EBU R 128-2011 and questions

[davidka]

Hi. I've just seen the upcoming plugin of Steinberg, and that raised up some questions about this new loudness normalisation standard.

So, basically, will this mean that broadcasters who adopt this new standard will have a consistent, yet dynamic loudness level?

I can speak only for Hungary, but here, TV commercials have at least 1.2x the loudness of the actual programme. This makes them annoying, and scary as sh*t when they pop up, especially at night. Will they fix this issue?

Also: Am I wrong, if I hope that it may increase the abundance of truly dynamic music in the future?

[BertKoor]

I've just seen the upcoming plugin of Steinberg

Not longer upcoming, it's published:
http://www.steinberg.net/en/products/nu ... meter.html

will this mean that broadcasters who adopt this new standard will have a consistent, yet dynamic loudness level?

The EBU R128 recommendation was incepted in 2010 and last revised in Augus 2011. So not really something new... Adaptation of the norm by the market is a different story, things don't go that fast.

TV commercials have at least 1.2x the loudness of the actual programme. This makes them annoying, and scary as sh*t when they pop up, especially at night. Will they fix this issue?

If all broadcasted material is within the EBU norms, then the issue with louder commercials will be fixed indeed.

Also: Am I wrong, if I hope that it may increase the abundance of truly dynamic music in the future?

This norm is for TV broadcasting only. It will be quite a shock for producers and customers when applied to audio CDs and distributed MP3 files, since content will be so much softer (around 10dB at least I think) than what we are used to in the last decade. It's like stepping back to the first days of the compact disk.

[BDeep]

My knowledge on the subject is far from complete, but I'll tell you what I know. The rec128 measures the volume of programme material over the entire length of the programme, and calculates one number for the entire file. This is opposed to the current method where peak is measured and a programme may not overshoot that peak at any given time.
This allows for more dynamic mixing of material and should equal the perceived loudness across, say, a documentary and a commercial block.
Most European commercial and public broadcasters will comply with this new standard, although I'm not sure if all of them will, and in what timeframe (also, no idea on the situation in Hungary).
In theory, radio stations should also comply with the new standard. When that will happen one day, it wouldn't surprise me if new albums will be mastered less loud than currently, as there is no benefit to do so anymore. However, with radio stations the situation is a bit more complex when compared to television, as they compress their signals quite heavily as to obtain a sound coloration of their own. Also, I believe for transmitting through the air, signals preferably shouldn't be too weak.

[BertKoor]

The rec128 measures the volume of programme material over the entire length of the programme, and calculates one number for the entire file. This is opposed to the current method where peak is measured and a programme may not overshoot that peak at any given time.

It's a bit more complex than that.

EBU R128 and Techreports 3341/3342 specify how "loudness" should be measured, and what the maximum and minimum values should be for broadcast material. There's a momentary loudness, short-term loudness, and integrated loudness. Momentary is based on the last 400 ms, short-term is based on 3 seconds, and integrated loudness for the whole programme with 2-fold gating applied. For each of them there are different norms, but the integrated loudness is the most important one. It's interesting to note that the absolute (inter-sample) peak value should be below -1.0 dB FS.

[Compyfox]

There is another thing as well that is bothering me:
According to Bob Katz, he recommended a different calculation with focus on the lowend.

His K-System recommendation is built upon a calibration of the reference level and a 600ms measurement time. How does this port over to the EBU (ITU) measurement? I don't think that they apply certain filters like the weighting filters (also see here) on DR (Crest Factor) meters like the bx_meter.

Or is it really that simple?


Else I wholeheartly agree with you - ITU 1770 / EBU R-128 is not really made for production. But broadcast wise it makes more sense to turn down the output gain. And this can be done at the three stages of the send stages (studio, preparation, broadcast module).

[davidka]

Thanks for the quick answers, much appreciated!

Maybe Bob had fewer friends among the decisive folks. I can speak only for IT tech standards: whiches were much more about the we game call 'monopoly' than technical improvement.



About Hungary in short: Our government is gathering money everywhere. Partly to get us out from the IMF threat, but mostly to get more money for themselves.

They created a Censorship Office to overwatch all the media content. Including Internet. I don't want to go into details as they may see it.

The only good thing came out of it is that they measured signal levels and are preparing to charge all the stations where the loudness acts like I've described above. I suppose it goes without saying that our stations will happily adopt R128.

[Compyfox]

Maybe Bob had fewer friends among the decisive folks. I can speak only for IT tech standards: whiches were much more about the we game call 'monopoly' than technical improvement.

Don't get me wrong here, Mr Katz not only started a certain revolution with his K-System, he also focuses more on R-128 now.

But both systems are flawed. The K-System (according to his comments) since it's not accourate enough, and the ITU-1770-2 system, since it uses a boost in the upper frequencies prior to measuring.


In the end it's a measurement tool debate yet again.
For music production, it's always the digital meter vs. one of the quasiPPM's
For loudness it's VU vs. K-System vs Dynamic Range vs. EBU R-128.


Personally I'm no broadcaster, I am an engineer. And here I stick to three meters absolute max:
- Digital Peak metering and loudness metering (VU, 300ms rise/fall) in terms of recording and mixing
- Digital Peak Metering and K-System metering while mastering


The rest is just bonus and aimed at different end applications. They(!!!) should be cleared first, before any firm comes along and says "this is the only standard - work in that one from now on".

[Dean Aka Nekro]

Maybe Bob had fewer friends among the decisive folks. I can speak only for IT tech standards: whiches were much more about the we game call 'monopoly' than technical improvement.

Don't get me wrong here, Mr Katz not only started a certain revolution with his K-System, he also focuses more on R-128 now.

But both systems are flawed. The K-System (according to his comments) since it's not accourate enough, and the ITU-1770-2 system, since it uses a boost in the upper frequencies prior to measuring.


In the end it's a measurement tool debate yet again.
For music production, it's always the digital meter vs. one of the quasiPPM's
For loudness it's VU vs. K-System vs Dynamic Range vs. EBU R-128.


Personally I'm no broadcaster, I am an engineer. And here I stick to three meters absolute max:
- Digital Peak metering and loudness metering (VU, 300ms rise/fall) in terms of recording and mixing
- Digital Peak Metering and K-System metering while mastering


The rest is just bonus and aimed at different end applications. They(!!!) should be cleared first, before any firm comes along and says "this is the only standard - work in that one from now on".

Unfortunately that is exactly what TC Electronic are encouraging us to do Compy man, Well that is how every presentation of thier new(ish) meter comes accross and of course they will do this as they no doubt poored alot of time into it and thats money. Wonder if I can get a free copy if I have proof of purchase of the old Finalizer they sold me but yeah it has me at the same time. Just agree to a sensible, decent standard for audio/music we can all use as reference = problem solved, It won't happen though it never does!

[Compyfox]

Thing is, EBU R-128 (or ITU 1770-2 for that matter) is just too soft on the long run. I works for broadcast (since the measurement is on a long-term basis), but not for music - unless it's squashed to sh't.


The K-System was a great recommendation - suddenly all are high about EBU R-128. I think I've said it before.

We (engineers) are engineers, not broadcasters. Two different workbranches.

[Compyfox]

I read myself further into the topic, still a lot to learn it seems.

The EBU R-128 is a longterm metering analysis tool (see Leq) with a two stage input filter (LR2LB) while the K-System Meter is basically a non-filtered (flat) RMS meter with adjusted referene level and sample time (sometimes called "gate"). The first stage of the RLB (or the revised L2LB for that matter) filter is aimed at the Fletcher-Munson-Courve at about 100phon (about 100dBSPL) and should resemble a similar courve to how our ears respond to high volume content.


I did tests yesterday with the official ITU-1770-2 files, ToneBooster's TB EBULoudness, Brainworx bx_meter (K-System) and Klanghelm's VUMT (set up to the K-System).

The testfile I used was a 20s/-20dB sine switching to additional 20s/-30dB sine. I got readout differentces of over 8dB from ITU-1770-2 (EBU R-128 with just a different scale) to the K-System, depending on the K-System mode of course.




Reason:
The EBU Meter uses two filters pre-RMS measurement per channel. The first stage is a highshelf at about 1kHz with a 2nd order IIR and +4dB boost, the second stage is a 2nd Order (12dB/Oct) LowCut with -3dB "knee" at about 60Hz. On top of using momentary measurement of only 400ms rather than the known K-System's 600ms. Additional to that, we have the short-term (3 seconds) and Integrated (longerm) measurements (start to end).

According to Mister Katz, this (the input filter) is more suitable for RMS measurement since his meter works woth a flat response (broadband) and the setup/calibration depends on C-Weighted SPL (LeqC) meters.




Conclusion:
I need to do further tests if a filter like recommended in the ITU-1770-2 papers are really what's helping getting a more healthy level and more detailed readouts for that matter. But I have a cuple of doubts on the long run:

1) the EBU Meters seem to be calibrated to -23dBFS (-5VU at 18dBFS reference), this is even lower than consoles and analog gear calibrated at -18dBFS (0VU). In a pure digital realm, this gives you a lot of headroom, but if you incorporate analog equipment, you can run into trouble. It is also somewhat obsolete if you level in your signal with both a digital meter and a VU.

2) -23dBFS (0LU) compared to K-14 (-14dBFS) is a 9dB difference. It makes sense for broadcasting, but for producers and mastering engineers it's a bit overkil, even for those that use K-20 already.

3) The K-System offers a level fluctuation of +4dB for mezzoforte passages. It is advised to use that area rarely - yet it can be used to "get out more". ITU/EBU is less forgiving +/-1dB fluctuation is allowed on short-term measurements (which can be okay, considering the 3s timeframe).

4) Tools that do longtime measurements and offer a multichart in the end are expensive. As of this moment, only TC and Nugen seem to offer that. The EBULoudness by ToneBoosters only shows and offers ca 40s. Most other tools are just compact stereo measurement tools and don't give you any usable info on longterm measurements

5) The EBU measurement is not really made for live applications, and for mastering it's a bit counter intuitive. In order to know what fluctuations your production has, you need to run it through the measurement tool in real time. Take a CD production with 74min for example - you need to run through it at least once(!) - and depending on the tool you used, you can then carefully adjust your files. But since only the big players offer longterm charts or adjustment tools, it's a PITA really.

6) the only good thing with the EBU meter is the "TruePeak" readout. Then again I did some readups on that as well. If I got it right, TruePeak is basically an oversampled digital peak measurement to rule out every overshots and rounding errors. Then again, the maximum recommended limit is -1dBFS (digital). A value you will barely reach with K-14, and almost never with K-20.




SUMMARY:
Is the K-System dead and will EBU R-128 be the ultimate solution?

Yes and no. The K-System does seem to need another configuration since Mister Katz changed his mind and I somewhat agree with him that both DR Meters and K-System Meters respond too heavily on lowend energy (which is less dangerous than mid/high frequency energy). The LRLB filter would take care of that, but I need further tests to see if the recommended loudness limits would go lower, or higher again. Brainworx made the first step for it's DR-Meter and offered LeqA and LeqK weighting, but only for the crest factor measurement.

I think we're on the right track regardless. Most consumed are still the radio and TV, and here the EBU R-128 recommendation seems to do the trick to finally offer a more consistent level between program material and commercials - especially if broadcast through digital means. Even though with recent tricks to evade that (run a show, then show one commercial, start the full commercial batch, run the show, show one commercial only, show another episode of the same show), it's a bit hard to control since these "oneshot" commercials are usually shorter than 20seconds. I do have my doubts regarding cinema content however, especially with the +/- 1dB fluctuation limit (short-term basis).

Personally I still say the K-System is a not to underestimate tool. It's simple to understand, it offers three steps to go back to the "better days of making audio content" and it's finally on everyone's lips.




IS THE PANIC-MAKING (by Steinberg especially) ADJUSTED?
Funk no! Get out of the house more!

No seriously - unless you're a broadcaster (technician), trailer cutter or DVD/BD editor, stick to different means (a plain VU, K-System meter, careful gain staging). If you're a musician only, all of this doesn't concern you. If you're an engineer however, broaden your mind. You might not need it now, but you'll never know when this is the case.

[insonicbloom]

I read myself further into the topic, still a lot to learn it seems.
I did tests yesterday with the official ITU-1770-2 files, ToneBooster's TB EBULoudness, Brainworx bx_meter (K-System) and Klanghelm's VUMT (set up to the K-System).

The testfile I used was a 20s/-20dB sine switching to additional 20s/-30dB sine. I got readout differentces of over 8dB from ITU-1770-2 (EBU R-128 with just a different scale) to the K-System, depending on the K-System mode of course.

why don't you actually record some music instead of "doing tests"
bollocks to the "K-System"
bollocks to the "loudness war"
bollocks to being told what my mixes should comply to.
and most of all Bollocks to Bob-I-Haven't-Ever-Mastered-anything-decent-Katz

try mixing and mastering with your ears not your eyes and you might not have these problems.
hell, try actually mixing and mastering something!

[knockman]

what we really need is a "Don't get me started" forum.

[manducator]

So, I think I do understand what the EBU R128 is about. It's a broadcast standard.

But does it mean we have to mix and master our tracks not going over -23 LUFS? If I would upload my songs that low in volume, I will get many complaints.

So how should hobbyists like me deal with this?

[Compyfox]

This is a revival of a 3 year old thread - a lot has changed since then.

So, I think I do understand what the EBU R128 is about. It's a broadcast standard.

It's not a "standard" per se, it's a recommendation. A lot of broadcast studios adapted this in silent agreement.

But does it mean we have to mix and master our tracks not going over -23 LUFS? If I would upload my songs that low in volume, I will get many complaints.

So how should hobbyists like me deal with this?

Two possibilities:
a) you don't and stick to the old VU/Digital meter (work around -18dBFS) prior to giving your track away to mastering

b) you can use the EBU R-128 meter as "more advanced meter" - in this case, use a custom reference level (i.e. -16LUFS) and work towards this.


If you take a dive into my KVR Marks, I've wrote a couple of posts on this topic up until this point.

[manducator]

Thank you very much, compyfox! I will check out the links.