Normalize items to RMS level (using the info from "get peak info" command

[Jeff0S]

From musicbynumbers@gmail.com on March 09, 2010 02:06:52

Xenakios's command "get peak info" (that also gets rms info from an item)
could be used to have rms normalizing over multiple selected items. This
could work out all the rms values of the items, set them all to the same
level and then using the loudest peak in the loudest item it could
relatively pull them all up in volume.

Original issue: http://code.google.com/p/sws-extension/issues/detail?id=60

[Jeff0S]

From swstim on August 18, 2011 11:01:30

Ok! Added actions:
"SWS: Normalize items to RMS" (entire item)
"SWS: Normalize items to peak RMS" (max RMS over window)
Set the window with "SWS: Set RMS analysis/normalize options"

If you want to make them as loud as possible without peaking then follow with native "normalize multiple items to common gain". r534 v2.1.0.2

Status: Fixed

[Jeff0S]

From musicbynumbers@gmail.com on August 18, 2011 11:29:36

BLOODY GENIUS!!!! :) THANKS!!!!!!!

Tim, question, is there a difference then between using different window sizes when not in a real-time environment? I didn't look into this closer after my last comment and be good to know if you found there was a reason? :)

I guess choosing an RMS allows us to calibrate to other meters that respond close to say human ears or to say the K-metering system but as far as I know taking the value over the whole item is the most accurate but may differ from real-time (well within a "window") meters.

Anyway, bloody cool stuff! :)

[Jeff0S]

From swstim on August 18, 2011 11:51:26

MBN,

Well, windowed vs full-windowed are kind of different concepts here. Non windowed takes the RMS of the entire item, windowed takes the RMS many times, for small-sized chunks, then picks the highest.

The best was to describe is with an example. Say you have an item that's a drum hit, 1s long. Pretty much all of the energy is in the first 100ms. If you calculate the RMS for the entire item, it ends up being pretty low because 90% of the item is quiet. If you do a windowed RMS with window of 100ms, the first window is large (where the energy is) and the rest get progressively smaller, so the first window's RMS is used.

Which action you pick depends entirely on what you're doing! I can see both being useful.

[Jeff0S]

From musicbynumbers@gmail.com on August 18, 2011 12:15:01

Guess it's a bit like the gate function then in LUFS measurements but slightly more simple :) Will try both on a few things and let you know if there's any weird stuff going on!

[Jeff0S]

From musicbynumbers@gmail.com on August 18, 2011 13:00:35

Just been testing it and so far it seems to work really well :)

had a question, does the "analyse" action now work on the RMS settings to get it's window size from or does it do the whole file? I guess it's the whole file which is probably the most useful for looking at say the overall crest factor of a track? :)

[Jeff0S]

From swstim on August 18, 2011 13:03:00

Analyze is just the whole file (at the moment).

[Jeff0S]

From musicbynumbers@gmail.com on August 18, 2011 13:09:07

cool, no probs just good to know, this is seriously useful! :) I often have to compare files for lots of reasons and balance things up and although you still have to tweak a little by ear this makes the process so much faster! :)

again, thanks!