We’ve been exporting mainly mono files, since we produce single narrator audiobooks. Every client has a different Integrated LUFS target requirement, so we have to be constantly double checking if we master them at the designated target.
The problem is that the handy built-in Loudness Analisys tool seems to be showing a result 3dB louder than the actual measurement, making us confused (due to that, we’ve delivered a couple of productions 3dB quieter than required). Can someone tell us if we’re missing something? Or is this an actual bug?
If you sum a signal with itself it will be 3dB louder. Maybe some conversion from stereo to mono for loudness analysis that doesn’t happen during export?
Can you detail precisely how to replicate what you did with a minimal example?
When the export of the mono file is finished and the pop-up Loudness Analysis is shown (image bellow), it states that the Integrated Loudness is of -14.5 LUFS, when is in fact -17.5 LUFS.
The same discrepancy occurs if I copy the exported file into a channel, select it, and choose Loudness Analysis from the Regions menu.
Does that mean that when you copy the file into a channel, then the loudness analysis shows -17.5 LUFS?
Or do you mean that when you copy the file into a channel, the Ardour loudness analysis again shows -14.5 LUFS?
This is the kind of problem I like to have: Robin (x42) is too busy to respond on here, but this just got checked into the code repository:
Author: Robin Gareus <robin@gareus.org>
Date: Thu Feb 2 17:57:02 2023 +0100
Fix EBU-R128 loudness analysis for mono signals
This removes the special case which assumed unity gain
when reproducing mono files on a stereo system.
ITU-R BS.1770 however specifies a channel weight of 0dB for
left, right and centre, regardless of the total channel count.
Tech 3344 6.16 mentions a 3dB attenuation to maintain the
loudness level of a mono audio signal in multi-channel signals,
and Tech 3343-2016 further specifies that "Ideally, a downmix
operation should be loudness-agnostic".
Don’t know what is up with the weird bolding, seems that the web form treats pre-formatted text like it should be source code and starts highlighting keywords, so just ignore that and concentrate on the fact that this is fixed, either build from source, grab a nightly build tomorrow, or wait for the next minor version release.
Use ```txt to start a plain text quote. If you do not specify anything after the triple backtick, some heuristics are used to determine the language to highlight.
In a typical playback situation a mono file will still be played to a stereo system. The result will not just be played on the left speaker only. This is commonly done simply by replicating the signal to the other channel, which adds 3dB of signal pressure level.
So the old (incorrect) code added that to the measurement.
The EBU spec however notes that the correct way for the playback system in that case is to attenuate the signal by 3dB:
6.16 Behaviour of mono audio at stereo and multi-channel outputs
The equipment shall ensure that a correctly signalled mono audio
signal with a loudness level of -23 LUFS will have the same loudness
measured at the stereo or multi-channel outputs as a correctly
signalled stereo signal with a loudness level of -23 LUFS.
If this is not performed in the internal decoder unit itself,
the general approach is to reduce the level of the decoder output
by 3 dB before passing it to the left and right channel outputs
This section was not present in the original revision of the original document (https://tech.ebu.ch/docs/tech/tech3344v1_1.pdf, 2011), which was the current version when Ardour’s EBU R128 analysis was implemented.
This is no excuse, but it offers an explanation about the previous state of things.
If the mono file were imported into a track and sent to the master bus via the default panner, that would lower the level to L and R by 3dB, correct? Was the concern about replicating the mono source with no attenuation related to playback devices?
Aside from the Loudness Analyzer issue but still related to the mono production workflow of the audiobook world, we would like to know if there’s a way to make the export of mono files pan-independent.
Currently we have been panning hard left to be able to get the correct loudness.
A side note: If you want to mix in mono you can remove the second input of the Master Channel. This will remove all the panners. See Mixing a 4 Channel Project - #11 by THM (but remove a channel instead of adding two, of course).
You can either bypass the panner right-click on the panner > bypass), or follow @THM’s sugestion.
By default every track has as many outputs as the master-bus has inputs.
If you want to work entirely in mono, you can directly create a session with a mono master-bus: New Session > Advanced Session, and later in the dialog pick 1 channel for the master-bus.
