Why 18?

[SouthSIDE Glen]

I've been researching the whole gain staging thing in an attempt to document it a bit better for the rookies to this racket, and frankly there's a bit of "common wisdom" that's passed around by most of the veterans here (I have been guilty of it myself) to which I can find no evidence to support; that a de facto standard for conversion from dBu to dBFS is +4dBu (0VU) to -18dBFS, though that calibration can vary from device to device.

Frankly the only published standards (EBU/ISO, SMPTE) do not match this figure. Does anyone have any tips or links documenting where the -18dBFS spec comes from? Is it simply because manufacturers think that 18dB of crest factor is "better" or an easier sell than 16dB (the SMPTE standard) or 14dB (the EBU standard)? While I personally like the 18dB of headroom myself; it makes more sense to me to have more peak headroom, it is a rather arbitrary number as far as I can tell. Why not -20dBFS, for example? While there is gear that'l go that far, it seems to be the exception at the extreme, not the de facto standard that -18dBFS appears to have become.

Or is there actually a third official standard that I can't find that specifies -18dBFS?

Any standards buffs out there?

G.

 

[boingoman]

Here is a stupid stupid guess.

LEDs and 3db increments. Very easy to meter -18db.

I can do better, I'm sure.

 

[The Audio Cave]

Although I can't find a direct reference to it, it appears that 18dbfs is an AES/EBU standard. I won't pay $25 for the document to find out for sure.

Apparently that's also the reference level used for broadcast post audio. Why -18 instead of -20/-16? I have no idea.

You can image how audio recorded on a converter referenced at (calibrated to) 20dbfs might look on another system calibrated to 15dbfs if the audio was tracked up to 2dbfs.

 

[RAMI]

As long as you record just slightly below 0 (24 bit digital), you'll be ok.











(just kidding)

 

[masteringhouse]

Glen,

BK has a pretty good chapter in his book regarding levels and in-house standards. If you haven't seen it already, check out:

http://www.digido.com/modules.php?name=News&file=article&sid=8

Also see:
http://www.magnisystems.com/application%20notes/sdm-560xapp1.htm

 

[SouthSIDE Glen]

Although I can't find a direct reference to it, it appears that 18dbfs is an AES/EBU standard. I won't pay $25 for the document to find out for sure.

Apparently that's also the reference level used for broadcast post audio. Why -18 instead of -20/-16? I have no idea.

Well, all the info I have found so far seems to contradict that.

To the first one, the only EBU/IEC/ISO (I have not seen AES included in there anywhere yet) standard I can find equates +4dBu to -14dBFS, with 0dBFS obviously then topping off at +18dBu.

And the only other standard I have found goes to SMPTE - which you'd think would cover broadcast post - also specified by a different IEC subsection, which equates +4dBu to -16dBFS, not -18.

I was thinking perhaps AES had something different perhaps which goes to -18, but like you I'm an even bigger cheapskate than the AES is (which is saying a lot.)

BK has a pretty good chapter in his book regarding levels and in-house standards.

As long as he doesn't try to make me drink that K-metering koolaid of his on the way there . I'll take a look.

G.

 

[Farview]

As far as i know, there is no real standard. My Motu is calibrated at -15db. John's Lavery's are -18db. It really seem like it's whatever the manufacturer feels like calibrating at.

 

[jmarques]

Glen,

BK has a pretty good chapter in his book regarding levels and in-house standards. If you haven't seen it already, check out:

http://www.digido.com/modules.php?name=News&file=article&sid=8

I haven't been here that long, but I don't think Glenn is all that interested in Bob Katz theories. At least when it comes to the K system.

 

[jmarques]

As long as he doesn't try to make me drink that K-metering koolaid of his on the way there . I'll take a look.

G.

Ah crap I thought I got my post in before this. LOL

 

[masteringhouse]

Sorry, I edited the previous post to include a better and more succinct explanation.

In North American video production facilities, the reference level is typically +4dBu and clip level is +24dBu, but there are a few that still run +8dBu and +28dBu respectively. The latter is a carryover from Ampex machines. Note that in both cases, the operating level in digital terms is -20dBFS so there is 20dB of headroom above the reference level.

There is an EBU standard that calls for an operating level of +4dBu and a clip level of +22dBu. This results in a digitally operating level of -18dBFS and 18dB of headroom. This practice is used in the UK, France, and by NHK in Japan.

This BK article is still worth reading though.

 

[SouthSIDE Glen]

Well, I just read the Katz article. The interestings in there are that

A) he recommends against the -18 calibration and argues for -20.

B) he seems to imply that the -18 de facto has its roots in the fact that that is what a lot of the original DAT machines were calibrated to. *Yikes*.

And no K-oolaid anyhwere. I got nothing against Katz in general, it's just his K system that drives me up a tree, that's all.

OK, I'll take a look at the other article now...

G.

 

[jmarques]

Hey I got BKs book and I think it's a excellent book to read, but Glenn has made some excellent post in the past regarding the K system.

Although I like that whole idea calibrating my monitoring to the K-system. I don't think it takes into account when you mixing music. Sometimes I feel the K-system feels to quite for the type of music I'm doing and I can't get something to sound the way I like to hear it, LOUD!!

Ok. Sorry I'm getting off topic.

-18dbfs??? Good question. Don't know.

 

[SouthSIDE Glen]

Yeah, I really don't want this thread to devolve into another K-System debate; BTDT in a few other threads already

OK, I read that magnisystems article, and I gotta tell you, I'm more confused than ever. Along with the quotes that Tom cited is the following quote:

"Since 0dBFS is the only thing that can't change in a digital world, one can look at SOL as either -20dBFS, -18dBFS or -12dBFS (depending on country)."

That does not jive at all with the IEC standards that I have found elsewhere (the several subsections of IEC 60268), that spec it at -16 for the US (per SMPTE) and -14 for Europe (per EBU/IEC).

So, basically, depending on which source one believes (and they are all pretty reliable sources normally), the standards exist at every even number from -20dBFS to -12dBFS, even within the same standards committees.

This is worse than figuring out the instrument frequency ranges for the instruments was, and that was conflicting enough!

G.

P.S. Anybody wanna bet against the idea that the buzzwords used in this thread show up ina a Walters thread sometime in the next 2-3 weeks?

 

[bennychico11]

To the first one, the only EBU/IEC/ISO (I have not seen AES included in there anywhere yet) standard I can find equates +4dBu to -14dBFS, with 0dBFS obviously then topping off at +18dBu.

And the only other standard I have found goes to SMPTE - which you'd think would cover broadcast post - also specified by a different IEC subsection, which equates +4dBu to -16dBFS, not -18.

You're talking about an input leveling standard but you refuse to accept a monitoring standard, Glen??!

anyway, where are you reading all this?
I'll add some links I have piled away:

Specification of the AES/EBU Interface

Alignment to SMPTE RP155, alignment level is 20 dB below maximum code
Alignment to EBU R68, alignment level is 18.06 dB below maximum code

Loudness Control In Digital Broadcast

The analog level required for 0 dBFS can vary from country to country. EBU R68 is used in most European countries. This standard specifies +18 dBu at 0 dBFS, while US installations use +24 dBu for 0 dBFS. In Japan, France and some other countries, converters may be calibrated for +22 dBu at 0 dBFS.

In European broadcast, the reference level is –18 dBFS (EBU R68), in the US –20 dBFS has been chosen (SMPTE RP155).

PBS Technical Operating Specifications

The operating level for reference tone and legacy analog system calibration is -20 dBFS per SMPTE RP155.


Audio Levels and Readings
pg. 8 contains this graph:





There are standards...but I don't think everyone follows them. Also, some manufacturers who do, may be in a different country than you. For example,
RME out of Germany says:

Input level for 0 dBFS @ +4 dBu: +13 dBu
for one of their units.

With +4 dBu selected, the according headroom meets the latest EBU recommendations for
Broadcast usage. At -10 dBV 12 to 15 dB headroom are common practice, each mixing desk
operating at -10 dBV is able to send and receive much higher levels.

Alesis says:

Analog I/O spect: +4 dBu nominal level (for -15 dBFS digital level) on balanced 1/4" TRS connectors

Personally I run test tone at -20dBFS...and keep my RMS levels around there. When recording spoken voice I may keep it a little hotter than that...but my MIX levels are around -20dBFS RMS.

 

[SouthSIDE Glen]

You're talking about an input leveling standard but you refuse to accept a monitoring standard, Glen??!

Two different things altogether, Ben . On the level "standard", it's simply a matter of providing a translation between two (or more) different scales of measurement, rather like the conversion between Farenheit and Centegrade; the air temperature remains whatever the air temperature is; all we're doing is measuring it. K-oolaid, however, is trying to set a standard for telling me what to set the actual air temperature to, with zero regards for my personal needs.

Besides, all I'm trying to do is put together a thing on gain structure for rookies, There was a thread here not long ago where someone posted a chart from SOS that showed "standard" dBFS scales that did not quite look "right" even though they came from IEC standards documents. Now we all know that the manufacturers treat these standards the way a cop treats turn signal laws and do whatever thay please as far as calibration (now I see that Alesis has added an odd number to the mix too) but I was hoping to at least be able to confirm some official, *basic* standard for a similar style chart, from which I could describe that "while X is the standard, your gear may vary in calibration yada yada yada". But I gotta have something on the digital side of the equation to get the idea of gain structure across properly.

But I think I'm coming to the conclusion that there are so many standards that there is no standard. Even within a single standards committee there exist different standards that contradict each other. Then when you bring other committees and organizations into it, the "standards" cover the gamut of possibilities.

I guess I'll just pick one or two of the "standards" from the most legitimate sources from a hat for the illustration and explain from there.

anyway, where are you reading all this?

All over the place, from SoundOnSound articles to IEC publications to Gearslutz transcripts to some .edu reference sites, to, yes, even Wikiality. And the info I'm giving here is *after* filtering out the numerous sources I found containing other yet more conflicting information that was not supported by any documentation or reputation whatsoever. Sorting signal from noise on the Internet can be an exhausting endeavor. I long for the days of editorial responsibility.

G.

 

[bennychico11]

All over the place, from SoundOnSound articles to IEC publications to Gearslutz transcripts to some .edu reference sites, to, yes, even Wikiality. And the info I'm giving here is *after* filtering out the numerous sources I found containing other yet more conflicting information that was not supported by any documentation or reputation whatsoever. Sorting signal from noise on the Internet can be an exhausting endeavor. I long for the days of editorial responsibility.

Well without downloading the SMPTE and EBU white pages myself, it appears SMPTE RP155 and EBU R68 are the recommended standards. Which are -20dBFS and -18(.06)

But I think I'm coming to the conclusion that there are so many standards that there is no standard. Even within a single standards committee there exist different standards that contradict each other. Then when you bring other committees and organizations into it, the "standards" cover the gamut of possibilities.

of course it's up to manufacturers to decided whether to follow this standard or not. And I agree it looks like it's a hit or miss when it comes to this. I think this is the whole reason we need to look at the specs. Instead of saying "+4dBU = -18dBFS"...we should just say "look at the specs of your gear." Perhaps the -18dBFS number everyone is recommending is just a safe, rough average. I don't think any of us need be incredibly anal when we record and make sure the meter always falls around there as if the end of the world will happen if it doesn't.
I think the point most of the members here who recommend it are trying to make is, "stay the hell away from 0dBFS when trying to record!"

 

[MCI2424]

I've been researching the whole gain staging thing in an attempt to document it a bit better for the rookies to this racket, and frankly there's a bit of "common wisdom" that's passed around by most of the veterans here (I have been guilty of it myself) to which I can find no evidence to support; that a de facto standard for conversion from dBu to dBFS is +4dBu (0VU) to -18dBFS, though that calibration can vary from device to device.

Frankly the only published standards (EBU/ISO, SMPTE) do not match this figure. Does anyone have any tips or links documenting where the -18dBFS spec comes from? Is it simply because manufacturers think that 18dB of crest factor is "better" or an easier sell than 16dB (the SMPTE standard) or 14dB (the EBU standard)? While I personally like the 18dB of headroom myself; it makes more sense to me to have more peak headroom, it is a rather arbitrary number as far as I can tell. Why not -20dBFS, for example? While there is gear that'l go that far, it seems to be the exception at the extreme, not the de facto standard that -18dBFS appears to have become.

Or is there actually a third official standard that I can't find that specifies -18dBFS?

Any standards buffs out there?

G.

Some info here:
http://emusician.com/met/emusic_plays_others_staying/index.html

I guess no real standard exists yet. The equipment designer knid'a sets his inputs for -18 at 0db (plus or minus) or not (plus or minus).

 

[SonicAlbert]

Besides, all I'm trying to do is put together a thing on gain structure for rookies, There was a thread here not long ago where someone posted a chart from SOS that showed "standard" dBFS scales that did not quite look "right" even though they came from IEC standards documents.

Well, this is one non-rookie that will appreciate what you are doing! That chart also looked wrong to me, and I think it is great you are trying to sort it out.

May I suggest you pick one or two "standards", as you are doing, but then also show where the rest of the "standards" lay against that. So in chart form (a correct one as I know you will do) we can see how the numbers line up.

I might suggest digital "0" as being the standard reference, as that is what most people here will be dealing with in their DAW's.

 

[Dogman]

I think 18 is the number because if she's any younger, you might be in some trouble with the law.

Oh wait....wrong forum....

 

[SouthSIDE Glen]

I just found another article that lies somehwere between distressing and hilarious (I prefer to take the hilarious POV myself ):

http://www.rane.com/pdf/bewilder.pdf

If you want to know just what I've dealt with in this search, this article sums it up very nicely. It also shows the problems associated with the IEC standards that are the basis for that SOS chart and article cited in that other thread.

Man, there was a famous article back in the 60s or 70s by Isaac Asimov called "Forget it!" where he found an old math textbook from the early 1800s or so that had a table of units of measure. He wrote this whole article going on about things like "firkins of beer vs. firkins of ale" and as to whether you use the "London firkin or the country firkin" and so on. It was a hilarious article about a hopeless attempt by the author of that book to make any sense out of the different units of measure. I used to have it in an Asimov book, but it got misplaced over the years. I have not been able to find it in the net anywhere. If someone finds it, I'd appreciate a link. Not only is it a funny article, but it puts across exactly the feeling I've had with this topic.

Good to see how much more advanced we've gotten in 200 years .

And I agree it looks like it's a hit or miss when it comes to this. I think this is the whole reason we need to look at the specs. Instead of saying "+4dBU = -18dBFS"...we should just say "look at the specs of your gear." Perhaps the -18dBFS number everyone is recommending is just a safe, rough average. I don't think any of us need be incredibly anal when we record and make sure the meter always falls around there as if the end of the world will happen if it doesn't.
I think the point most of the members here who recommend it are trying to make is, "stay the hell away from 0dBFS when trying to record!"

Yep, I agree with this 100%, and all that has to be explained. I expected that going in to this. I was just hoping that there was a definitive or at least de facto standard or two to start with in the chart, one that stood out authoritatively, that I could use as the launch point for the discussion.

It is clear to me now that it's a far muddier situation than that. Seeing as how it's a kind of "choose your poison" situation, that the IEC "standards" are apparenly somewhat problematic in their design (and a bit lacking in headroom, if you ask me), and that there are indeed a signifigant number of manufacturers and engineers who prefer either -18 or -20 as the reference, I'll go with those in the chart and text the real explanation out from there.

Now what did I do with that Photoshop eraser...

G.