The SWS/S&M extension is in my opinion an essential extension for the Reaper DAW, my favourite piece of software for editing podcasts. One of the things it offers are custom actions for loudness normalization. I learned about this when watching the excellent video training by Brendan Baker (which I’ve mentioned before):
In Reaper when I select an audio item and click on the N key it normalizes that item to -23 LUFS Integrated. Although I am using many of Baker’s key commands, I changed this one. Instead of using the action ‘Custom: Reset Volume and Normalize to -23 LUFS NORMAL’ I am now using ‘Custom: Reset Volume and Normalize to -23 LUFS (B-format)’. Let me explain why.
Problem: between stereo and mono there’s a 3 dB difference
When you drag a mono audio file into Reaper the audio is send to both speakers left and right with the same amount. So a mono file gets doubled and is in fact twice as loud as the original file is (3 dB = twice as loud).
Let’s say I have one track in mono which is -26 LUFS Integrated (mono). And another track in stereo which is -23 LUFS Integrated (stereo). When I import both tracks in Reaper both tracks will be as loud because the mono track gets doubled, 3 dB louder.
In almost all playback systems (both analog and digital) a mono file will be doubled in loudness because it is send to both speakers. Because of this when rendering a podcast at -16 LUFS Integrated this means that the mono version of that podcast needs to be -19 LUFS Integrated.
For a test in Reaper I created two files:
- a stereo white noise file
- a mono version of the stereo white noise file
The stereo version is twice as loud.
When I import both files into Reaper it looks likes this:
The mono file has a wave form which is as thick as both wave forms of the stereo file combined. Which makes perfect sense since both channels show the same loudness as Reaper will send the output of the mono item to the left and right busses, doubling its volume.
The problem with the ‘Custom: Reset Volume and Normalize to -23 LUFS Integrated NORMAL’ action
When I apply the action ‘Custom: Reset Volume and Normalize to -23 LUFS Integrated NORMAL’ to both items this is what happens:
As you can see the stereo item is set 3 dB lower, twice as soft. Is this correct? Of course not! As I mentioned above: both files are sounding exactly as loud. So setting the volume 3 dB lower creates a difference in loudness between both items.
What’s going? Well for both items the volume is set to -23 LUFS Integrated, -23 LUFS Integrated stereo and -23 LUFS Integrated mono to be precise. But since Reaper will double the output for the mono item this will be twice as loud. So the -23 LUFS Integrated mono item will sound as if it is a -20 LUFS Integrated file.
The action ‘Custom: Reset Volume and Normalize to -23 LUFS Integrated (B-format)’ is perfect
There’s another action we might try: ‘Custom: Reset Volume and Normalize to -23 LUFS Integrated (B-format)’. It uses the B-format method which is a special Mid/Side measurement technique:
Sum Mid and Side to obtain the stereo left and then sum Mid and the Side with polarity reversed to obtain the right side.source
When I apply the action ‘Custom: Reset Volume and Normalize to -23 LUFS Integrated (B-format)’ to both items this is what happens:
When applying this method it results in the same volume setting for both items. But because this method applies the special Mid/Side summing to a mono channel as well, what we get is -20 LUFS Integrated stereo (two times -23 LUFS Integrated).
For me it is not an issue to work at a reference level of – 20 LUFS Integrated. Knowing that when I click on the N key the item is set to -20 LUFS Integrated is perfect. Whether it’s a mono or a stereo item.
However when using the ‘Custom: Reset Volume and Normalize to -23 LUFS Integrated NORMAL’ action you would end up with mono items sounding twice as loud as stereo items.
I love the custom action ‘Custom: Reset Volume and Normalize to -23 LUFS Integrated (B-format)’ for setting any item to the same loudness. This makes editing a lot easier. So what I often do is divide a long item in smaller items which I then normalize using the N key. This speeds up the proces of getting all spoken parts sounding more or less as loud.
It’s also easier to work with compressors on channels this way. I am focussing on getting a -20 LUFS signal from my compressors. So it’s like -20 LUFS in and -20 LUFS out. I’ve set the Master Fader in Reaper so that it shows me the LUFS Short Term values (3 seconds) and have the threshold set at -19.8 LUFS. So it is super easy to see if the voices are at -20 LUFS.