How does Match Loudness actually work?

[HastieBeatz]

Hello!

I'm looking to set up a process for normalising my files, but in a pretty manual way. Basically, I want to know all the steps that are taking place throughout, rather than letting Audition use an automated process. The Match loudness process is close to what I'm looking for, but I'm thinking of the following basic steps:

1. Hard Limit - figure out what level to hard at based on a particular stat of the file.

2. Normalise to -2db

3. More? let me know if I should be adding.

It's step 1 I need some help on. What stats would you use to figure out a starting point for the hard limit? I know you're going to say 'it depends, needs to be done by eye' but the trouble is I need to write a script around this to allow it to be done automatically.

So, the Amplitude statistics give me a lot - do one of the RMS measures give me a decent starting point? Eyeballing it, the Maximum RMS Amplitude looks like it's in the right ballpark, for example.

Any help would be appreciated!

[davescm]

Hi - I've moved your post to the audition forum

Dave

[SteveG_AudioMasters_ · Accepted solution]

'Match loudness' works fundamentally by assessing the overall levels in your file, but it's measured in a way that gives a perceived loudness value, which isn't the same as any of the individual signal levels. It's loosely based on a thing called an Leq meaurement - you can read about that here: Equivalent Continuous Noise Level (Leq)

The thing about using the hard limiter is that you won't be measuring the file to establish what its overall level is, but you'll be altering it, and how 'loud' that makes it depends entirely on the type of file it is. For instance, if you do it to music you will be distorting the intended dynamics of it in a way that many people would find unacceptable - but if you did the same thing to speech, then the results would be far less noticeable, if this was done with a little care.

The direct answer to your question though is that the figure that gives you the most information is that one at the bottom, where it says ITU-R BS1770-3 loudness, with a LUFS number. In theory, all you need to do is move the overall levels of each file up or down until they all measure the same LUFS level - at that point they will all have nominally the same level of loudness, and you won't have to alter the dynamics (which generally is undesirable, if the material actually sounds okay). LUFS units are basically dB, so to make a -16 LUFS section a -23 LUFS one, then lower the overall level by 7dB.

[SteveD333]

I restore old vinyl for a hobby, and sometimes make my own CD compilations.  I use the Match Loudness function to level the overall loudness of the CD as a whole.  There’s no hard limiting or normalizing.  I just drag a CDs worth of restored content into Match Loudness, set Match To: ITU-R BS.1700-3 Target Loudness to -13LUFS, Tolerance 0.5 LU and the Max Peak Level of -0.1 dBTP.  I then export all the files into a new folder. Using that content to create the CD.

It’s something you can experiment with as much as you like, as long as you don’t overwrite the original files when you export, until you get something you’re happy with.  As I say, I use -13LUFS, and it works for me most of the time and is my default starting point.  I think -16LUFS is the more common setting, but each to their own.