, cured their cat's illness, etc...

If the -18-for-sound-quality thing is for real, then I accept that my three tests on two interfaces are certainly inadequate to disprove it, and that my very limited experience with other interfaces hampers my perspective; but to me the onus is on anyone who advocates for -18 as a sound quality issue to demonstrate it with some audio files. I've searched the internet high and low and find no examples or people willing to provide them. Some people scoff at ABX testing and disregard math/statistics. All of that makes me a little more suspicious about it being an urban legend. :-)

I'm very willing to be proven wrong.

(Incidentally, my previous test in the other thread included a metering error on my part which meant that the "-18 RMS peak" files were peaking at more like -12 or -14; i've reposted the corrected audio files; no change in the results that i can hear.)
clepsydrae is offline   Reply With Quote
Old 09-01-2014, 12:29 AM   #2
reapercurious
Human being with feelings
 
reapercurious's Avatar
 
Join Date: Jul 2007
Posts: 1,149
Default

two words that changed my mixing and shut down a lot of my inner chatter about the zillions of contradictory things i have read

Master. Fader.

and to be fair, and sort of address what I think you're saying. if you want to mix quiet, then crank up your monitors till your ears almost bleed, then its easy to leave headroom.
reapercurious is offline   Reply With Quote
Old 09-01-2014, 02:38 AM   #3
brainwreck
Human being with feelings
 
brainwreck's Avatar
 
Join Date: Jul 2006
Posts: 12,070
Default

clepsydrae, in general, I agree with what you are saying here, and I certainly agree with your want to hear something concrete, rather than taking someone else's word for it. For myself (after doing my own experimentation), recording levels with an interface are much more of a practical concern than a quality concern. Take di guitar, for example, I have concluded, through experience and testing, that as long as I give myself a good safety margin for avoiding clipping, and that I'm not having to fiddle around the input/output levels on the amp sim too much, the quality of the incoming signal level will be fine.

For the example which I posted in the other thread, the -9dbfs RMS sound clip was too hot, in that I had to avoid playing too hard so as not to clip. But I wanted to prove a point: that tracking at a higher RMS level than the 'rule' will not absolutely destroy the incoming signal. Does it sound a little different? Yes. Might one person prefer a lower level based on sound quality alone? Yes. Was the sound destroyed? No. In practice, though, I never track that hot (for practical reasons), so the difference in sound between -18dbfs RMS and -9dbfs RMS doesn't come into play.
__________________
You say you want a revolution, well you know, we all want our products.
brainwreck is offline   Reply With Quote
Old 09-01-2014, 04:57 AM   #4
kindafishy
Human being with feelings
 
kindafishy's Avatar
 
Join Date: Mar 2009
Posts: 3,468
Default

Never heard of this rule. Seems a bit nonsensical to me.

When recording, I play hard into my interface, get it just to the point of clipping and then back it off a bit.

As far as the level that is captured in REAPER, I don't see how it matters. The digital level is relative so the s/n ration is the same whether you are recording at -12dB, -18dB or -30dB RMS.

I like to record with my peaks around -6dB because it is practical from a listening standpoint.

When mixing, I do normalize to -18dB RMS, but this is because of plugin input gain staging.

Does not apply to recording though.

This is all my take on it anyway. I wanted to join the conversation as I am interested in seeing where it goes. I will be happy to be corrected on anything.

Phew, glad I've never heard this rule before because I am quite impressionable.




...doh!
kindafishy is online now   Reply With Quote
Old 09-01-2014, 05:15 AM   #5
Giano
Human being with feelings
 
Join Date: Mar 2014
Posts: 427
Default Kenny Gioa

I remember Kenny Gioia saying in one of his tutorials, when he was recording/tracking vocals that the average Input Level should be be between -18 and -12 db and peaks should not exceed - 7 dB.
He refered to the LED of track level meter in Reaper (w/o fx!).

Last edited by Giano; 09-01-2014 at 07:52 AM.
Giano is offline   Reply With Quote
Old 09-01-2014, 06:07 AM   #6
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 10,244
Default

Quote:
The -18 dBFS rule -- myth or reality?
^Neither, it's a rule of thumb and it's RMS not peak, this confuses things unnecessarily by using the wrong terms.

Shame on you for leaving out the "of thumb" part which I have painstakenly defined and used multiple times now. It's the appx level your signal would land in the DAW if you were at unity externally, that's what makes it sensical. However, this is the only post I'll make because it is a....

RULE OF THUMB

Quote:
A rule of thumb is a principle with broad application that is not intended to be strictly accurate or reliable for every situation. It is an easily learned and easily applied procedure for approximately calculating or recalling some value, or for making some determination.
You also omitted RMS again which I also bolded multiple times in the other thread for very good reason, it is NOT accurate when omitted. And some of your assumptions are projections. I'll own the embellishment of "trashing the converters" but the rest is just plain common sense.

So please, while you guys hash this out without me, would you please do me the simple favor of actually thinking instead of seeing the word "rule" then allowing your head to explode before understanding what comes after? Pretty please? With sugar on top?

Quote:
When recording, I play hard into my interface, get it just to the point of clipping and then back it off a bit.
Guess what general range that ends up landing in as RMS give or take a couple DB?

Most sound cards (as in rule of thumb) fall into the -18 RMS range when striking the best compromise between 0 dbFS and the analog side of their hardware hence the reason someone can follow this and be fairly comfortable they are within the specs of their gear:

Quote:
Analog levels - http://en.wikipedia.org/wiki/DBFS

dBFS is not to be used for analog levels, according to AES-6id-2006. There is no single standard for converting between digital and analog levels, mostly due to the differing capabilities of different equipment. The amount of oversampling also affects the conversion with values that are too low having significant error. The conversion level is chosen as the best compromise for the typical headroom and signal-to-noise levels of the equipment in question.

Examples:

EBU R68 is used in most European countries, specifying +18 dBu at 0 dBFS
In Europe, the EBU recommend that -18 dBFS equates to the Alignment Level
European & UK calibration for Post & Film is −18 dBFS = 0 VU
UK broadcasters, Alignment Level is taken as 0 dBu (PPM4 or -4VU)
US installations use +24 dBu for 0 dBFS
American and Australian Post: −20 dBFS = 0 VU = +4 dBu
The American SMPTE standard defines -20 dBFS as the Alignment Level
In Japan, France and some other countries, converters may be calibrated for +22 dBu at 0 dBFS.
BBC spec: −18 dBFS = PPM "4" = 0 dBu
German ARD & studio PPM +6 dBu = −10 (−9) dBFS. +16 (+15)dBu = 0 dBFS. No VU.
Belgium VRT: 0dB (VRT Ref.) = +6dBu ; -9dBFS = 0dB (VRT Ref.) ; 0dBFS = +15dBu.
__________________
Massaging electrons sure is a lot of fun.

Last edited by karbomusic; 09-01-2014 at 09:16 AM.
karbomusic is online now   Reply With Quote
Old 09-01-2014, 08:47 AM   #7
ashcat_lt
Human being with feelings
 
Join Date: Dec 2012
Posts: 425
Default

The real point of the whole thing is to run your analog gear at sensible levels to avoid adding analog distortion on the way into the box. This usually means keeping your average levels around 0dbVU = +4dbu.

The first part of this that gets fun is that so many folks nowadays don't actually have an analog chain with meaningful meters. You've got an all in one interface which might have a green light to tell you you've "got signal" (whatever that means) and then a red one to tell you you're clipping. Many don't even give you that courtesy. Most folks are not looking at a nice big analog needle, or even LED tree that tells them where the nominal level is, so they are pretty much forced to gauge it by the meters in Reaper, and that leads to the second fun part...

Every interface is calibrated differently. If you actually did set up say an analog mixer feeding your line level inputs, and set your levels on the mixer to 0dbVU, you could see in Reaper anything from -24 to maybe even -12dbfs. If you've got one calibrated at -24, and you push your average levels up to -18dbfs with peaks hitting around -6dbfs, then your mixer would be averaging at +6dbVU with peaks at +18dbVU, and that could be pushing the limits of many analog stages. How do you know how yours is calibrated? Either test it, or look at the spec sheet.

But that's the third fun part: Every spec sheet is different. Some call it headroom, some just list a maximum input level. Sometimes it's listed in straight db relative to the nominal level, sometimes it's listed as dbu, or dbV, and that's not always even the same scale that they used to specify nominal level. Some even just list it as a voltage, so that you have to go find a volts>db calculator.

Does it sound better? That's kind of subjective isn't it? It completely depends on the behavior of the analog chain. Some analog stages might sound kinda good when pushed up into distortion, some just sound like crap, and some folks might like the overdrive from this one or the other one, and some folks might thing it sounds better if it comes in as clean as possible. I'm not talking about "clipping" here. As mentioned above, it's quite possible to distort your analog chain without ever coming close to 0dbfs and incurring "digital clipping", and a lot of analog gear will start to get non-linear well before it even hits its own maximum output level.

The only right answer to this is to learn your gear. Learn to read the spec sheets, interpret and interpolate all the different ways that these things are presented. More importantly, test your gear, run various signals through at various levels and listen to the results, because in the end that's all that really matters.
ashcat_lt is offline   Reply With Quote
Old 09-01-2014, 09:07 AM   #8
beingmf
Human being with feelings
 
beingmf's Avatar
 
Join Date: Jul 2007
Posts: 2,521
Default

ashcat_lt, that's 99% what I wanted to write as well.

Of course the OP isn't sure about his monitoring situation - that makes the whole process a little difficult. But once you've got your room acoustics and monitors right, it's a pleasure to decide where to hit the AD in order to get the "best" signal for the situation.

[on a side note: tape was so much more fun when it comes to finding the "sweet spot", oh my... ]
beingmf is offline   Reply With Quote
Old 09-01-2014, 09:15 AM   #9
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 10,244
Default

Quote:
ashcat_lt, that's 99% what I wanted to write as well.
I too think he worded it quite well. I'd only add that in my humble opinion most users here own sound cards that would fall into that same general range. Being that they don't have the analog side of the metering as ashcat pointed out (I was hoping someone would), -18 RMS is often where they fall. At least the ones I have had my hands on either showed this in the spec sheet or is how it measured when I used my VU meters to set the incoming signal and actually measure it. Happy to be wrong but that's what I've seen much of the time in practice and in print. That doesn't make it a hard fast rule and why general was the chosen term.
__________________
Massaging electrons sure is a lot of fun.
karbomusic is online now   Reply With Quote
Old 09-01-2014, 09:28 AM   #10
Cosmic
Human being with feelings
 
Cosmic's Avatar
 
Join Date: Jul 2010
Location: Online
Posts: 3,874
Default

Practice.

The only way to find out what is correct for you is to learn it
Cosmic is offline   Reply With Quote
Old 09-01-2014, 10:13 AM   #11
Aesis
Human being with feelings
 
Join Date: Jan 2011
Posts: 182
Default

You are way overcomplcating this.

Take a synth with a sine wave, run it into the preamp of the interface with something like voxengo span open. Increase the volume till it peaks and see if any harmonics actually pop up, then alter the frequency and do the same. The funny thing is, no one has bothered to do such a simple test to actually see if there is any basis to to this stuff. Some interfaces are probably much better than others. And of course distorting past the low pass filter is where the really ugly sound comes from in most sources.

I believe that the claim has for some reason become way over exaggerated, especially in this era of ultra saturated sound.

Quote:
but it sounded like you expect perceptible differences in audio quality
Well if you hear no difference, should you trust your ears or some ppl who claim stuff who haven't even got an ounce of proof really?

Last edited by Aesis; 09-01-2014 at 10:20 AM.
Aesis is offline   Reply With Quote
Old 09-01-2014, 10:21 AM   #12
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 10,244
Default

Quote:
The funny thing is, no one has bothered to do such a simple test to actually see if there is any basis to to this stuff. Some interfaces are probably much better than others
That's because of three reasons:

1. Specs are published so you can follow them removing the need to spend all your time testing so you can make music instead and know your covered. Millions of dollars were spent doing this testing for us to allow just that, follow the spec.

2. It's not just about the interface, it's about the analog circuits and gear before it.

3. It's a nominal range for nominal reasons. Had anyone just followed that, there would be no need for the complexity.

Quote:
Well if you hear no difference, should you trust your ears or some ppl who claim stuff who haven't even got an ounce of proof really?
If what you immediately hear is the only thing you care about, you should go with your advice and not worry about it. Simple enough no? Record at -9 and have a grand time.

The tires on my car are spec'd at 155MPH, I know I can test them and run 180 but it would be silly to jump out of the car and exclaim that the 155 spec is BS due to that test. In the long run, it's best to follow the spec and not worry about it.
__________________
Massaging electrons sure is a lot of fun.

Last edited by karbomusic; 09-01-2014 at 10:31 AM.
karbomusic is online now   Reply With Quote
Old 09-01-2014, 10:28 AM   #13
Aesis
Human being with feelings
 
Join Date: Jan 2011
Posts: 182
Default

Quote:
Originally Posted by karbomusic View Post
That's because of three reasons:

1. Specs are published so you can follow them removing the need to spend all your time testing so you can make music instead.
I don't see any specs on distortion at specific amplitudes. Maybe some manufacturers publish such figures in which case there is no questions about any of this in the first place, but most don't.

Quote:
Originally Posted by karbomusic View Post
2. It's not just about the interface, it's about the analog circuits and gear before it.
Well, that would be included in my test.

Quote:
Originally Posted by karbomusic View Post
]
3. It's a nominal range for nominal reasons. Had anyone just followed that, there would be no need for the complexity.
I don't think this is necessarily the case.

Quote:
Originally Posted by karbomusic View Post
If what you immediately hear is the only thing you care about, you should go with your advice and not worry about it. Simple enough no?
I like proof, personally. It's simple and effective.
Aesis is offline   Reply With Quote
Old 09-01-2014, 10:39 AM   #14
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 10,244
Default

Quote:
Originally Posted by Aesis View Post
I don't think this is necessarily the case.

If you don't think so, all you need to do is show that +10 to +18 dB VU can be considered just as "nominal" spec wise as running any other way. If everyone agrees that is what nominal means in the analog world, I'm good with it. While we await that consensus, there is surely nothing wrong or misleading about advising someone to set their gear to that middle ground/nominal area.
__________________
Massaging electrons sure is a lot of fun.

Last edited by karbomusic; 09-01-2014 at 11:27 AM.
karbomusic is online now   Reply With Quote
Old 09-01-2014, 11:39 AM   #15
Aesis
Human being with feelings
 
Join Date: Jan 2011
Posts: 182
Default

Quote:
Originally Posted by karbomusic View Post
If you don't think so, all you need to do is show that +10 to +18 dB VU can be considered just as "nominal" spec wise as running any other way. If everyone agrees that is what nominal means in the analog world, I'm good with it. While we await that consensus, there is surely nothing wrong or misleading about advising someone to set their gear to that middle ground/nominal area.
Of course not, it makes sense that the middle range is more linear than the top because most analog components are that way.

However, I have seen credible persons (professionals in this field) saying that recording at higher levels than that will RUIN recordings. These kind of over-exaggerated claims just don't hold water. Meanwhile, no one actually has tested just how nonlinear the near clipping range actually is on various dacs. If it turns out to be very slight difference who even cares? Especially when more dynamic stuff like an orchestra can benefit from more gain.
Aesis is offline   Reply With Quote
Old 09-01-2014, 11:48 AM   #16
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 10,244
Default

Quote:
Originally Posted by Aesis View Post
Of course not, it makes sense that the middle range is more linear than the top because most analog components are that way.
Great, maybe one of my previous clarifications was unclear then since the above is what I've been trying to get across much of the entire time. It's why I kept using terms like range, nominal, rule of thumb, common sense etc. I'll do my best to step away from it now because all the ABX talk is unrelated to what I was expressing.
__________________
Massaging electrons sure is a lot of fun.
karbomusic is online now   Reply With Quote
Old 09-01-2014, 11:55 AM   #17
drtedtan
Human being with feelings
 
drtedtan's Avatar
 
Join Date: Jul 2010
Posts: 156
Default

Quote:
Originally Posted by Aesis View Post
I don't see any specs on distortion at specific amplitudes. Maybe some manufacturers publish such figures in which case there is no questions about any of this in the first place, but most don't.
How can you have a spec for distortion at a given amplitude when every source signal you track will be different and therefore clip differently. You'd have to reference that distortion measurement to a specific type of signal, say a 1K sine wav or something for it to be repeatable and meaningful.



Quote:
Originally Posted by Aesis View Post
However, I have seen credible persons (professionals in this field) saying that recording at higher levels than that will RUIN recordings.
Those claims are typically based on how the A/D converter sounds at various levels and/or that engineer's experience tracking as hot as possible to avoid the noise floor back in the days of 16 bit PCM digital recording. If you've tested your gear and found a specific level to work for you, go with it. Especially in cases of recording a DI guitar to run into a sim because that initial signal will be so distorted by the time it reemerges from the amp sim that the original signal isn't as important as it would be for, as an example, an acoustic guitar or vocal track.
drtedtan is offline   Reply With Quote
Old 09-01-2014, 12:27 PM   #18
G-Sun
Human being with feelings
 
G-Sun's Avatar
 
Join Date: May 2010
Location: Norway
Posts: 3,230
Default

I record peaks between -24 and -9dbFS.
End of story.
G-Sun is offline   Reply With Quote
Old 09-01-2014, 12:59 PM   #19
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 372
Default

Thanks for the thoughts, all:

Quote:
Originally Posted by brainwreck View Post
Does it sound a little different? Yes. Might one person prefer a lower level based on sound quality alone? Yes.
If you say so :-) -- I have never heard this effect. IMO, what this debate could really use is you (or anyone) repeating your experiment with a single performance multed to two inputs, and posting two normalized files so we can blind ABX them and see if anyone can discriminate between them better than chance.

If anyone thinks there are requirements to being able to tell the difference (high-end monitoring, or it only matters when the signals are used in a mix, or only matters with certain types of signals, etc), please state them and we can do that experiment, too.

Quote:
Originally Posted by kindafishy View Post
As far as the level that is captured in REAPER, I don't see how it matters. The digital level is relative so the s/n ration is the same whether you are recording at -12dB, -18dB or -30dB RMS.
We're talking here about the interface between analog and digital. Obviously you can't just record at arbitrarily low levels without noise repercussions. My interface happens to reveal its noise floor a little when going in with -18 dBFS RMS peaks. Maybe better interfaces don't. But all interfaces will at some point, as I'm sure you already know. I just noted it as a potential drawback to recording at -18 dBFS RMS peaks, albeit a tiny and probably insignificant one.

Quote:
Originally Posted by karbomusic View Post
...it's RMS not peak, this confuses things unnecessarily by using the wrong terms.
Apologies for the thread title. In the post, I refer to "-18 dBFS RMS peaks" ten times; six of those just as "-18" and four of them I specify "RMS" and sometimes "RMS peak", which is how the rule/rule of thumb was presented in the other post. I used "-18" as shorthand, because "peaking at -18 dBFS RMS" is cumbersome. Seemed clear enough to me, but I'll be explicit from now on.

Quote:
Shame on you for leaving out the "of thumb" part which I have painstakenly defined and used multiple times now.
Sorry for any offense -- I left it out because I wasn't addressing only you, but the "rule" in general, which is indeed a "rule" for many people. I took pains in my post above not to ascribe any particular stance to you: "maybe I misunderstood and you were only referring to clipping", "even if you don't, there are plenty out there who do", etc. I don't feel that shaming is warranted, here; there was nothing disingenuous about my post.

Quote:
So please, while you guys hash this out without me, would you please do me the simple favor of actually thinking instead of seeing the word "rule" then allowing your head to explode before understanding what comes after?
We're having this discussion in the context of lots of people who do consider it more than a rule of thumb. I at least am only worried about the presentation of the claim/rule/rule of thumb in the context of purported discernible differences in the character of the recorded audio. Restricted to just that issue, whether someone presents it as a just-in-case guideline or a hard and fast rule is just a matter of degree.

Quote:
Originally Posted by ashcat_lt View Post
Does it sound better? That's kind of subjective isn't it?
Sure -- but whether it sounds different at all or anyone can distinguish the difference isn't so subjective. :-)

Quote:
Some analog stages might sound kinda good when pushed up into distortion, some just sound like crap, and some folks might like the overdrive from this one or the other one, and some folks might thing it sounds better if it comes in as clean as possible.
Quote:
a lot of analog gear will start to get non-linear well before it even hits its own maximum output level.
I believe that about analog gear in general, but do you think that's true for prosumer/pro ADC interfaces?

Anyone out there with one of these devices that sounds like crap when the levels get hot, or with a device that sounds at all different, please post two audio files of the same signal to two inputs, one gained at -18 dBFS RMS peaks and one close to but not over clipping, and let's hear it. I believe this is a possible scenario, I just get the impression that it's nearly non-existent among most gear. Let's hear the examples and I can stop yammering about it.

Quote:
Originally Posted by karbomusic
If what you immediately hear is the only thing you care about, you should go with your advice and not worry about it. Simple enough no? Record at -9 and have a grand time.
I worry about immediate and long-term differences, and I can find neither, so I do record hotter. I believe it's worth getting to the bottom of the claim, though, to save myself and anyone else the wondering. I like to be intimiately familiar with my tools.

Quote:
Originally Posted by karbomusic
The tires on my car are spec'd at 155MPH, I know I can test them and run 180 but it would be silly to jump out of the car and exclaim that the 155 spec is BS due to that test. In the long run, it's best to follow the spec and not worry about it.
Let me adjust the analogy (in a fanciful way that is obviously not accurate re: tires): people on the internet say you have a smoother ride, the tires last longer, and you get better fuel efficiency if you drive at 70mph instead of 60mph. You accept the smooth ride claim and the longevity claim, but you want to test the fuel efficiency claim, so you do some tests with your car and find out it's not true. You ask others to test their cars, but no one will, though they (not accusing karbomusic) claim it's true as a general rule, and some even say that doing the test is meaningless or wrong for some inscrutable reason. I think anyone would naturally start to wonder if the fuel efficiency claim had any merit.

It irks me when people (not accusing karbomusic) repeat a common claim on the internet, assert that it is important, and when anyone wants to check out that claim they suddenly say "hey it's not important, just use your ears and get back to making music and stop sweating the small stuff".

The -18 dBFS RMS peak thing contains a simple claim/rule/rule of thumb that says that some/many/most ADC interfaces will sound different in some way when receiving signals at that level.

There are other reasons for the recommendation, having to do with matching analog gear, levels in the DAW, etc, but the part about the audio sounding better, or different at all, is a simple, testable claim.

I'm not arguing, and am not aware of anyone who is arguing, that recording at that level should never be recommended, just that it's suspiciously hard to find anyone willing to challenge the audio quality aspect of it.

You can find supporting examples for almost any audio claim that has merit. Not all are testable, but this one is.

Incidentally, karbomusic or anyone else: could you quote your interface manual on this issue? I'm curious to know how they word it; do they just recommend -18 dBFS RMS peaks (or wherever they spec it) as a general rule without explanation, or do they make statements about audio quality?

My Firestudio Mobile manual makes no mention of recommended recording levels at all. It specifies maximum input levels only.

Quote:
Originally Posted by karbomusic
there is surely nothing wrong or misleading about advising someone to set their gear to that middle ground/nominal area.
Advising that is fine with the right wording. If someone advises it because they say it sounds demonstratively better than recording non-clipped hotter recordings, I have a problem with that (until proven wrong, which hopefully will happen any time now).
clepsydrae is offline   Reply With Quote
Old 09-01-2014, 01:21 PM   #20
beingmf
Human being with feelings
 
beingmf's Avatar
 
Join Date: Jul 2007
Posts: 2,521
Default

Quote:
Originally Posted by clepsydrae View Post
My Firestudio Mobile manual makes no mention of recommended recording levels at all. It specifies maximum input levels only.
Yes, and that's it.



If you max out the inputs, you get a THD of 0,5%, that's a value you wouldn't want on stacked tracks!
beingmf is offline   Reply With Quote
Old 09-01-2014, 01:23 PM   #21
whiteaxxxe
Human being with feelings
 
whiteaxxxe's Avatar
 
Join Date: Jul 2014
Location: United States of Europe, Germany, Mönchengladbach
Posts: 434
Default

Quote:
Originally Posted by G-Sun View Post
I record peaks between -24 and -9dbFS.
End of story.
such posts as yours ruin the whole discussion. :-)))))))))))))))))))

on Gearslutz such a thread can easily go over a 1000 pages ...

I am with you, G-Sun. but there are always people that think analog, so to say. and mix that up with some half-science and come to crude conclusions. that wouldnt be so bad, if they wouldnt spread their wisdom in all these audio-forums, spiced up with wrong citations or citations from the wrong context of famous people or so-called pros.

ask Ethan Winer or Paul Frindle for the self-dynamics of threads to such issues as here ... its unbelievable.
whiteaxxxe is offline   Reply With Quote
Old 09-01-2014, 01:32 PM   #22
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 10,244
Default

Quote:
Incidentally, karbomusic or anyone else: could you quote your interface manual on this issue? I'm curious to know how they word it; do they just recommend -18 dBFS RMS peaks (or wherever they spec it) as a general rule without explanation, or do they make statements about audio quality?
Mine shows +19. Specs don't recommend sound quality, they list test results and you use experience and interpretation to make a good judgment call just like we would with any datasheet. The AD converter has an analog side to it which is also bound by the same laws of physics the rest of the analog world is. That doesn't make any statement other than a reminder that we aren't yet out of the analog woods so to speak until it is actually 1s and 0s. I don't need to (nor am I compelled to) test it to be OK with trying to stay in the you guessed it, nominal range. We're both on different philosophical pages here.
__________________
Massaging electrons sure is a lot of fun.

Last edited by karbomusic; 09-01-2014 at 02:19 PM.
karbomusic is online now   Reply With Quote
Old 09-01-2014, 02:02 PM   #23
Reno.thestraws
Human being with feelings
 
Reno.thestraws's Avatar
 
Join Date: Nov 2009
Location: Belgium
Posts: 6,714
Default

It's simple as that

Reno.thestraws is online now   Reply With Quote
Old 09-01-2014, 02:13 PM   #24
Reno.thestraws
Human being with feelings
 
Reno.thestraws's Avatar
 
Join Date: Nov 2009
Location: Belgium
Posts: 6,714
Default

Check out this video tutorial

http://www.puremix.net/video/recording-levels.html

17$ and you'll understand everything
Reno.thestraws is online now   Reply With Quote
Old 09-01-2014, 02:38 PM   #25
ashcat_lt
Human being with feelings
 
Join Date: Dec 2012
Posts: 425
Default

Quote:
Originally Posted by Reno.thestraws View Post
It's simple as that
It's not exactly that simple. In actual practice there is no standard as to where any of these scales line up one to the other. Every console (or other analog stage) will have a different amount of headroom, and every interface will have a different maximum input level. When any two different pieces of gear have the same specs, it's only by accident.
ashcat_lt is offline   Reply With Quote
Old 09-01-2014, 02:48 PM   #26
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 372
Default

(@Reno.thestraws -- I don't see how that graph addresses the audio quality issue I'm talking about, so I assume you were addressing someone else...)

Quote:
Originally Posted by beingmf View Post
If you max out the inputs, you get a THD of 0,5%, that's a value you wouldn't want on stacked tracks!
I'm definitely no expert, but my very crude understanding is that 0.5% THD+N at 1k is inconsequential (and anyway that spec is only for a sustained sine wave at max input level). See this interesting study re: THD+N in music:

http://www.axiomaudio.com/distortion

Maybe you can provide some testable audio examples demonstrating otherwise? :-)

And the "stacking" thing is another myth altogether, as far as I have learned from those smarter than me. See Ethan Winer's take on it here. Even if it wasn't a myth, we could test the effect by stacking a bunch of tracks and witnessing the results, but apparently no one wants to do that, or ever has done that, and posted their results.

Curious to know what the device makers themselves might say, I emailed a few audio device manufacturers. The content of my email is at the end of this post, for reference.

RME responded almost immediately, so, big kudos to them. Here is what they said (responding only to my below email, not to any forum threads, etc):

Quote:
Originally Posted by RME
there is no such difference, certainly not by design. I would personally tend to call such claims voodoo...
I'll let you know if any other manufacturers get back.

Quote:
Originally Posted by karbomusic
Specs don't recommend sound quality, they list test results and you use experience and interpretation to make a good judgment call just like we would with any datasheet.
Specs don't recommend for sound quality, but I asked about device manuals, most of which certainly make all kinds of recommendations about sound quality. You'd think some of them would recommend -18 dBFS RMS peaks or similar, if it was really a thing? I haven't seen anything like that, but I'd be very interested if it's out there.

Quote:
The AD converter has an analog side to it which is also bound by the same laws of physics the rest of the analog world is. I don't need to (nor am I compelled to) test it/prove it to be OK with trying to stay in the you guessed it, nominal range.
What I believe we are all compelled to do is to be prudent about what we recommend to others and the reasoning we give for that advice. If we're advising people to do things which are a waste of their time, we should stop (not saying you did this). I want to know if recording at -18 dBFS RMS peaks makes any difference to audio quality. Whoever is on board with the mission of getting to the bottom of this question, please lend a hand.

--------

Quote:
Originally Posted by clepsydrae's email
Hi there -- question about RME interfaces for your techs, if they have the time:

There is a claim made often on the internet that many audio interfaces sound "better" in various ways if gainstaging is such that signals coming in to the box result in RMS peaking at about -18 dBFS. This is specifically claimed in contrast to recording "hot" signals that peak just under digital/analog clipping.

There are many reasons for the recommendation, including matching analog gear voltages, convenience of workflow once in the DAW, and so on.

I'm curious only about the audio quality aspect of the claim: would your gear sound any different if signals are recorded at -18 dBFS RMS peaks vs. recorded at -0.5 dBFS (non-RMS) peaks, provided that no clipping has occurred, and once normalized in the DAW, etc?

If there would be a difference besides the noise floor, under what circumstances would this difference be discernible? (e.g. high-end monitoring? expert listeners? certain types of audio? etc.)

Thanks for your time!
clepsydrae is offline   Reply With Quote
Old 09-01-2014, 03:14 PM   #27
ashcat_lt
Human being with feelings
 
Join Date: Dec 2012
Posts: 425
Default

Quote:
Originally Posted by clepsydrae View Post
What I believe we are all compelled to do is to be prudent about what we recommend to others and the reasoning we give for that advice. If we're advising people to do things which are a waste of their time, we should stop (not saying you did this). I want to know if recording at -18 dBFS RMS peaks makes any difference to audio quality. Whoever is on board with the mission of getting to the bottom of this question, please lend a hand.
I think that if you're advising anybody to do anything other than use proper gain staging in the analog realm and let the digital levels fall where they may, then you're kind of wasting everybody's time. Even better advice: Turn the knobs till it sounds good!

I'm trying to figure out how to do this test that you propose in any meaningful way, but it's tough to do without adding more variables to the equation...
ashcat_lt is offline   Reply With Quote
Old 09-01-2014, 03:31 PM   #28
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 372
Default

Quote:
Originally Posted by ashcat_lt View Post
I think that if you're advising anybody to do anything other than use proper gain staging in the analog realm and let the digital levels fall where they may, then you're kind of wasting everybody's time.
The definition of "proper" being the devil in the details, here. Regardless, I certainly don't plan to go on a mission admonishing people to record as hot as possible. What I'm on about is the parade of posts in forums that say "you're recording more than -18 dBFS RMS peaks? That's the problem with your kick drum/guitar/mix/stereo serparation/punch/high-end/whatever... record lower and everything will change for you."

Quote:
I'm trying to figure out how to do this test that you propose in any meaningful way, but it's tough to do without adding more variables to the equation...
I think the easiest may be to use a mixer, send a mono signal in, and send it out two outputs that you can adjust the balance between... adjust the overall level and the balance such that one goes to -18 dBFS RMS and the other to just under clipping (analog or digital), and go into inputs in your interface that are at unity gain (or which lack gain knobs). If you worry that individual deviations in the two outputs from the mixer or the inputs into the interface might confound things, repeat the test after switching the routing up. Or just use a splitter cable and mult to the inputs and gain one down?

Record, normalize, and render (preserving 24bit, ideally).

Sound good, or am i overlooking something? Did you see the test I posted in the other thread? Was it not meaningful due to some mistake I didn't think of?
clepsydrae is offline   Reply With Quote
Old 09-01-2014, 04:50 PM   #29
ashcat_lt
Human being with feelings
 
Join Date: Dec 2012
Posts: 425
Default

But does that tell us about the mixer or the converter? Or do we not care? Maybe the point is just to evaluate the system as a whole? I might try a couple things...
ashcat_lt is offline   Reply With Quote
Old 09-01-2014, 06:51 PM   #30
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 372
Default

Quote:
Originally Posted by ashcat_lt View Post
But does that tell us about the mixer or the converter? Or do we not care? Maybe the point is just to evaluate the system as a whole? I might try a couple things...
I see your point... i suppose it's impossible to decisively eliminate a confounding variable in such a situation: you want to compare two things that are identical except for their level, but getting one source to two different levels will always involve some other tech. I say go for it anyway... if there is no perceptible difference then it's moot. If there is a difference, it could be investigated, though if there was a difference my money would be on the converter, not the mixer. Hard to imagine a little attenuation causing noticeable changes.

Look forward to hearing about whatever you try...
clepsydrae is offline   Reply With Quote
Old 09-01-2014, 07:05 PM   #31
Bristol Posse
Human being with feelings
 
Bristol Posse's Avatar
 
Join Date: Jan 2011
Location: Southern California
Posts: 432
Default

I generally shoot for line level RMS of 0VU (+4dBu) because I have an extensive analog tracking/mixing chain

How that shows up in reaper depends on which converters I happen to be using and how I have them calibrated

It can be anywhere from -19dBFS (RMS) to -9dBFS (RMS). I don't care about peaks so long as I don't clip

if you're not using a lot of analog gear in tracking and mixing then it becomes less relevant, unless you are using really bad interfaces that add a lot of noise and THD at higher levels or have poor S/N performance

I think the argument in the DAW ITB realm is mostly about retaining headroom and keeping the faders close to unity in mixing for easier use due to better resolution and fine adjustment closer to unity on the Fader

Also Waves and UAD (maybe slate and others too) for the most part use -18dBFS as simulated 0VU for their emulation plugins where they will give the most linear performance with best S/N ration (like analog gear)
__________________
An Infinite Number Of Chimps

Last edited by Bristol Posse; 09-01-2014 at 07:12 PM.
Bristol Posse is offline   Reply With Quote
Old 09-01-2014, 07:39 PM   #32
drtedtan
Human being with feelings
 
drtedtan's Avatar
 
Join Date: Jul 2010
Posts: 156
Default

Quote:
Originally Posted by clepsydrae View Post
-18 dBFS RMS peaks
What is this -18 dBFS RMS peak you keep speaking of?

Do you mean -18dBFS RMS?

Or perhaps -18dBFS peak?

It can't be both, so go research what RMS, peak and nominal mean so you can then come back here and have an intelligent conversation.
drtedtan is offline   Reply With Quote
Old 09-01-2014, 07:45 PM   #33
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 372
Default

Quote:
Originally Posted by drtedtan View Post
What is this -18 dBFS RMS peak you keep speaking of?

Do you mean -18dBFS RMS?

Or perhaps -18dBFS peak?

It can't be both, so go research what RMS, peak and nominal mean so you can then come back here and have an intelligent conversation.
No need to be insulting. The idea, as was expressed in the previous linked thread, is that the level should be such that an RMS meter, such as that found on Reaper's master fader, would peak at -18 dBFS. RMS meters aren't calculating a single RMS value over the entire audio program, but work on a moving window, so they can indeed have a peak.
clepsydrae is offline   Reply With Quote
Old 09-01-2014, 08:03 PM   #34
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 10,244
Default

Quote:
Originally Posted by clepsydrae View Post
The idea, as was expressed in the previous linked thread, is that the level should be such that an RMS meter, such as that found on Reaper's master fader, would peak at -18 dBFS.
I replied to the very first occurrence of -18 dBFS in that thread and corrected part of that assumption before it could ever have been made, or at least tried to get things in the direction of not thinking the above:

http://forums.cockos.com/showpost.ph...7&postcount=35

I can't find where anyone meant what you wrote above, I did search quickly though. Are you sure statements like this aren't throwing you?

Quote:
Maybe people don't understand the RMS part? Set to -18 db RMS and see where peaks go.
The above means set your preamp so that the RMS value in reaper shows -18 dBFS RMS, then look at the peak meter right beside and see where they are; with an electric guitar in a DI and normal playing those peaks would be above -18dB which is the same reason that when setting RMS to -9 there are overs. The reason that and my first quote above was written was to point out not to look at the RMS value as a peak but do look at the peaks in relation to the RMS value.
__________________
Massaging electrons sure is a lot of fun.

Last edited by karbomusic; 09-01-2014 at 08:27 PM.
karbomusic is online now   Reply With Quote
Old 09-01-2014, 08:38 PM   #35
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 372
Default

Quote:
Originally Posted by karbomusic View Post
I replied to the very first occurrence of -18 dBFS in that thread and corrected that assumption before it could ever have been made:

http://forums.cockos.com/showpost.ph...7&postcount=35

I can't find where anyone said what you wrote above, I did search quickly though. Are you sure statements like this are throwing you?
Your post you link to is in response to Reno.thestraws who said "Record at -18 RMS dbfs at max."

Apparently I misinterpreted that to mean that the RMS value should be, at max, -18, and I take it you're saying he meant it should be roughly at -18, not at max -18. Either way, the RMS meter on the samples I posted (which are complex, finished, compressed mixes) was not bouncing around much at all: only about 3-6 dB (just checked them again.)

The first draft of my samples were RMS-meter-peaking around -13, so that's around ~-17.5 "average RMS". The current ones were RMS-meter-peaking just up -18, so let's call that ~-22.5.

Both sets of samples were indistinguishable from those peaking at -0.5. I presume no one is suggesting that the effect is gone at -17.5 and gone at -22.5, but present somewhere in between, so I can't believe doing another test with a small variation in RMS meter level is going to matter. If anything, the current samples (~-22.5) should exhibit more of the possible positive effect (if a tiny bit more noise) since they are further away from the supposedly "distorted" upper end of the analog input section.

I note that, according to my web logs, zero people even downloaded the examples I posted, which is an interesting comment in itself, either on the subject at hand or on a lack of faith in my methodology. :-)
clepsydrae is offline   Reply With Quote
Old 09-01-2014, 09:29 PM   #36
Lawrence
Human being with feelings
 
Join Date: Mar 2007
Posts: 15,620
Default

Quote:
Originally Posted by ashcat_lt View Post
The real point of the whole thing is to run your analog gear at sensible levels to avoid adding analog distortion on the way into the box. This usually means keeping your average levels around 0dbVU = +4dbu.
I've said it before and I'll say it again... all of this crap is really for consumers. People with outboard gear that transitioned from analog recording set their levels on their preamps on VU meters. With a correctly aligned system everything downstream from the outboard takes care of itself, all by itself, because that's exactly what the scale was designed for, to accommodate everything up to high headroom pro gear.

Everything else is people using cheap gear with 3 stage led's and not knowing what level to record at and obsessing over it... while recording with $200 USB devices who's converters are actually at -15 (because they never bothered to check or don't know how to check) and making -18 a rule, when it's not.
__________________
We can easily forgive a child who is afraid of the dark; the real tragedy of life is when men are afraid of the light. - Plato
Lawrence is offline   Reply With Quote
Old 09-01-2014, 09:43 PM   #37
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 10,244
Default

Quote:
Apparently I misinterpreted that to mean that the RMS value should be, at max, -18, and I take it you're saying he meant it should be roughly at -18, not at max -18.
Yes.

Quote:
The first draft of my samples were RMS-meter-peaking around -13, so that's around ~-17.5 "average RMS". The current ones were RMS-meter-peaking just up -18, so let's call that ~-22.5.
I never really replied about those tests since we were speaking of instruments that need to be amplified, and that setting gain in nominal ranges (let's call it 0 dbVU just to simplify) usually results in the "range" I had been speaking of. That's what some of us mean by "let the values fall as they may" so we don't have to modify our carefully chosen analog gain structure just to satisfy the DAW. Whether someone finds a difference in that or not by the time it arrives in the DAW makes no difference to me.
__________________
Massaging electrons sure is a lot of fun.
karbomusic is online now   Reply With Quote
Old 09-01-2014, 09:50 PM   #38
Lawrence
Human being with feelings
 
Join Date: Mar 2007
Posts: 15,620
Default

Quote:
Apparently I misinterpreted that to mean that the RMS value should be, at max, -18, and I take it you're saying he meant it should be roughly at -18, not at max -18.
That's called "The Gearslutz Rule". It went viral in about a month.

A guy, a pro engineer of some esteem and talent, talked about levels and what happened was that people who had no actual clue what he was really talking about just ran with it (-18) and now it's all over the net and newbies are latching onto it like the Holy Grail.

I - still to this day - read people talking about trimming all of their peak levels across 24 tracks or whatever to "-18" before they mix because (they thought) famous engineer x said so.
__________________
We can easily forgive a child who is afraid of the dark; the real tragedy of life is when men are afraid of the light. - Plato
Lawrence is offline   Reply With Quote
Old 09-01-2014, 10:58 PM   #39
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 372
Default

Quote:
Originally Posted by Lawrence View Post
...and now it's all over the net and newbies are latching onto it like the Holy Grail.
Thing is, it's not just newbies, c.f. John Scrip of MASSIVE Mastering (post from 2006). He does make the case for the levels being more convenient and sensible for analog interoperation, but note also the unambiguous claims of not just a detectable difference but a big difference in sound quality specifically due to "overdriving" the analog input stage of the converter. Specifically he claims that many other engineers find the difference stunning, so it's not that it requires golden ears and a perfect environment to hear it.

The thread goes on and on, with many posters (many with multiple thousands of posts under their belts) concurring heartily. John says the audio at -18 has "The best sound, the lowest distortion, the best signal-to-noise, the most clarity, the most focus - all of the "good" things." Recording hot yields audio that is "distorted, pinched, veiled, dynamically questionable", etc.

(There are no audio examples provided in the entire 16-page thread. There are also 2 or 3 mentions of pro audio interface manuals recommending that you record hot without clipping, which are presented as lamentable cluelessness on manufacturers' part rather than a counter-argument.)

Maybe it was a big deal for 2006-era hardware? Maybe it's vestigial from the analog days where older engineers got used to analog input stages being sensitive? I sure wish someone would do some tests and post examples so we could get a better idea of how common it might be among devices.

Another poster: "Tracked an orchestra last night - One sustained section hit around -10dBFS. Spooked the hell outta me... You could hear the tonality changing (not for the better either). Luckily, it was a short section and no one else is ever going to notice. But I still wish I would've been a few dB lower on the input levels... Darn timpanists..."

This stuff is everywhere on the 'net, and not just from teenagers on gearslutz, which is why it concerns me. (No offense to teenagers.)

I originally started posting about this precisely because so many experienced people swore it was the gospel truth, and that's why I'm still curious. I've just grown a bit more skeptical as the evidence-free threads have worn on.
clepsydrae is offline   Reply With Quote
Old 09-01-2014, 11:21 PM   #40
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 372
Default

Here is a very interesting paragraph from an article in Sound on Sound about analog/digital levels. They mostly aren't addressing recording levels but rather alignment of analog/digital levels. This is from May 2000 (!) so ADC's are obviously even better now... even so:

Quote:
Originally Posted by SoundOnSound
Building in this kind of allowance for headroom is essential when recording unpredictable material which may well contain unexpected transient peaks of substantial level. However, it is totally unnecessary when working with controlled, post-produced material which has benefited from compression or limiting to tame transient peaks. With peak levels ironed out, the requirement for large amounts of headroom is removed and, following the long tradition of 'louder is better', it makes sense to adjust the overall level of the music to peak as closely as possible to the maximum level — 0dBFS.
(emphasis mine)

So I take it that their opinion is that the converter's analog stage will do nothing negative to the signal. And I further infer that when recording a new track the signal should also be as hot as possible, provided that it will not clip. Avoiding clipping may result in -18 dBFS RMS average level anyway, but no reason to avoid hotter, according to what I take from that article.
clepsydrae is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 04:58 PM.


Powered by vBulletin® Version 3.8.4
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.