Mixing Levels: K-System vs Peak for cues??

[Len911]

Maybe I'm crazy,lol, but I use either the Sonnox Limiter or Elysia Alpha.

I have the Rme Adi2 volume set where I want it, and try out the presets on either one, maybe do a little fiddling with the Elysia. That's after I have the volumes adjusted in Cubase of course.

I usually don't have the compression or loudness I hear on most other soundcloud tracks, though there certainly are the presets for it.

Whether you use the K-system, or peaks, or whatever reference you use, it's still reference. My reference happens to be the 10 o'clock volume position on the rme, and a nice signal at 0db on the cubase faders.

Sonnox Limiter http://www.soundonsound.com/reviews/sony-oxford-limiter

...featuring a detection algorithm that is capable of spotting not only sample values that would hit 0dBFS, but also inter-sample peaks that might clip the D-A converter when the signal is reconstructed. The output level meters go to +6dB in order to display these peaks when the Recon Meter button is selected, and a second button labelled Auto Comp automatically brings the level down for an instant whenever such a peak is encountered, providing either a safety net or a second layer of limiting, depending on how far you're pushing the output level. The ability to detect and neutralise illegal inter-sample peaks can come in very handy even if you don't plan to use the limiter function at all.

[andygabrys]

I am sorry for muddying the water by not being more specific.

Where I wrote something like -77dB it should actually be 77dB. A measurement of loudness that theoretically has no maximum.

And in places where I have indicated an average loudness of -14 dB that is in reference to dB FS (Full Scale) where 0 dB is the theoretical an practical maximum.

And yes as Mojo has indicated, CREST FACTOR is more of less the difference between an RMS average loudness (say that -14 dB FS) and peak level (of the track) in question.

Hope that is better.

One more thought: although it's hard to figure out a situation where the Peak Level of a track would be lower than 0dB FS that is exactly what happens now in iTunes (with Soundcheck turned on) and on YouTube and Spotify.

In an effort to level the apparent loudness of tracks being played these platforms turn down the output of the track so that it appears as loud to the ear as much less hot tracks.

Case in point: Foo Fighters "the Pretender".

Listen to the Audio as bought from an audio CD or from the iTunes Store (with Soundcheck turned off) and then you the same track on YouTube and Spotify. I am sure you will hear a big difference. Now turn Soundheck on in iTunes. It will most likely sound pretty similar to YouTune and Spotify. Keep your monitor volume the same while doing all these listens.

So if the actual master has peak level of -0.1 dB FS, and a Crest Factor of -8dB FS, the YouTube and Spotofy versions might be peaking at -6dB FS while still having a CREST FACTOR of 8dB, and a resulting RMS of -14 dB.

Over to you Mojo. Make clear what I have bungled

[cardell]

I find this to be a very interesting subject! It took me a few weeks to [begin to] understand the K-System.

Many great, useful concepts in it.

Stuart

[mojobone]

I am sorry for muddying the water by not being more specific.

Where I wrote something like -77dB it should actually be 77dB. A measurement of loudness that theoretically has no maximum.

And in places where I have indicated an average loudness of -14 dB that is in reference to dB FS (Full Scale) where 0 dB is the theoretical an practical maximum.

And yes as Mojo has indicated, CREST FACTOR is more of less the difference between an RMS average loudness (say that -14 dB FS) and peak level (of the track) in question.

Hope that is better.

One more thought: although it's hard to figure out a situation where the Peak Level of a track would be lower than 0dB FS that is exactly what happens now in iTunes (with Soundcheck turned on) and on YouTube and Spotify.

In an effort to level the apparent loudness of tracks being played these platforms turn down the output of the track so that it appears as loud to the ear as much less hot tracks.

Case in point: Foo Fighters "the Pretender".

Listen to the Audio as bought from an audio CD or from the iTunes Store (with Soundcheck turned off) and then you the same track on YouTube and Spotify. I am sure you will hear a big difference. Now turn Soundheck on in iTunes. It will most likely sound pretty similar to YouTune and Spotify. Keep your monitor volume the same while doing all these listens.

So if the actual master has peak level of -0.1 dB FS, and a Crest Factor of -8dB FS, the YouTube and Spotofy versions might be peaking at -6dB FS while still having a CREST FACTOR of 8dB, and a resulting RMS of -14 dB.

Over to you Mojo. Make clear what I have bungled

Where to begin? The original question was kinda fuzzy, in the first place. Back when we were making records, as opposed to digital files, we had a similar problem with this thing called a 'broadcast limiter', something that's still in use at terrestrial radio, where if you if you traded too much dynamic range for loudness, you'd end up working against yourself, and The Pretender is one example; Metallica's "Death Magnetic" is another famous one.

Broadcast limiters are there to prevent overmodulation of the FM signal per FCC regulations and they don't work in quite the same way as the leveling algorithms used by iTunes and Spotify, but the upshot is roughly similar; you need to maintain a crest factor of 4-6dBFS. Not an exact figure, because there are different algorithms, (different limiters, too, and some stations fudge) but if there's less than 3dB, your odds are pretty good of having the leveler turn the whole thing down, and now your peaks are lower than everybody else's.

For a full explanation of these matters, try this: http://bit.ly/2biU0O8

It's Bob Katz' introduction to TC Electronic's Finalizer, and it's probably the best short article on the subject of mastering-the relevant part begins on page 9, but imo, you'll wanna study the whole thing. Uh, by "you", I mean y'all, not you, Andy.