1
$\begingroup$

I am developing a loudness meter conforming to the EBU R128 recommendation. Currently, it works with stereo files but not with mono files.

In the ITU Recommendation, where the metering algorithms are defined, weightings for the individual audio channels are given (table 3, page 6). From that, I assumed that the weighting for a mono channel is 1.0 (0 dB).

As you can see in figure 1 on page 3, after applying k-filter, mean square and channel weighting to each individual channel, the resulting values of each individual channel are summed to one value. Logic dictates that when having two channels of the same audio signal, the value will be higher than when just having one.

As a consequence, a mono file will result in short-term loudness values, that are 3 LU lower in comparison to a stereo file with same content (on both L and R channels). This is clearly wrong, because established meters like "Klangfreund LUFS Meter" and "MeldaProductions MLoudness Analyzer" will output the same short-term loudness with both files.

How am I interpreting the algorithm definition wrong? How to deal with mono signals?

Side note: It seems that FFmpeg once had a very similar issue, but I could not find the solution to it.

$\endgroup$
2
$\begingroup$

Use a channel weight of 1.0 for your mono signal.

The weighting of 1.5 presented in ITU-R BS.1770 addresses the fact that a signal coming from behind is perceived louder by a human than if it is coming from the front (Reason: To better detect an enemy sneaking at you from behind). With a single mono signal this fact doesn't need to be addressed.

The 3 LU difference you mentioned also makes perfect sense from a physical point of view. One speaker in front of you compared to two (equal) speakers only provides half the energy and therefore the SPL will be lower by 3dB.

I'm honored that you use my LUFS Meter to compare to, but I'd recommend to also use the test files from ITU and EBU. The LUFS Meter manual contains the links to these test files.

Are you aware that the sources of the measurement algorithm of the LUFS Meter as well as some development notes are available? (Sadly I'm not allowed to post a third link... They are at docs/developmentNotes/)

Disclaimer: I'm the developer at Klangfreund.

| improve this answer | |
$\endgroup$
  • $\begingroup$ Thank you for this very helpful answer, Samuel. I think I now know why Klangfreund and other VST meters show the same value for mono and stereo files. It's probably because the DAW I'm using (Samplitude) instantaneously doubles all mono objects to both L and R output channels (i. e. making it louder), before routing it to the VST plug-ins. So everything makes sense. $\endgroup$ – Seb3736 Aug 6 '16 at 14:49
  • 1
    $\begingroup$ @Seb3736 You're exactly right. Actually, a plugin tells the host in which channel configurations it operates, e.g. 1 in & 1 out, 2 in & 2 out, etc. But sadly quite a few hosts ignore this and a stereo instance of a plugin will be used on a mono track (or other strange things happen...). $\endgroup$ – Samuel Gaehwiler Aug 6 '16 at 15:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.