dpMeter2 is used as a control room insert.
/Magnus
Looking more closely at file “0799TPlimit” (signal of 0.799 dBTP with Cubase’s Brick Wall Limiter (BWL) set to limit at -1.0 dBTP) - I think the BWL is processing correctly after all:
I propose for discussion (and trashing of the logic if wrong) …
-
dpMeter2 has a measurement/display error: though the BWL is limiting to -1.0 dBTP, dpMeter2 it is reading -0.8 dBTP. Why is this more likely a dpMeter2 problem than a BWL one? Because Cubase’s control room, downstream of dpMeter2, is reading -1.0 dBTP.
-
Cubase’s BWL input display is in error. It is being fed a 0.799 dBTP signal, but is displaying a signal amplitude of only 0.6 dBTP. I don’t know if that’s because it was such a short spiky signal, or …? (I can’t remember what the “link” button does either …).
But even though it’s input display is off, Cubase’s BWL processing seems to be accurate, as the Cubase Control Room meter indicates it is getting a signal of -1.0 dBTP.
Just my two cents, if this thinking is wrong I wouldn’t be surprised!
Just thought I’d run this through the Waves Loudness Meter plugin as well… it reports 0.7dBTP and 2.0dbTP.
Running Statistics on the audio clips (in Cubase, Audio->statistics) gives 0.56dBTP and 2.01dBTP.
Also, Youlean Loudness Meter reports: 0.8dBTP and 2.0dBTP.
With the BWL set to -1.0dB: Waves gives -0.9dBTP, Youlean gives -0.8dBTP, Control Room gives -1.0dBTP (same as above).
But, very interestingly, I got completely different Integrated LUFS: Cubase Control Room gave -14.7, Cubase statistics gave -36.15, Waves and Youlean both gave -37.7. The Control Room result seems very strange to me.
Mike.
posted in error
I think the nature of the signal being tested, very “spiky” (see Magnus’ file attachments), may not lend itself to meaningful LUFS calculations, as the calculations are based on averaging over a timescale that may not apply to signals like that. For example, from Sound On Sound mag, Feb 2014, The End Of The Loudness War? :
… After this simple filtering, the next stage determines the average signal power using a mean-square calculation (ie. the average of the squared values of sample amplitudes). The power is averaged over a 400ms measuring period which is updated every 100ms …
Interesting stuff! Seems everyone is using a different dBTP algorithm, and (at least for this one signal) Youlean’s is the most accurate.
Oh … that assumes the signals themselves are labeled correctly! (Oscilloscope, anyone?).
Tough again to say for sure why the results are different, except maybe just another manifestation of different algorithms being used by different companies to estimate dBTP.
I wish someone with an oscilloscope had a moment to run these through!
Agree. You cannot test without the right test signals.
IMO those files are not a lot of use for testing loudness and true peak meter levels. Probably better to try a reputable source like the following instead:
How to use these test signals is explained in EBUTech 3341 (see table 1):
AFAIK the Cubase loudness meters function correctly and reliably in accordance with the expected results of these test signals.
You may well be right. On the other hand, Cubase’s ctrl room meter and Brickwall limiter likely use the same dBTP algorithm. If so, they would show the same.
/Magnus
Those are interesting results, Mike, thanks for taking the time. As alexis says below, however, the integrated loudness is a measurement over time, meaningful for measuring and setting the perceived overall loudness for (in the context of music) a whole song. For a mere spike of a few samples like the test file, it is not useful. Further, if not reset, I believe Cubase’s Ctrl Room loudness meter will display the integrated loudness accumulated since the last reset of the session.
But the variation between the dBTP values of various plugins is fascinating. I guess dBTP is an extrapolation that varies with algorithm and upsampling, thus resulting in a spectrum of results. The test files seem to imply that some of these plugins are more accurate than others. The variation is yet another reason to master to at least -1.0 dBTP.
/Magnus
Thanks, stingray, for the links and input. I found and downloaded the EBU test files a couple of days ago. However, I didn’t find the tech document you link to, and without it, the files weren’t of much use. I see now, that there are signals included for measuring true peak levels.
/Magnus
At the link page there’s an open file button… but here’s a direct link to the PDF:
Table 1 in this PDF provides a guide to using the test signals.
As a comparison I’ve prevously checked the Cubase loudness meters against RME Digicheck’s meters and they produce very similar results (sometimes very slight differences in the integrated values if I remember correctly). As mentioned, IMO the Cubase Loudness meters are accurate and reliable.
Thanks, stingray, that is good to know. Does that apply to true peak levels too, and the Brickwall limiters true peak capacity?
/Magnus
Interesting to see, on the downloadable EBU .pdf stingray linked to, the acceptable measurement tolerance for dBLUFS measurement/display is +/- 0.1.
And for dBTP it is +0.2 dBTP to -0.4 dBTP!
So … are the different measurements in Magnus’s initial posts of test signals measured by dpMeter2 and Cubase’s Control Room, some processed by Cubase’s Brick Wall Limiter, therefore all within spec?
(… even if the spikey test signal Magnus used wasn’t an “official” one …)
Yes, true peak measured the same.
Yes, there are tolerances. However, I’d expect to see the same on all meters when using using the EBU test signals. Using other signals like an impulse or a signal of only a few samples then you may see differences. I don’t know if those particular signals bear a relation to these tolerances…
I may be thinking of it incorrectly, but … I see the inclusion of tolerances in the standard values (Table 1 of the .pdf) as freedom to develop whatever monitoring/display system one can (like dpMeter2, Cubase Control Room, etc.), and it would be considered valid as long as it reproduces those standard values to within the stated tolerances.
In other words, I therefore would expect some variability between measurements of different plugins when using the EBU test signals. But I’m open to criticism of this point of view!
I never even considered that Cubase’s Control Room measurements could be out of spec before this thread! I think I am going to just assume they are within spec unless someone shows they are not … life is too short to challenge all of one’s assumptions!
Well if you get the time to do some tests the results will confirm this (or not)…
Gotta leave it for the next guy up! 