1. Correct.
2. if #1 is correct then this point cannot be correct!
3. OK, let's create an example to illustrate why not, let's take a drum kit: A drum kit has an enormous crest factor, it's comprised of a number of large transient peaks with relatively low level sustain and short decay times. Let's we have a consistent (doesn't change volume) drum kit recording with an arbitrary crest value factor of say 30, we'll call this "A". Let's say we apply compression which results in a crest factor of 10 and then take some sections of the recording and reduce them in volume by a value of 10, we'll call this "B". Obviously "A" has a far bigger crest value than "B" but "B" will sound like it has far more dynamic range than "A".
3a. What I'm saying has been known for decades but what you're saying has never been known because it is incorrect (or has been known by those who have a similar incorrect belief)!
4. As you say, the sausage shape in a DAW is caused by many peaks in the song having been compressed/limited but again, if your point #1 is correct then dynamic range is not affected "to a substantial manner" by these limited peaks! As loudness and therefore perceived dynamic range is largely unrelated to peak voltage levels, then the sausage shape may or may not be indicative of a small dynamic range. What do you think is the correct answer to this question: Which is louder: X. A signal with 3dB of compression, limited to -0.5dBFS or Y. An uncompressed signal which peaks at -3dBFS?
5. How's that going to work, when peak level and loudness are not directly related? If we peak normalise our most dynamic track to -0.5dBFS, what are we going to do with our less dynamic track which peaks 1dB higher, we can't have +0.5dBFS! I'm afraid none of this makes much sense, you're not considering what loudness really is, how LUFS works or the relation (or lack of it) with peak levels.
G
But dynamics, as we hear them, are affected by presence of both peaks(percussion, hand claps, etc,) and changes in the average levels(remember my 'soft verse vs loud refrain' example?). Compression and limiting tools can be used to even out both.
THIS:.....
Is what I have been trying to express verbally, albeit with disastrous results! (With eternal gratitude to Mr. Robert Katz)
Apply the following assumptions to the RH graph in this illustration:
1. The top of the vertical black line represents digital peak - '0dBfs'
2. The horizontal black line represents the average level.
3. The colored vertical bars represent the Average and Peak levels of songs on a typical album.
Now correct me if I'm wrong, but I believe that the most dynamic 'songs' are to the left, and less dynamic ones are to the right, as arranged on the graph.
THIS is what I have been stumbling to say all along: If the left-most song is the most dynamic, it is peak normalized just shy of 0dBfs on the scale in this graph. IT's average loudness becomes the average loudness for the entire album, and thus the horizontal black line is being used in this example to represent that. All the OTHER, less dynamic, songs songs are raised or lowered, as need be, so that their averge loudnesses all approximate, both by ear and meters, the loudness of that left-most track.
MY issue was confusing the argument with what to label the measurement at the horizontal black line: dBfs, LUFS, R2D2, or something else?
Suppose, in that graph, HYPOTHETICALLY, the average loudness of that most dynamic(left hand) track happens to fall around -7dBfs? How is that -7 expressed as a measurement - LUFS? dBfs? We're assuming operating in the digital realm here, so what'll it be?
I hope the use of Bob's graph here clarifies what I was trying to convey over the past 5-6 posts.
*Apologies for all my misspellings - I know in your eyes they do nothing to aid my crediblity! lol*