Spot on!
And Timo: Great news!
Bye / Tumppi
It’s not really “completely different”. Apart from the delivery requirement metering using various algorithms are tools and tools only. I use the tools you use in my way. I’m just a bit fed up with people telling others how to use their tools. There’s only one correct way to use any tool: The way to get the job done conveniently, accurately and if possible enjoyably… All else is just reactionary nonsense…
I’m not following you here …
When you need to deliver content according specific spes, then there is no “alternative” way to use the tools.
R128 is a broadcast spec, so anyone delivering content needs to make sure that his “master” is 100% within those specs.
So R128 implies that the audio content is delivered @ -23 LUFS, so their is no other way to use the tools (=R128 or LUFS metering) so they reflect exactly what the delivery requirements are.
Apparently there’s something I am missing from your statement, but how do you want to use your tools -in this case LUFS metering- in a way that suits your “free interpretation of the standard”, while still being correct to the delivery standards.
(Don’t get me wrong, I am not looking for an argument -please not-, I’m just trying to understand what you are saying.)
Fredo
Well, for example:
Say that the specification says an average level of -24dB LKFS, +/-2dB, over a certain period of time.
Say now that you have a metering plugin that measures exactly that.
You could mix this “to your ears” in your calibrated room that you know, and get a mix that you’re happy with. THEN you go and evaluate your file, see where the average is, and adjust via an offline gain-change (that thus affects the entire file equally).
The “tool”, in this case the meter, is used by you retroactively, to see what you mixed.
I can reach exactly the same equally “legal” result by watching this meter as I mix. The result is the same, and average level of -24dB LKFS.
Thus, the tool we both need is used differently: Realtime versus Offline.
It’s actually very simple.
R128 is a delivery spec. If your content is not delivered correctly, the distributor will “normalize” it to the spec.
Older -brickwall limited- content will sound not near as good as content mixed without limiting or heavy compression.
Though, this content will also be “leveled” according R128 before broadcast.
Once the specs are enforced by the broadcast station, everyone who delivers content will have to make sure that the content is according specs. (R128). If not, the broadcaster will “normalize” it so it fits the specs.
The good thing about R128 is that there is a loudness measuring algo build in, which fits more or less the Fletcher-Manson curve and which produces a fairly objective measurement of loudness.
In the end it is comparable with mixing for cinema. You have tons of headroom (around 20dB if I am not mistaking), which means that these gunshots and explosions can be loud, while dialog can stay at a reasonable, comfortable listening level. So as soon as you start to mix by ear (dialog on a comfortable level), then everything falls in place. And even if your mix is a few LUFS below or above specs, after normalizing (which would be the same as adjusting your monitoring levels) your mix will be -exactly the same as before, though over the entire length of the program material a tad louder or quiter. In other words, this is a delivery spec which encourages everyone to make realistic sounding mixes and penalizes everyone who search for “being louder than the competition”.
From what I have heard, heavely compressed and limited commercials end up about as loud as non-limited mixes; however the inherent “audio quality” of the heavely limited commercials becomes clearly sub-standard. As it has always been … except for those who think that louder is better.
Fredo
And?
You asked me to show how the tools we use can be used different ways to reach our desired goals. I did.
So?
While you were posting this one, I was answering to the poster above you (Brede); just bad timing. We posted at the same time. I edited my previous post so it is more clear. Sorry 'bout that.
In answer to your questions, see below.
Say that the specification says an average level of -24dB LKFS, +/-2dB, over a certain period of time.
The spec is -23 LUFS over the total length of the program material.
You could mix this “to your ears” in your calibrated room that you know, and get a mix that you’re happy with. THEN you go and evaluate your file, see where the average is, and adjust via an offline gain-change (that thus affects the entire file equally).
Yeah, that is correct.
I can reach exactly the same equally “legal” result by watching this meter as I mix. The result is the same, and average level of […].
Not saying that you can’t, but there are a couple things that need to be said here.
As long as you are not far off the spec, you are correct.
However, when your mix is far off the specs, it will sound vey differently.
Since there is a loudness measurement going on, program material that is mixed “loud” will have much less bass than program material that is mixed on “reasonable” levels. The louder you mix, the less dynamic range you will “allow” your mic to have, etc … The only way to mix “correctly” according the specs is by mixing at the same level as the specs intentions. You can mix “the old” or any other way, no problem … only, you risk that your mix is not percieved as you have intended it to be percieved.
Bottom line is that the level you mix on has a great influence on how your mix sounds; that is something we all know, but very few adapt in real life. Except for a few people who work according K-Metering (Bob Katz) and those in Post (calibrated rooms at 85, 83 or 79 dB), it is pretty hard to understand how this will work.
Again, if you are only a few LUFS (=NOT dB!!!) off, your are very correct, then the meter doesn’t matter too much.
Thus, the tool we both need is used differently: Realtime versus Offline.
There, I think is the misunderstanding.
R128 measurement is measured over the complete program length, so if you want to have it technically correct, there is no such thing as realtime measurement. The correct numbers and readings can only be shown at the end of the program material, which is more or less the same as offline.
Fredo
Ah, that explains it….
I was using LKFS instead of LUFS simply because I was looking at the Dolby Media Meter and that’s what popped into my head. It does however not change anything.
Well “duh” monitoring level has an influence on how we mix. That’s obvious. But I really don’t see the problem here.
- “only a few LUFS off”… well, you’re either within spec or not, so I don’t understand the point.
- So what if stuff that’s mixed “loud” has less bass, does the spec define that?
two other points
-
I work to satisfy my clients, not technicalities. I only follow specs because if I didn’t, my clients would be unsatisfied. So, because of this, when I’m working on an ad campaign or series I want to deliver all at the same loudness. It has nothing to do with the final specs and everything to do with them not coming back to me with “why does this sound softer” type comments. Like I said, I think the calm act allows for +/- 2LKFS, which is a fairly wide range. If they pop in one spot that’s averaging -22 followed by one at -26 they’re going to be annoyed.
-
Which leads to the fact that we aren’t robots and our ears tire. Not everything I do (unfortunately) is “artistic”. I’m a technician a lot of the time. Thus when my ears get tired my levels and dynamic range change. Meters help me in these cases. There’s not a human in the universe who will care if I mix with my Dolby Meter running or if I mix to my ears if it’s a crappy ad for some company everybody hates.
It’s no misunderstanding: You adjust after the fact because you mix solely “by ear”, and I chose to use the meters to get it where I want it from pretty much the beginning. I don’t need to “normalize” (ignoring also what this does to absolute peaks, yes?). My final mix is as desired and within specs.
Clients happy.
I’m paid.
I’m happy.
It’s interesting to see that everyone is talking about LU and nobody seems to mention the LRA parameter. On page 18 of tech doc 3343 the EBU goes into detail about recommended loudness ranges.
For the metering part of Nuendo it would be GREAT if there would be a nice visual feedback of the programmes loudness range.
Ollie
So this is what I understood from EBU specs: -23LUFS (Gated - Europa)/-24LKFS(UnGated - America) or 0LU/0LK (which is the same) must be the integrated value (which is the avarage of the hole program). So I do not think that you have to keep your programe at 0LU/0LK in every moment. Otherwise you would lose the dinamic which is the hole purpuse of the new standard. This is why you should calibrate your monitoring to 0LU/0LK (filtered pink noise) = 79/82/85dB SPL. After that you just have to mix by ear. In the end, if your integrated value of LUFS/LKFS is not -23/24, than you have to normalize (offline) or change the gain before the loudness meter and do a realtime recheck.
As for the dinamic range, LRA, EBU specs stated a maximul of 25LUFS/LUKS recomanded. This is for most dinamic cinema content. For TV i do not know but I think is much smaller. The allowence of +/-2dBLUFS/LKFS (also for the integrated value, not the short-term one) is for live events where you cannot mix perfectly to the standard.
This is what I understood from the EBU specs. There is also a webinair well made by them:
http://tech.ebu.ch/events/webinar_loudness1/cache/off?id=13224
I am currently testing the NuGen Audio VisLM which has a great interface with all the data available and the also the LRA displayed. It would be great if Steinberg would implement something like that.
I am still searching for how will this new standard be implemented in post (cinema) as there are movies without dialogue. In this case, I believe you just mix by the ear and afterwards you do not normalise but just say to the distributor that your integrated value is lower and it should stay that way. And also, for cinema do -23LUFS/-24LKFS = 85dB SPL calibrated on each speaker with a 500-2500Hz pink noise?
Regards to all.
PS: Sorry for my bad English.
Actually, as far as I recall having read the purpose of it in the US wasn’t dynamic range but “absolute loudness” of commercials relative to “regular content” programming. In other words too many people had complained that commercials were too loud, so now the loudness is “set” through this new standard.
Hmmm… Not sure I’m understanding what you’re saying now. Are you saying that the EBU allows for the difference between loudest and softest loudness to be 25LUFS? So if your program has an average of -23LUFS (throughout) you can have a “maximum average” of -10.5LUFS? And minimum “average” is then -35.5LUFS?
I totally agree. A LUFS/LKFS meter/number display, with offline analysis/processing capability as a plugin.
Come to think of it, integrating the ability then to do a simple gain change would be nice. So for those of use that like to mix by ear, measure afterwards and do a gain change, that could all be done within one plugin. Should be very simple once the material is measured. Just gain all up/down by XdB to reach desired average. If the spec has a “hard” maximum peak value I’m sure they could integrate a simply brick-wall limiter as well. Convenient no?
Does it even apply to cinema at all? I thought it didn’t. I thought you just calibrated your monitoring environment and mixed by ear, levels “unimportant”…?
Seems pretty much spot-on to me!!!
You have understood the specs very good. You are spot on.
I am still searching for how will this new standard be implemented in post (cinema) as there are movies without dialogue.
AFAIK, this standard will not be adapted by the cinema-world, though even without doing anything, the outcome of both (cinema 85dB reference monitor “mixing by ear” and R128 mixing) will produce more or less the same results.
Fredo
Yes, you are right. For TV advertising and documentaries this is the main purpuse.
I was speaking from a cinema perspective, sorry.
This is what I understood. This is the reason for the LRA (loudness range) value. They were measuring The Matrix movie content and got to a conclusion of a 25LUFS\25LKFS LRA recomandation.
Yes, that would be great. That to state here that the gain must be made conform to LUFS/LKFS not dBFS, because they do not relate to each other.
I believe it must also apply to cinema because you send movie content to TV, so you have to relate to something.
I do not know for sure. I think that this standard can and would be best to be implemented to all the audio media. This way we could stay focused on the design more than on the tehnical side when we have to do corss-media transfers.
Thank you. 
I did not know about that. These are also my thoughts.
You are both correct. The standard is meant to iron out the differences in level between program material, and it has the advantage that it “promotes” dynamic range. As I said before, heavely limited/compressed program material will quality-wise be significantly disadvantaged in favor of program material with a wide dynamic range.
No, LU-metering is a new, differently weighted, loudness display. In no way related to PPM or traditional loudness meters. LU= Loudness Units. These Loudness Units are measured over the complete length of the program material. So it’s actually pretty hard to mix to LUFS in real time. At the end of the program material, the outcome would have to be in the range of -23 LUFS. So if your mix ends up at -25LUFS, you can simply nromalize it to -23LUFS. However, this process will raise your loudest dynamic peak by some extend, so it might become too loud then. Or the other way around. (Though a gate is part of the spec, so that should not be an issue) When your mix ends up @ -20LUFS, after normalizing, your dialog might be to quiet. So therefore the recommendation to adjust monitoring levels, so you can forget the meter and mixing by ear becomes the most easy and accurate way of mixing.
Depending on how you look at it, the analysis/normalizing is somewhat part of the EBU/ITU standard.
So yes, that is what we should expect to see.
See my answer above.
But we have to make the difference between Cinema for Theatre and the same cinema mixes that end up on DVD or TV. It is common practice to reduce the dynamic range of the original theatrical mixes so they fit the DVD/TV -world better. There was a time where a completely different mix wa made, but that is a thing of the past. Mostly they just compress/limit the cinema mix a tad.
That being said … since we have a mini loudness war in theatres too … we might get back to making a different mix that is meeting the R128 standard.
It doesn’t, but as explained earlier, these mixes will not stand up anymore against the other program material; while in the past (loudness war), they did have the advantage of sounding louder than the other program material. So, in fact these mixes will be penalised quality-wise. They always were, but weren’t percieved like that in the heat of the battle.
I understand that, and I am not saying that you are doing something wrong.
R128 is a delivery spec that is entended to be applied and enforced worldwide. So in the near future, if you mix a commercial, a promo, a TV feature or series, or whatever … you will have to deliver according the R128 specs.
So to avoid confusion, I am a big fan of R128 because not only will it end the loudness war, it will also “force” all of us to concentrate on the quality of the program material, opposed to focus exclusively on “being loud”.
Really don’t want to get into an argument here, but mixing at moderate levels should prevent exactly that.
I see what you are saying, and I have been advocating this half of my life. (Taking pauzes to “rest my ears”)
Until I started working in post, with fixed (low) monitoring levels.
Please don’t take this personally, I am not saying that you are monitoring too loud. But ear-fatigue is a typical problem of (too) loud monitoring levels. As long as you mix in the “sweet spot” of the Fletcher–Munson curve, your ears should maintain the equal-loudness balance on all frequencies without degration or fatigue. Of course, there is human fatigue … Let’s drop this part of the discussion, it is not my intention to become personal, and I fully realize that we can’t have this discussion without becoming personal. Whatever works for you; and your way might work better than “my way”. All I am trying to say is that there are many more arguments in favor of monitoring at the “correct” levels than monitoring at the level “that suits me best”.
That should be possible for music content, but IMO not possible for Post-content.
What if you are within specs, and the last 5 minute of the movie (or whatever) is one big fight scene with plenty of gunshots and explosions? You can’t possible anticipate for that.
You can rightly argue that “mixing by ear” can’t anticipate for that neither.
That’s why I think that there is no point in trying to match the meter readings during the mix. That’s why I advocate “mixing by ear”.
The procedure in post should be as follow:
-Mix the program material by ear.
-Measure R128
-Normalize the mix to -23LUFS (or whatever number is enforced)
-Next pass, apply adjustments.
-Measure R128
-Normalize to -23LUFS
-Next pass … etc …
Anyway … that’s my idea about R128 and how to introduce it into the workflow.
But the bottom line is: whatever works for you.
Regards
Fredo
Ups… I did not read the “average” word in Lydiot sentence.
To state all I know, you have to have an overall LUFS/LKFS average (Integrated loudness) of -23/-24, as Fredo said. Your Short-Term LUFS/LKFS can be in an maximum (recomanded) 25LUFS/LKFS range (meaning shouts and wishpers, explosions and fight scenes, etc.). Also the LRA range is measured based on an “algorithm” so any short high-level sound will not be counted. This LRA measurement is for letting you know if a dynamic compresion is necesarly or not.
But this is for a film with a big dinamic range. For TV related media this dinamic range is to much.
Hope I understood right.
Well if quality was the primary concern then we wouldn’t have had the loudness wars to begin with.
The reality of it is that I’m not competing with myself to create a good sounding mix, I actually think I can pull that off. Instead, I’m competing with other engineers who are selling their services and doing things their way. Thus my only concern, as I said, is pleasing my client. If they want a mix with a narrow dynamic range then that’s what they get. I might mention that I personally prefer a wider range, but if they’re liking it, I shut up. That’s all there is to it.
This may sound “insane”, but I’ve actually spoken the truth to several clients, trying to inform them about the technical aspects of mixing, and lost the gig! In one case it was about loudness and the balance of frequencies, broadcast processing etc and predicting the results once the spot was on air. Some people you just can’t reason with. So you’re left with either telling them just enough to save your own arse should there be a problem down the line (“he told us so”) or telling them what they want to hear.
Also, I never argued for “heavily limited/compressed program material”.
And that’s exactly what I thought.
We agree again.
But see here you’re making a blanket statement that I think is untrue. If we take 1000 post engineers and have them listen to 10 different mixes with different dynamic ranges, and have them say what they prefer, we may indeed find a sweet spot for what they like. But I propose that the “range of ‘good’ dynamic range” is wider than you imply. Just because it’s “loud” doesn’t mean it’s so heavily processed and limited it’s garbage.
Secondly, I think it’s highly dependant on the content, and I was referring to commercials.There’s a huge difference between a commercial adhering the CALM act loudness having the correct average but with the dynamics of a cinematic drama, and a cinematic drama with the exact same average and dynamic range. In the latter case it’ll be appropriate whereas in the former it’ll be inappropriate and sound simply “off”, because commercials aren’t intended to sound like that.
I totally see what you’re saying from a technical standpoint, but this isn’t about that.
Absolutely, and I never monitor loud for more than at most 1 minute (post). However, I live in NYC and the city and the lifestyle its residents have (me) sometimes just tire my ears. This is about more than monitoring when working. Or have you ever had a cold and had your ears get “stuffy” without thinking about it?
And I agree with that. I don’t think it affects my point in the least. If my ears are tired and I’m monitoring at “correct” levels I’ll still hit the “feels too loud” level faster (or opposite if my ears are clogged), resulting in my material either having a different dynamic range or just lower average loudness. A glance at the meter on days like that helps me. Even if I’d mix entirely by ear I could hypothetically adjust my monitoring level that day simply because my monitoring is telling me my hearing is off that day - know what I mean? Also reminding you that my workflow on some types of content makes it more time-efficient for me sometimes to work with my eye on the meter.
As for not making it “personal”: When you advise people you’re using terms and statements that are “absolute”. Of course the only logical conclusion is that views that don’t conform with the absolute are inaccurate. Not saying that’s your intent, but that’s the way it looks. Perhaps it’s not you, maybe it’s me.
You’re absolutely right. I actually mentioned that “This is for mostly short duration stuff though, like commercials and infomercials. For programming (lifestyle tv for example) I’m leaving a bit more dynamics, but deliver the same average.” It probably got lost in the discussion.
Yes, for feature length stuff, for things that should have a wide, natural “good” dynamic range I mix by ear solely. For commercials and infomercials of short duration however it’s not the same. And lifestyle TV falls somewhere between those two in my opinion. If I can work faster and produce an equally acceptable mix by using meters as I mix I will do so. It’ll save the client money which makes them happy. Nobody cares if the dynamic range of a daytime cooking show increases by 6dB and that makes it sound “better”. Nobody. But they do care about saving money.
But as I said, you might have thought I was referring to any and all post mixing, which I assure you I didn’t.
Well, I’ve not seen any requirement in any spec I’ve received for a maximum dynamic range. Out of curiosity, could you point to the specific document?
Sure. Here is it:
Other documents on EBU R128 can be found here:
Please give a feedback; I might have understood it all wrong.
Cheers!
Well, don’t take my word for it.
But if you take time to read the EBU production guidelines, you will bump into these exact same terms and statements.
Fredo
Really? Such as?