“0dBFS” sample peak is just that, sample peak, it isn’t the same as loudness. So perceived loudness can probably exceed “0”. There’s also “True Peak” which sort of reconstructs the actual waveform out of the sample values and see what the actual ‘true’ peak would be if you converted the signal. That too can go above zero.
You are talking about the difference between LUFS and the final dB Peak Level, in the Stereo Out?
This i sort of understand, from a producers perspective (im not an electric engineer, so i dont know 100% exactly how the convertion from Cubase engine to real world audio goes).
Where are you measuring the “4dB over 0” in Cubase?
A 16 or 24bit audio file cannot go over 0dbFS. Not possible, at least regarding sample values. If you measure “True peak”, as MattiasNYC mentioned, yes, that can definitely exceed 0dB.
If the file in question is in 32bit float format, then you can also get over 0dBFS. It is not recommended though, because there will be clipping at the DA converter, and it most likely won’t sound good.
You can get proper statistics about the original audio file by selecting it and then use “Statistics” in the audio menu. It will show you sample, true peak and LUFS values for the audio file. If the value for “max sample” is 0 or lower, then there is something in your Cubase signal path that is changing the audio.
I ran into something similar a few weeks ago while I was playing around with some factory content. I didn’t pay it too much attention as I figured I must have added gain somewhere in the signal path. But seeing this thread I went back and found the file. So anyone should be able to reproduce this.
If I put the file India Percussion 1.wav from the Loop Mash Factory Library onto a Track with both the Fader and Pre-Gain set to 0dB and play the file the Track will show a peak of 1.5dB. Looking at the file’s Statistics it shows the Right side peaking at 0dB with a True Peak of 0.02 (which is kinda sloppy on Steinberg’s part but not that close to 1.5). As you can see from these screenshots, no gain is being added.
Changing the Panner between Combined & Balanced produces the same results. It does convert the file from 16 to 24 bit when imported to the Project - but that shouldn’t impact the level. Real head scratcher???
@raino Lol that’s quite bizarre indeed.
Exported files in 24 bit or less can’t go past 0 dBFS so the 0.02 True Peak looks normal, even though I would expect it to be higher, maybe it has been limited with a True Peak limiter, but whatever, the meter should display 0, even if it was clipping at +8 during export.
What do other meters/plugins show ?
@WK1 Would you mind sharing a snippet of your track ?
Cannot find that sample here, but I tested with another sample that has a 0.0dBFS peak value, and all channel meters are at exactly 0.0.
I think that @MattiasNYC is on point here, maybe you have some pre-section filter somewhere in the signal path, as soon as I enable a LC/HC, there are overshoots.
That’s just an artifact of my Test Project sending other audio to Stereo Out. When I solo the test Track the Stereo Out level matches the Track.
No. This was a brand new default Audio Track added specifically to perform the test.
However playing around today I think I’ve found the cause - Musical Mode. The original Tempo for the clip is set to 90.670 bpm, which is also what I’ve set as the Project’s Tempo. The clip is 2 bars long and, as expected because of the matching Tempos, does not change length when turning Musical Mode on and off.
Look at this animated GIF. The first few times through the Loop Musical Mode is disabled and the meter peaks at 0 dB but then Musical Mode is enabled and the peak jumps to 1.5 dB. FYI, the overage occurs in the first bar around beat 2.
I was initially suspicious of this since my Test Project was at 120 bpm. Then I changed the Tempo to match the clip’s to avoid any possible stretching. But it didn’t really matter - if Musical Mode was enabled then it always peaked at 1.5 dB both stretched and not stretched.
Is this dBFS (full scale) or dB on the meter? It may be that cubase gives you some headroom on the meter before clipping.
Another option is that the file is exported in floating point format. Floating point can go above the “maximum” without directly distorting, although they will eventually have to clip when going into some kind of fixed point output, such as a DAC.
Although my guess is that the meters put 0.0 not at clipping, but a bit below.
Looks like it might be a bug to me. It’s possible that the musical mode option forces the audio to go through the time stretch code even if the tempi are the same and nothing needs to be done. In that stretch process there might be an artefact that accounts for the volume hike. Just to check - you have no event envelopes or offline processing, or anything else, affecting the audio?
I did try putting an Audio file of steady state brown noise on the same track and the meter levels did not change when I toggled Musical Mode on and off. Maybe Musical Mode does something funny only to transients?
In a practical sense this doesn’t really matter much. There’s ample headroom that the signal won’t clip. And using normal mixing levels starting with -14 vs. -12.5 isn’t going to impact a mix outside of a slightly different Fader level.
Maybe so, but I made a floating point file with a +/- 1.0 sine wave at 90 Hz, and imported it into a track in the default stereo mastering project setup, disabling the master plugins that comes with, and here’s what the meters say:
Why is it not “just the bits, mam?” Because recording started out in an analog world, and the transition to digital was long, messy, and gradual, and Cubase followed what actual recording engineers were doing, rather than trying to impose the desires of computer scientists on the business. I might have done the same thing in their shoes.