Average Audio Peaks
 

I've recently been asked to mix a spot for a network with a 'target loudness of -24LKFS with instantaneous audio peaks not exceeding -6dBFS with average peaks not exceeding -10dBFS'.

In the past when mixing to -24LKFS, I've seen specs stating average audio peaks around -17 or -18 dbfs..With this particular target at -24LKFS, doesn't an average peak of -10 dbfs seem a little hot? Perhaps room for peak effects at dramatic impact? A network precaution?

Cheers.

Re: Average Audio Peaks
 

Actually, -17dBFS for peaks with a -24LKFS seems low to me. I haven't seen a spec with peak limits below -10dBFS.

Re: Average Audio Peaks
 

This specs makes no sense. The point of a loudness spec is that the peaks become irrelevant. A peak limit of -1 dBTP or slightly lower is technically needed to avoid full scale. What I guess they are clumsily trying to express is the concept of dynamic range or Loudness Range, and how much they want.
New meters are coming out with that measurement, so for example you can have a programme with a LRA (loudness Range) of 20 dB (quite large and possibly not adequate for general tv broadcast) or mix the same with a lower LRA of 10 dB more or less.
Cheers

Re: Average Audio Peaks
 

I have seen this type of spec before, but it's been a couple of years now. In my case, "Average peaks" really meant the upper most swings of a "VU meter", or "NBC meter", which is based on RMS measurements. (BTW, LKFS is also an RMS-based measurement.) If that's the case, you've got some room for dynamics by keeping the RMS meter reading at that and below. Or, it could be someone's request that the peak levels generally stay around -10dbfs with only an occassional hit going higher. If it is the latter, your dialog dynamics will need to be more restricted, but at least you've got room for some music or sfx hits above that. I'd try to get some clarification about what they mean by "average peaks".

Re: Average Audio Peaks
 

Soundthinker, the issue was 'average peaks' and what is meant by the definition, not instantaneous peaks. If it was instantaneous peaks, then yes -17 is way too low..

Postman, I'm with you. It could mean one of the two situations..If it does mean that average peaks at -10dbfs translate to the upper most swings of the VU meter (RMS), then wouldn't it be difficult to manage -24LKFS for all material? (if it actually does in theory hit -10dbfs)

These 'average peaks' would have to be instantaneous as well in regards to RMS, or the overall LKFS would reading would shoot up way past -24LKFS.

Here's another look at the spec:
'all material to have a measured loudness of -24LKFS averaged over the entire commercial. instantaneous audio peaks not exceeding -6dBFS with average peaks not exceeding -10dBFS'

On top of that, these are the specs for commercials on this network. Maintaining -24LKFS on commercials is also a debated topic. On the other extreme, some people may even take these specs to mean that it's all fair game as long as the RMS doesn't go above -10dbfs and peaks are maxed at -6dbfs.

Thoughts?

Re: Average Audio Peaks
 

My thoughts are that it is a brilliant example of a terribly written specification.

I said I'd seen something like this before. I looked through my folder of specifications and found the document I was remembering. This is from a show mixed December 2009. These specs are no longer in use, they were revised last year to ITU BS.1770, but I thought it would make for interesting discussion:

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

3.1.2 Programs are to have average loudness levels that fall between -28dbfs and -20dbfs during the majority of a program as measured on a digital meter calibrated to the RMS/VU ballistic. Average loudness should not go above -17dbfs at any point during the program.
The bolded line above means "the VU meter cannot go above +3VU", since the calibration level is -20dbfs, and 0VU=-20dbfs sinetone.

3.1.2 Programs are permitted to have audio levels that regularly peak near but not above the following limits using a digital peak meter:
SD: -10dbfs
HD: -3dbfs

3.1.4 (excerpt) Programs...must have dialog levels with a value of -24dbfs+-2db. Programs may have peak music or effects levels up to the level limits specified in 3.1.3 during moments of dramatic impact as long as dialog levels are maintained as specified.

When I read your specification I think it is a similar circumstance, where the "average loudness" is probably based on RMS meters, not instantaneous peaks. My spec was to peg average levels to -17dbfs, your specs say -10dbfs. However, you'll never know what your specs mean until they explain it.

Re: Average Audio Peaks
 

Nice one Postman. Thanks for putting this up. I know the exact network this is taken from. With average loudness between -20dbfs and -28dbfs and average loudness limits at -17dbfs with dialog around -24dbfs it doesn't seem like the best written spec. However it sounds much more doable than average program material at -24LKFS with average loudness peaks at -10dbfs!

For two networks that are showing program average levels around -24dbfs, that 7db difference in average loudness peaks seems crazy to me.

Under this situation, we are also both assuming the 'average peaks' in my spec is the same as 'average loudness' limitations in your spec. As you said in your earlier post in reference to my original spec, the -10 dbfs could mean that most peaks (not RMS) are around this level, while instantaneous peaks remain around -6dbfs.

It's really interesting to get everyone's take on this. Hairy stuff.

Best,




Original spec:
'all material to have a measured loudness of -24LKFS averaged over the entire commercial. instantaneous audio peaks not exceeding -6dBFS with average peaks not exceeding -10dBFS'