Issue link: https://digital.copcomm.com/i/1500716
66 S U M M E R 2 0 2 3 I C A S Q U A R T E R L Y Program Loudness than the Target Level. If special processing is applied, this is anticipated and still within the spirit of R128." And, finally, "It must be emphasized, though, that choosing an anchor signal generally is an active process requiring input from an experienced operator. This approach should only be considered after operators and sound engineers have become very comfortable with the concept of loudness normalisation. Performed well, it may help to fine-tune the loudness normalisation of wide loudness range programmes according to the chosen anchor signal." Which, pretty much, means "you need to be good at mixing." Let's all count ourselves in that category, shall we… Another measurement that comes up on spec sheets these days is Loudness Range (LRA). So, let's look at LRA as case in point for what I'd consider an over- prescribed spec. LRA was developed by TC Electronic, and is defined by them as follows: "LRA describes the overall program material range: from the softest part to the loudest part. The range is quantified in LU, and to avoid extreme events from affecting the overall result, the top 5% and the lowest 10% of the total loudness range is excluded from the measurement. For example, a single gunshot or a long passage of silence in a movie would result in a very broad Loudness Range even though it would not be representative in the big picture." What this means in practical terms is that LRA is a moving target—it's possible to turn down the loudest sound in a mix in an attempt to reduce your LRA and actually end up with a larger LRA value. This is because by reducing the volume of something in the top 5%, it forces the calculation to reconsider what is between that and the lowest 10%, and your whole measurement starts from scratch. And this is true every single time you alter the dynamic range of the mix. I cannot stress enough that both A/85 and R128 do not recommend that LRA be used as a delivery spec, but rather as a guideline and an informational tool in the mixer's toolkit. R128 Tech 3343 puts it this way: "It is important to understand that it is impossible to define one maximum value of LRA for all broadcasters and all programmes. Furthermore, LRA is very useful as a mixing tool and should not be a brickwall parameter for delivery specifications of programmes." Despite this, Disney has included max LRA value of 20 LU in their delivery specs, and I can tell you from personal experience that this moving-target value can be incredibly hard to nail depending on the nature of the content. I do understand why Disney chose to include this, but you can have an LRA of 20.5 and literally spend a whole day trying to get it down to 20, and gain nothing that could be considered perceptible (I've been in this situation). They also suggest the use of the NUGEN LM Correct plugin to wrangle your print master into spec if need be. And while the results are impressive, it's a little bit of "voodoo magic" for my taste since you can't really tell what it's doing. Additionally, you can't recreate it for the Atmos deliverable unless you use a 7.1.4 re-render, which some say is as good as an ADM or DAMF, but I would disagree with that assessment, and I know that Dolby does, too. My own takeaway from this is that the more we're expected to chase after numbers for the sake of hitting them, the worse this problem is going to get with respect to consistency and faithful artistic execution of the desired mix. Scott Kramer advises, "We've known for some time that mixes with moderate dynamic range play better for Netflix members. One way for mixers to ensure that LRA generally stays out of enforced specifications is to use dynamic range sparingly and with intention." So, once again, we need to exercise appropriate restraint while we're mixing and use dynamics where appropriate. But it's our responsibility to push back on specs that are an unreasonable expectation or cause overruns on the stage that cannot reasonably be described as the fault of the mix crew. The knock-on effect of all this is how it affects dialogue intelligibility. The idea of loudness normalization is to make the perceived loudness consistent across the viewing experience. If you follow the assumption in R128 that the difference between these two measurements is going to be less than 1 LU, then this makes sense. But since this is frequently not the case, then the only approach that makes sense is dialogue anchor. As Scott Norcross from Dolby observed in our last article, if you "have a show with some dynamics and the dialogue is at -24, but then the relative level gate (full program mix loudness) might be 6 dB louder, so -18. So, to align the full program mix loudness, I would actually have to drop the whole thing by 6 dB, which would mean that your dialogue is now at -30 LKFS. The misconception is that A/85 specifies 1770-3 for everything, but it's only for short- form (commercials). For long-form content, you use the anchor element—and the anchor element is the dialogue. It is still a 1770-based measurement, but only measures the loudness of the dialogue/speech of the program."