EBU r128 Normalization and split stereo - is this a BUG?

One of our clients asked for a tv mix and a .L .R split stereo file.
So, after our mix we exported the file. LUFS normalization and split files option were selected in the export window.

We checked and measured the exported files. Apparently Nuendo did first the split and then the EBU normalization, so each file measured -23 LUFS. We recombined it in a stereo file and of course the output was louder than -20 dbfs.

Which for us it’s like a bug. If I have to deliver split files but EBU I would first normalize and then split those in multiple mono.
Has anyone encountered the same problem?


Haven’t seen that problem since I don’t typically normalize to LUFS, but I agree with the logic that a split stereo should be treated as stereo for normalization and simply render two separate files (left/right).

The implementation/design seems wrong.

Thanx Matthias.
Can you please try this in your spare time and let me know?
This is something that should be reprogrammed.


K, short export;

‘stereo’ when measured: -23LUFS
“L/R”: -23LUFS
"Split Channels: -26.27 and -25.77LUFS

In other words, I could not reproduce the error when using “statistics” inside Nuendo.

I then tried a different project with a longer timeline. Same 3LU difference.

How and where did you measure the exported audio?

Are you intentionally measuring lufs on the mono files and dbfs on the recombined stereo?

He probably is measuring recombined, which makes sense.

Some networks/production companies require interleaved stereo, some prefer .L/.R instead. So for QC I would imagine the delivered .L/.R files will be combined and measured as a stereo signal, not two separate mono signals…

Having said that I of course know how a lot of QC departments (mal)function, so maybe I’d expect nothing less than something illogical…