How accurate is SC LUFS meter?

Hi all,

Anyone one else compare the loudness meter in SC to other plugins/hardware?
I export at -23 LUFS from Reaper and ProTools 11.
In both cases, the loudness level was measured with the RTW meter (screenshot below),
the Youlean plugin and Melda plugin.
In all cases, they correspond to within 0.2 LUFS, where as importing the audio into SC,
it consistently shows 2 to 3 LUFS too loud.

SC version 20.02.17 running under MacOS and audio channels set to 2.

Anyone else finding this?

The RTW R128 meter I always use on live broadcasts and my reference is this one:

Screen Shot 2020-05-24 at 10.37.41 PM

Can you give more detail on how you make the comparison? Shotcut does not report a “final” loudness measurement of a clip. The Audio Loudness Scope was designed to give real-time status on a clip while it plays. But the values depend on when the clip starts and stops. It resets its values if any frames are skipped or if seeking occurs.

Hi @brian

Thanks for the quick reply.

On the DAW, I have the Youlean and Melda plugins on the master output.
I let the whole song play, so as not to skew the readings.
Every 30 seconds or so, I compare the Momentary and Integrated loudness.
of each of the two plugins, they correlate with each other.

I export the audio as a signed stereo PCM (wav) LE, 24 bit at 48KHz sample rate.

I then play the same song via the playout server and look at the readings at roughly the
same time points, the RTW meter reads the same.

I then import the audio into SC and play it.
At no point do I pause, skip or scrub.
The readings are always too high by 2-3 LUFS.

Do your analyzers agree with this online analyzer?

I analyzed a file online and it gave me:
Momentary Max = -5.78 LUFS
Short Term Max = -7.21 LUFS
Integrated = -8.44 LUFS

Then I played the file in Shotcut with the loudness meter open and it also showed an Integrated loudness of -8.4 at the end of the file. Shotcut does not report an max momentary or short term. So I do not know how to compare those.

look at the readings at roughly the same time points

I feel skeptical of this process. With the file I was testing, the short term and momentary loudness values were moving around very quickly. I do not know how one could visually compare to another meter in real time.

Shotcut uses this library for loudness measurement:

That library has unit tests to check that it conforms to the standard reference files. So I have high confidence in the underlying implementation.

If there is a problem in Shotcut, it could be in its improper use of the underlying library. But I would need a better test case in order to qualify an issue and resolve it.

Also, I never intended for the Loudness Meter to be used to analyze complete files. So I would also want to better understand this use case.

I got -23 and the online one showed -22.64
Good enough for me.

That depends on the audio.
The test song I was using, is a typical modern (very compressed) song, so the readings
do not fluctuate that much.
Rather easy to see in realtime in this case.
Reaper allows side by side view of plugins (floating GUI), so a quick glance is all that is needed.

Shut SC down and restarted it and all is well.
Could have been some weird bug, can’t say, it is now reporting exactly what everything
else is.

So to answer my original question, how accurate is the loudness meter is SC?
I would have to say very and the library is obviously properly implemented.
However what still bothers me is, why it was showing the wrong values until I restarted it.

Will keep my eye on it and report back if it happens again.

Are you saying that there is some sort of time or ram limitation?

I mean that it was not designed to specifically process a file from start to end and report final results. It is designed to process the audio that comes through the preview.

This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.