EBU R-128 meets K-System v2, a possible future for the loudness debate (Loudness War)

[Compyfox]

This thread is intended as continued discussion for several debates that constantly pop up here on KVR, and also GearSlutz. Recently especially in here, which was a debate of "why" I don't go higher in terms of a certain loudness value.

I also wanted to see if this whole topic maybe evokes by itself, but I think I have to be more upfront in terms of presenting a certain idea, that currenty stirs up the engineering world as we speak.



Last year, at the end of August, especially Europe and a lot of selected countries all over the world adapted the so called "EBU R-128" loudness standard. In short, it's a standard intended for a unified loudness while broadcasting. This saves you from heavy loudness jumps while watching TV, or even at cinemas. Even though it is still happening to a certain extend.

The thing is however, that the music realm was almost left out of the context. At first, it is totally logical since this standard is aimed at broadcast environments, not engineers. But the Loudness War doesn't find an end, if only the broadcasts are normalized.


Example:
Take a recording from the 90ies with a huge dynamic range (or crest factor for that matter) and a loudness of (fictional) K-17. Like: Michael Jackson (the "HIStory" collection) or Dire Straits (Money for Nothing). Then run it through a normal broadcast chain (studio -> studio comp -> preparation station -> broadcast station). The endresult is a fairly normalised song with tons of transients that doesn't sound squashed or pumped.

Now in direct contrast, let's take a look at the "Day that Never Comes" song by Metallica, mixed and mastered in 2008, with a loudness of about K-5(!) ran through the same chain until it reaches the listeners ears. The endresult is an uber-loud, pumping and noticable distorted playback. Even with normalisation schemes. Why? Because the original song was too hot mastered to begin with. So even with Loudness Normalisation schemes, you loose out on the transients and in worst case, you get audible pumping due to the broadcast studio compressors.




Is there a Solution?

Currently there are no real schemes in motion to fight against that dillema. In my opinion, the war needs to be extinguished right at it's source - and that would mean while post production, and especially mastering. As usual, the opinions have become a habit, and debates about this topic are heated. Of course, and thankfully, there are certain people these days that think different on that behalf.

One prime exampe is Skip Burrows with his infamous post over at GearSlutz called "The Reason Most ITB mixes don't Sound as good as Analog mixes". He encourages to stick to gain staging schemes and not push everything to 0dBFS within the DAW or outboard gear.

Another example is Bob Katz, who recently switched from using his own K-System to using configured meters of the EBU R-128 standard. But as with the first introduction of the K-System, his ideas are still pushed aside as "utter nonsense".


I heartfully have to disagree.


Through the course of the last summer, and at the time the drama came up that "EBU R-128 will be the only defacto standard in the audio realm", I sat down for a couple of weeks to analyse the meter. And to find a way to use it for music post production and mastering as well. I had heated debates with Mister Katz and some of the developers of the EBU R-128 in the process, but it paid off. As with Slate Digital briding the gap towards understanding proper gain staging again, Mister Katz directed me to incorporating the K-System idea with the EBU R-128 meter standard.

In July 2012 I released a white paper that collects all these ideas, and bridges the gap between musicians and broadcasters. I call it K-System v2, which is silently accepted by Mister Katz (since it was originally his idea, but I developed it further).

You can find the white paper on my technical blog:
http://techblog.studio-compyfox.de/
(PDF, 144kb, 20 pages, duplex-print ready)


In short:
This white paper picks up the color code idea and the reference level shift of the K-System v1, but ports it over to the EBU R-128 "Momentary Loudness" meter.

The K-System v1 meters are at 600ms integration time, and like the VU (ANSI C16.5-1942, British Standard BS 6840, and IEC 60268-17 standard), it's non weighted. But their main flaw (just like the DR-Meters) were that they responded too strong on lowend content. However, the most dangerous content to our ears are the upper frequency ranges.

The k-weighting of the EBU R-128, or ITU-R BS.1770-2 standard for that matter, utilize a lowcut and highshelf filtering scheme prior to measurement. Due to this, the meters don't respond as strong to bass intensive material but more balanced (dare I say "compensated") and more "true" to our ears. The ballistics also shift back to 400ms rise/fall.

More about EBU R-128/ITU-R BS.1770-2 can be read at:
http://tech.ebu.ch/loudness/




But... this doesn't end there.

There will be certain engineers and "garage producers" again that think "meh, why do I need this?! It's not important". Wrong way of thinking in my opinion. We still have the problem with the broadcasting. But also (if not especially) while playback on home devices (CD, MP3 players, DVD players, etc). This is where the so called "Music Loudness Alliance" comes into play.


Who are they?

The alliance is formed of the very same people that originally worked on developing the EBU R-128 standard. With this alliance, they try to offer a solution for loudness normalisation while playback. Meaning: built in schemes in MP3 players that let you change the volume of your devices, but the overall loudness of all your tracks on the player stay the same.

There is a lot more involved (like: playback in loud environments), but this would clearly go beyond the scope of this discussion. Please educate yourself with the whole scheme over here:
http://music-loudness.com/




This brings me to certain comments I made in various posts all over KVR: K-16(v2)

K-16v2, or -16LUFS, is the planned average loudness for music for all kinds of players. Rumors have it that iTunes (software) already implemented loudness normalisation schemes to stay around that value. Though I can't confirm it yet. It came up around summer 2012 as well, while I had contact with the developers of the EBU R-128 standard, and the "Alliance".

To some people, this is a rather drastic drop from used to loudness values like RMS -7/-6/-5. Though think about it... The lower the loudness, the higher the dynamic range, the more "dynamic" a production sounds - and you don't torture your ears as much. You can still turn up the volume with your playback devices. Though granted, certain digital system have a limit or the DAC can't take a specific signal (see "Music Loudness Alliance" proposals). But clipping peaks is still better than clipping the whole average signal.

The original concept was to actually go down to -23LUFS, just like broadcasting. But this caused a major uproar in both the industry and musicians alike. Hence the introduction of the "Alliance". I heard from befriended engineers that this whole topic was also heatedly debated at the last AES (2nd half of 2012), and all participants agreed that there need to be subsequent works on the matter. And this is where my white paper chimes in yet again.


In my white paper, I offered a so called "timeline" to gradually reduce the loudness from K-12 and higher, down to K-16 as good balance between broadcasting mixes (read: movies documentaries, etc) and consumer mixes (read: music). Funny enough (and at that time unknown to me), this resolves around the same value that is rumored to be "final" with iTunes. And even more surprising, this is exactly the limitation of full dynamic EBU R-128 mixes.

Full dynamic EBU R-128 can have a loudness fluctuation of up to about +9LU. At least according to the official white papers of the standard (EBU R-128/ITU-R BS.1770-2). So "K-16v2 Dynamic" would mean, that the upper limit is actually -16LUFS, and the average signal can be anything between -23LUFS and -16LUFS, which is a drastic fluctuation (hence the "Integrated Loudness" with the +/- 1LU limit!). Though I'd only use K-16v2 dynamic for mixing broacast content (movies) and orchestra, not for popular music like rock/pop/electronic.




"But sir, K-16 is still too quiet".

Dear Sir and Madam - sorry, not at all. I redirect to the "Gain Staging" debates again. Yes, the ones where you actually use a reference level for both recording and mixing. And that might usually be -20dBFS (SMTPE), or more common -18dBFS (EBU) for loudness, and -9dBFS for digital peaks. (see: PPM on Wikipedia)

Agreed, ITB we can pretty much do what the heck we want, but if we use the best of both worlds, OTB and ITB, we can't go that route. And we shouldn't either. Especially with the recent swamping of analog type plugins. Which will only grow in numbers through the course of the next years (I'm sure of that).


Now let's assume we stick to the rules that the individual channels don't exceed -18dBFS (RMS, bass intensive material) or -9dBFS (transient heavy material) while peaking in. Then we mix to our hearts content, but don't exceed +1 VU (again, reference level -18dBFS) on the summing bus with fortissimo (read: very loud) passages of the song. What do we have? Exactly - a theoretical K-17/AZ-0 mix. But this doesn't end there.

K-16v2 has an average loudness limit of -16LUFS. But like the K-System v1, there is still the amber zone for fortissimo passages. And that can reach up to -14LUFS. Though unlike the K-System v1, I advise NOT to constantly stay there. It won't be possible either.


So, if you mix at -18dBFS/0VU, then master it for K-16v2, your average signal will be around -18LUFS to -16LUFS (mezzoforte passages), and peak up until -14LUFS (fortissimo passages) - though I won't kill you if you go to -13LUFS for a very short while (the Spagetti God will handle that). You have a (theoretical) "dynamic" loudness range of over 6dB, and your peaks can go haywire up until the allowed -1dBFS. The same works for K-14, only shifted by 2dB of course.

The endresult is a (moderate) loud production, no clipping. You don't even need a drastically setup brickwall limiter - if none at all. Best of both worlds. And, it's compatible to future planned loudness normalisation schemes. Not to mention that it's easy to print on every medium without further drastical editing: vinyl, tape, MP3, HD audio, etc.

Is it too quiet? Turn up the volume.




"I like your ideas sir, but what meters can I use? How do I need to setup my equipment?!".

Setting up your equipment is still a heated debate. The K-System uses a pink noise to calibrate your speakers to an output level of 86dB RMS (C-weighted). But to some, this is too loud. Doesn't hold you off from using it though, or calibrating the speakers to whatever loudness you feel most comfortable with (if setup loud).

Most important is the calibration of the meters however. They are your visual guides, your savety rope, telling you what's going on and what your upper limit is.


There are several tools on the market that can be fully customized to your needs. In the VST realm there is:

- Grimm Audio LevelView (apparently, even a S variant now, for 140EUR excl VAT)
- TC Electronic LM2 and LM6
- Nugen Audio VisLM (though the scale can't be shifted properly)

The first VST plugin that offers the K-System v2 presets "built in":
- ToneBoosters EBU Loudness/Compact (15 EUR, bundle)

With more and more developers interested in adapting the K-System v2 idea...

If you don't want to spend a fortune, or don't need 5.1 measurement, logging and all that jazz, I recommend to take a closer look at ToneBoosters for the time being.




SUMMARY:

We are not debating in whether or not a certain production "needs" to be loud in order to groove. This can be done with proper mixing. Neither do we debate if a mix must have a certain loudness in order to be "enjoyable" while out on the streets. For this, we either have volume knobs or at a certain forseeable future, the "Loudness Normalisation for Music on Playback".

We are debating in terms of how to fight against uber-loud productions in general. Or right at it's root - read: post production and mastering. What to do to steer back. Even though the K-System v2 is another interim solution, it is a proposal to drive back on the long run.

Something that is IMO simple and actually works.

You can stick to it and don't have to sweat if there is a big-bang in the audio realm regarding loudness schemes. Or you can ignore it and need to adapt from one day to anther. Nobody is forcing you to anything - currently.



But... think about it.

Let's assume(!!!) the "Playback Loudness Normalisation" by the Music Loudness Alliance will be put into place by Summer 2013 (assuming that the software providers play along, there is suitable hardware and the masses are covered). Now play your K-5 mastered song right after a K-14 mastered one. Even thuogh the normalisation process offeres a balanced listening experience, the K-5 mix will sound degraded.

So why not improve on this now rather than realising this problem through (possible) sales or negative feedback at a later state?



"BRAIN - it's the new mojo" (tm) - not how loud your productions are, or what gear you used in order to get there.






TL;DR:

The K-System v2 is an upgraded setup (color codes, reference level) to measure loudness with the EBU R-128 Momentary Loudness Meter instead of using unweighted meters at 600ms and reference levels of -20/-14/-12dBFS.

It is not aimed to be used while recording/mixing. For this I highly advice to use the VU (300ms/-18dBFS reference) and Digital Meter combo. Only use the K-System v2 meter for checking the mixdown, or while mastering.

If you are not into Post Production or Mastering, this does not(!) convern you.


Want to know more:
http://techblog.studio-compyfox.de/

Maybe also educate yourself with:
http://tech.ebu.ch/loudness/
http://music-loudness.com/

And/Or follow my KVRmarks:
http://www.kvraudio.com/kvrmarks.php?u=9761


Interesting YouTube material:
Loudness: War & Peace by Bob Katz, Nov 2011
TC Electronic Loudness and the EBU R128 Broadcast Standard by Florian Camerer, Jul 2011 (I also recommend "Bob Katz about Loudness War" and the "Rome Calling" video series with Thomas Lund
Loudness War: Dynamic vs. Massive compressed mixdowns



Thanks for reading and debating.



EDIT:
Added link, changed a paragraph for easier understanding.

[paterpeter]

Interesting read, thanks a lot. I'll go through your whitepaper as soon as I find the time.

Maybe the wrong place, but do you know what's the difference between the LU K16v2 and LU K16v2d presets in TB EBULoudness?

Edit: ah wait, it's for K-16 v2 dynamic as you mentioned it, right? The one with the larger light green area.

Edit 2: On page 8, you say "If you want to use the histogram as visual guide, the mezzo forte passages should not exceed +2 LU, ideally +1.5 LU."

That should be "fortissimo" in place of "mezzo forte", right?

[Compyfox]

Maybe the wrong place, but do you know what's the difference between the LU K16v2 and LU K16v2d presets in TB EBULoudness?

Of course, since I created them in the first place.


The main difference are the color codes.

The light green color zone for the K-16v2 preset is from -3 LU (-19 LUFS) to 0 LU (-16 LUFS). Suitable for most music content IMO, and close to what you're used to the old K-System - just a tad shifted.

The "light green" zone tells you where your music should(!) be at maximum for forte passages (in this case at/around 0LU). The 3 LU "yellow zone" gives you enough headroom for fortissimo passages. Instead of the 4dB "yellow zone" from K-System v1, you now have two 3dB zones (6 LU).


The light green color zone for the K-16v2 Dynamic preset should be (couldn't test it yet, not the time for it) from -7 LU (-23 LUFS) to 0 LU (-16 LUFS).

The "Dynamic" preset is intended for full dynamic content like movie/TV mixes, or even ultra dynamic orchestra that you want to clock in at K-16v2. Not to mention is this also intended for understand the difference between EBU R-128 as "broadcast metering standard" compared to music mixing.

Example again:
If you mix a movie for example for Youtube, and your average loudness hovers around the crossover point from dark green to light green (which is -7 LU or -23 LUFS), but the forte to fortissimo passages (including FX and music) don't go higher than the light-green zone of "K-16v2 Dynamic" (read 0 LU or -16 LUFS), then you should be(!) within specifications. The "yellow zone" is then the indication "wait... too loud!".



It's described in short on page 19 in my white paper.



Again, the whole K-System v2 idea is another interim solution to drive back the loudness. There is a lot in motion in this section currently and a lot can change. I just tried to find a middle way.

[Burillo]

small and possibly unrelated question - why this "loudness alliance" thing when there's ReplayGain already?

[Compyfox]

small and possibly unrelated question - why this "loudness alliance" thing when there's ReplayGain already?

Because replay gain can be evaded (read: turned off), changed and therefore faked. There is no set standard, or no way to "fix" this value. This is also a concern by the "Alliance" and their white paper.

To in order to get a consistent and enjoyable loudness throughout, an external/independent system that analyses the files beforehand, while ignoring ReplayGain settings or specific loudness, needs to be implements.


Another thing to debate. But I agree with the Alliance.

[paterpeter]

Edit 2: On page 8, you say "If you want to use the histogram as visual guide, the mezzo forte passages should not exceed +2 LU, ideally +1.5 LU."

That should be "fortissimo" in place of "mezzo forte", right?

Did you see this?

[Compyfox]

Sorry, I did not see the edits. But you're right - it should mean "fortissimo" passages if you measure "Short Term Loudness" with a histogram.

I actually confirm this a bit further down with the short summary on the very same page in the PDF.


I really hope that certain tools will be able to write histograms for the Momentary Loudness as well/instead of the Short Term Loudness. It was actually possible with the pre 2.6 versions of EBU Loudness. But that feature is (again) on a larger FR list for a possible v3.

[kylen]

Compyfox - congrats on some very nice work trying to affect, leverage & drive loudness standards for music recordings.

I myself appreciate the need for loudness for several reasons that come up often in my world. 1) Playing a song in 'rotation' among other songs; 2) Comparing demo samples in on-line forums; 3) Inserting commercial reference material into my mastering project - sort of flushes the ears and gets me in the right ballpark.

I'm sure other folks have similar reasons for appreciating loudness vs volume discussions. Someone in the other thread mentioned the 'music while driving in a car' vs noise syndrome.

Here's a question for you:
The various 'systems', whitepapers and books that contribute to the current state of the whole loudness discussion mention crest. What standard or refence do you use that describes crest, dynamic range, etc. in your white papers?
IOW what do you take crest to mean.

I know how I've been using crest (dynamic range) and Bob Katz does mention leaving a limiter at -0.3dbm on the master to catch any possible peaks. But he doesn't really say that K-14 assumes a crest of -14dbm rms [lower] to 0dbFS (or -0.3dbm] [upper] or something like that. Anyway I can't find it. You know why I'm asking - what makes it mandatory to hi the upper side of the crest, what are your thoughts on this top end of the range - regarding the standards you're participating in?

[Compyfox]

Mister Katz mentioned in his documents on Digital Domain (and his book) that he reconfigured a Dorrough meter to his needs. This one is built upon the standards of VU's ballistics and reference level (ANSI C16.5-1942, British Standard BS 6840, and IEC 60268-17). Meaning: if you use a reference level of EBU (in this case -18dBFS), we talk about 0VU, which should readout 1,23V on output if a 1kHz signal is applied.

A bit complicated to describe here, but you let's simplify it and say that Dorrough meters can be setup to any reference level you like, while the ballistics can be 300ms (VU) or even 600ms (K-System v1). At least the hardware can be modified, I could only chance the reference level with the Waves plugin. Then again, I didn't go more in depth.


But to answer your question (and to my understanding):
A crest factor is the dynamic range from loudest signal (digital full scale) to average signal (RMS).

So if you (for example) have a signal that has an average signal level of -13dB RMS and the maximum digital peak is -0,5dBFS, then we have a crest factor of 12,5dB.



Now let's take a look at the K-System v1

K-14 has a reference level of -14dBFS (which would mean 0 VU at 600ms), and Mister Kats recommends not going higher than -0,3dBFS in terms of digital peak in order to have enough headroom for the DAC to not barf. At least to my understanding.

The K-System v2 is however built upon the specifications of the EBU R-128 or ITU-R BS.1770-2 standard. These specifications also say that the loudest digital peak should never exceed -1dB True Peak. IMO a good decision for all kinds of media to be save from overloads - especially old DAC's and certain MPEG audio encodings.

So if you use K-16v2, you have an effective dynamic range of 15dB. With K-14v2 it's 13dB, and with K-12 it's 11dB.


But even with a crest factor, or dynamic range for that matter, of 11dB and using K-12v2 (average level between -15LUFS to -12LUFS, maximum peak at -1dBTP), you barely need any drastic limiting. You literally cut off peaks only.

With K-14 already (either v1 or v2), changes are that you barely even reach -1dBFS digital peak. Unless you have rouge peaks in your production.

[paterpeter]

Did I get it right that v2 doesn't change anything about the way you should calibrate your monitors? I currently follow the standard procedure:
1. start with K20:
- -20dbFS pink noise playing through each speaker individually
- calibrate speaker so that this matches 83db c-weighted/slow (or a different value depending on hearing preference*)
- sums to 86db for the monitor pair
2. for K14, reduce monitor gain by 6 db (so creating a mix at 83db will require -14dbFS)

That is, for K-16 v2 I can still use this calibration method but set the monitor gain to -4db, right?

* Currently I typically calibrate using 77db per speaker as the reference level. When using K14 (= monitor controller set to -6db) I can crank it up to 83db from time to time to hear what it sounds like at that level.
I've got to say that it's great to know the physical loudness I'm working with. Before I never knew if I had my levels too loud for safe listening.

[kylen]

calibrate speaker so that this matches 83db c-weighted/slow (or a different value depending on hearing preference*)

I had to bag the 83dBSPL-C a while back, I can't focus my 'mastering' listening ears for very long at that level - they start to crush. Mind you a rock concert at that level is no problem, wierd I guess...

EDIT: I should point out that my level of 'mastering' occurs on nearfields, ASP-8's so now we have that variable to deal with also in regards to dbspl that is.

[kylen]

But to answer your question (and to my understanding):
A crest factor is the dynamic range from loudest signal (digital full scale) to average signal (RMS).

So if you (for example) have a signal that has an average signal level of -13dB RMS and the maximum digital peak is -0,5dBFS, then we have a crest factor of 12,5dB.

Yes, that's my understanding also.

Now let's take a look at the K-System v1

K-14 has a reference level of -14dBFS (which would mean 0 VU at 600ms), and Mister Kats recommends not going higher than -0,3dBFS in terms of digital peak in order to have enough headroom for the DAC to not barf. At least to my understanding.

Yes, I'm good here. Nice not to clip the DAC if possible. This also matches the metering in my DAW so I can tell where K-14 is (-14dbmFS rms) in the crest and also where the top most limit of my crest is (max -0.3dbm - because my mastering limiter brickwalls at -0.3db). Note: I just barely touch my mastering limiter, every few bars in a song are allowed to push into it about 0.5db - nothing continuous in other words.

Now here's the ambiguity (in my mind), in my world I hit K-14 just fine, but I also make sure that I hit the top of the crest regularly so that my K-14 dynamic range is full. rms -14db up to -0.3dbFS, a dynamic range or crest of 13.7db or thereabouts over the course of the entire song (program). The claim of 13.7db crest is then based on the premise of touching digital full scale (altered to -0.3db per the suggested 'spec', standard or best practice).

But I don't believe anything in the standards is requiring me to hit digital full scale (or -0.3dbFS). For example if I setup my gain and plugins to K-14 but the crest only hits -6dbFS then do I really have K-14. My thought is yes of course, as long as the program stays in my DAW. As soon as the next guy gets it and normalizes it to -0.3dbFS now my K-14 program just went up to about K-8. Because the crest or dynamic range of my program was always 8db.

So having blurted all that out I guess what I'm wondering is if there is anything in the spec that says K-14 MUST have a crest of 14db?

I need to reread your doc and my Bob Katz book and your true peak paper now that I've rephrased my question. A nit pick I guess but kind of has been on my mind ever since I tried to stop squashing stuff.

[kylen]

So having blurted all that out I guess what I'm wondering is if there is anything in the spec that says K-14 MUST have a crest of 14db?

Ahhh - here we go, Mastering Audio Chap 15 page 194 [BKatz]:
"...pop music with a crest factor of less than 14 db should not be mastered to peak to full scale as it will sound too loud..."

This is a suggestion in a Bob Katz book, not a standard enforced by any metering or compliance that I know of. I'll bet there are meters that can plot crest factor though...hmmm what say you Compyfox?

EDIT: Ok I might try this one, Compyfox I think you know about it from last year: http://www.pleasurizemusic.com/
You've got some others listed up there too...need to check.

We already talked about this one:
http://www.toneboosters.com/tb-ebuloudness/
I think it might haves just what I need - I didn't see that real-time analysis window - seems to have a histogram, true peak, loudness & crest factor(?) ...checking it out.

Well OK - looks like good old Voxengo SPAN gives me a good indication of crest factor along with K-scales: http://www.voxengo.com/product/span/
I wasn't sure the crest factor was right but it looks ok now...

[Compyfox]

Did I get it right that v2 doesn't change anything about the way you should calibrate your monitors?

Yes, you are correct on the callibration part. If you want to go for it.

But as you also pointed out, 86dB c-weighted can be tiresome for the ears. This is still a thing to debate, and IIRC the original 86dB was not necessarily meant for most studios. At least not small studios.

A more healthy level might indeed be 77dB c-weighted. Though whatever level you feel most comfortable with works. The meters are still your visual guide, ears can be fooled in terms of loudness.

But I don't believe anything in the standards is requiring me to hit digital full scale (or -0.3dbFS). For example if I setup my gain and plugins to K-14 but the crest only hits -6dbFS then do I really have K-14. My thought is yes of course, as long as the program stays in my DAW.

You are correct on the thought that you don't need to constantly have a crest factor of 13+ dB with K-14. No matter if within your DAW, or outside of it. It wouldn't even be possible most of the time (at least with current modern recording/mixes).

As soon as the next guy gets it and normalizes it to -0.3dbFS now my K-14 program just went up to about K-8. Because the crest or dynamic range of my program was always 8db.

This is where the Loudness Normalisation schemes for playback would do the trick. The poposal that the "Music Loudness Alliance" is currently making focuses on that issue.


Instead of applying replay gain, or peak normalisation, the system does "loudness normalisation" to have a consistent loudness over the whole playback list. It ignores the DR values.

This is why I earlier wrote: "Imagine a song from the 90ies, played side by side with a song from mid 2000s, however both share the same loudness of K-14 for example."

What would happen in this case, is that the loudness would be the same overall. But the missing transient peaks from the squarewave production result in a softer, more dull audio representation.

As example, another YT video from (now) six years ago that clearly shows this effect.

http://youtu.be/3Gmex_4hreQ

So having blurted all that out I guess what I'm wondering is if there is anything in the spec that says K-14 MUST have a crest of 14db?

As you found out yourself already, there isn't. The main debate in the end is "peak normalisation" vs. "loudness normalisation" on playback. Else you're well within specs.

This is a suggestion in a Bob Katz book, not a standard enforced by any metering or compliance that I know of. I'll bet there are meters that can plot crest factor though...hmmm what say you Compyfox?

Plenty around by now.


The one by "PleasurizeMusic" or better said the "Pleasurize Music Foundation" (Friedemann Tischmeyer) was the first mass-public aware Dynamic Range Meter, coded by Brainworx. In an earlier thread of mine (see my KVRmarks), I analysed what this thing is about.

The thing is that this plugin asks for annual fees. It's successor with a tad more features and different weighting, is the Brainworx bx_meter. So far, these two incarnations are the most simplest visual representations of the Dynamic Range of a song in real time. I only miss an update off the offline DR-Meter with Brainworx. Currently, that one is only available through PleasurizeMusic and is locked to 44kHz.


EDIT:
Just checked EBU Loudness with the new PLR (Peak-to-Loudness) meter. This one is not real time fluctuating per se, but going by "maximum lowest dynamic range ratio". Fine with me, it's just an additional indication anyway.

Overall, this meter is a very fair priced (dare I even say "too fair") all-in-one solution for all kinds of measuring needs. Need an entry level EBU R-128/K-System v2/True Peak and DR-Meter with 2.0/5.1 measurement capabilities, not to mention being cross-platform compatible? Spend 15 EUR and you're good to go.


Voxengo SPAN has the values showing somewhere below the RMS values. But I never liked SPAN to be honest.



Personally I don't think that the DR range is important anymore. It can even fluctuate drastically from production to production. A rock mix for example could have a DR of 5-8 even if the loudness is K-14. Electronic productions are similar. Now imagine jazz productions, or orchestra (highest I got with an orchestra recording from end 90ies was DR-13!).

There was never a set "standard" on what to shoot out for. And the one that once floated around "you have to have a DR of 14+" is in reality mostly not even possible. At least not with music.

I just see the DR or "crest factor" value as another information of where the signal is currently at. But compared to the EBU R-128 measurement, it's non-saying.




Oh and... could we maybe clear what you mean by dBm, kylen?

To me, dBm means " power ratio in decibels (dB) of the measured power referenced to one milliwatt (mW)". But if I talk hardware, I usually go by dBu or dBV.

Then again, let's not make this more complicated than it already is. Stay within the DAW, so we only have dBFS, dB RMS, LU (relative loudness units), LUFS (Loudness Unit Full Scale, like dBFS), etc.



EDIT:
Fixed broken link

[kylen]

But I don't believe anything in the standards is requiring me to hit digital full scale (or -0.3dbFS). For example if I setup my gain and plugins to K-14 but the crest only hits -6dbFS then do I really have K-14. My thought is yes of course, as long as the program stays in my DAW.

You are correct on the thought that you don't need to constantly have a crest factor of 13+ dB with K-14. No matter if within your DAW, or outside of it. It wouldn't even be possible most of the time.

I'm not talking about "...constantly have a crest factor of 13+ dB...".

Here's what I said [...nothing continuous...]:
"where the top most limit of my crest is (max -0.3dbm - because my mastering limiter brickwalls at -0.3db). Note: I just barely touch my mastering limiter, every few bars in a song are allowed to push into it about 0.5db - nothing continuous in other words."

Here's where I have a problem with all of this:

As soon as the next guy gets it and normalizes it to -0.3dbFS now my K-14 program just went up to about K-8. Because the crest or dynamic range of my program was always 8db.

This is where the Loudness Normalisation schemes for playback would do the trick. The poposal that the "Music Loudness Alliance" is currently making focuses on that issue.

Instead of applying replay gain, or peak normalisation, the system does "loudness normalisation" to have a consistent loudness over the whole playback list. It ignores the DR values.

This is why I earlier wrote: "Imagine a song from the 90ies, played side by side with a song from mid 2000s, however both share the same loudness of K-14 for example."

What would happen in this case, is that the loudness would be the same overall. But the missing transient peaks from the squarewave production result in a softer, more dull audio representation.

This is a judgement about "Loudness" without any clear criteria in the system about "Dynamic Range". It allows a program with a Dynamic Range of 6dB to be "Loudness normalized" against a program that has a Dynamic Range of 12dB.
Now imagine they're the same program - just produced at 2 different dynamic ranges. I'm not sure how this has fixed the Loudness Wars which I thought were about Dynamic Range and over-squished music. It appears that Loudness Normalization still allows me to squish my DR to 6dB as long as the Loudness Normalization algorithm can turn down the volume (normalize) to K-14 for example.

So having blurted all that out I guess what I'm wondering is if there is anything in the spec that says K-14 MUST have a crest of 14db?

As you found out yourself already, there isn't. The main debate in the end is "peak normalisation" vs. "loudness normalisation" on playback. Else you're well within specs.

I'd like to suggest that "Dynamic Range" be added to the debate.