Is this possible - limted track exceeding 0dB?

[mjbphotos]

So I took my mixes, rendered them to 24 bit stereo, then applied a very minor compression (per Ashcat's idea posted elsewhere), then the Kjerhaus Master Limiter to get the tracks up to -0.3dB but with only minor compression happening for the louder bits. Rendered and dithered to 16bit.
All sounded good to me, checked on my monitors, other systems and headphones.

A guy said one (yes, just one) of the tracks showed going into the red (he actually said 'exceeding Red Book standards', but then said it was on his meters as over 0.0 peak).
I understand my Reaper meters, at just one decimal place, may not be super accurate, but I also don't think they should be that far off, and for only one song!

He took one of my tracks that was somewhat lower in peak (around 0.7), EQed the high end up way too much, and boosted the volume/limited it so it peaked at 0.3dB (on my meters), but the regular (whatever its called) meter showed it going into the red. It also sounded like sh!t to me because of the high EQ boost it ruined the mix.

Thoughts/comments?

 

[ShanPeyton]

We may or may not be talking about the same thing here but when i started playing around with reverbs in reaper i noticed that during playback my master channel strip didn't log anything past -0.1DB.

After rending that same mix however, it logged a few peaks above 0db. It climbed as high as +0.7db i believe. I think someone told me it had to do with some settings within my reverbs. Differing bit rates or something? I am totally not being helpful at all, with my ambiguity here but it may be worth looking into?

 

[mjbphotos]

No, this was a mix already rendered and limited to 0.3dB per my Reaper meters.

 

[Track Rat]

If you pull up the mix in question in an editor and zoom in on the peaks you think may be clipping, what do they look like?

 

[ashcat_lt]

A fixed-point file can't go over 0dbfs. "True peak" meters might show that the wave could go over 0dbfs when reconstructed at the DAC, but that's not really what is actually in the file. A normal digital peak meter will show it hitting 0, but it just can't ever actually go over. /pedantic nitpicky crap

1) Are you sure that limiter is actually a brickwall/hard limit that keeps absolutely everything from going over the ceiling you set. A great many of these don't for one reason or another. Any RMS-averaging (even with lookahead to compensate for "lateness) make it impossible to say for sure because the limiting will essentially be frequency/time dependent. Likewise, any filter placed after the actual limiting will make it less than completely reliable. If this thing has any "warmth" or "analog" anything involved, this could be an issue. Oversampling also involves filtering, so that oversampled limiters don't always actually limit either.

Did you actually play all the way through all of these pieces with the mastering chain in effect and/or look at the peak meters in the render dialog to make sure it wasn't overshooting that limiter?

B) Sample-rate-conversion fucks up hard limiting for the same reason that any other filter operation will. You mentioned dithering down from 24 bit (the dither noise shouldn't be enough on its own), but didn't mention if there was an SRC process at any point along the way.

III) We are talking about .wav files right? They didn't get mp3ified along the way anywhere? There is no good way to predict what that might do to your peaks.

 

[mjbphotos]

It's the Kjaerhas Master Limiter, with a minimal compression (using ReaComp) before it - the way you had suggested adding 'just a little' to the overall volume. Yes, played all the way through, nothing showing on the render waveform indicating clipping.
And the 'clipping' the guy said he saw only happened on one of the tracks, although I gave them all the same mastering chain, just adjusted the limiting level as needed so bare minimum of compression was happening on each track.
Not sure what 'SRC' is, just used the Reaper dither option when rendering.
Yes, WAV, never MP3ed.

 

[Farview]

SRC=sample rate conversion.

It could simply be a difference in your meters. An "over" is actually just several full scale samples in a row. The number of samples needed to register on over differs from meter to meter.

The answer to the problem is to have the guy complaining about the over turn that song down by 0.3db and then both of you can get on with your life.

 

[ashcat_lt]

I kind of thought that I could get away with abreviating a term that I had used earlier in the paragraph...

You really do need to know what is happening and why if you're ever going to render anything again. One of my biggest pet peeves is limiters that don't actually limit! If I can't know exactly where my loudest peaks are going to be then I'm fucked and the "limiter" sucks.

Reaper's peak meters are dead-nuts-on by default. True, there's only one decimal place precision, but that can't possibly explain this issue. Now, if you've messed with the meter preferences... And the meters indicate exactly what the levels in the rendered file will be as long as you actually render what the meters are indicating. If the Master meter hits -0.3, and you render the Master Mix to the same sample rate as the project and go to an uncompressed format, the resulting file will peak at -0.3. Well, ok, somewhere between -0.25 and -0.34... Even if you can't trust the limiter, you can trust the meters.

...unless you've got a bunch of "live" randomization running on the mix. "Humanized" drum plugins, random oscillators, or whatever are kind of not supposed to do the same thing twice. Many kind of actually do play the same "random" sequence every time you rewind and hit play, but you can't always count on them. So, if you haven't previously rendered those things to stems, you can't really know for sure what might happen when you render it. Course the limiter should catch that, and the render meter should show it.

It is possible for a file that peaks at -0.3dbfs to push "inter-sample peaks" beyond 0, but we can't ever know for sure what that means until it hits the DAC. We can run various models to try to estimate, and sometimes even come close, but "true peak" metering just isn't fair.

I suspect dude did something stupid to your file and that made it go over, but we don't actually have enough info to do more than speculate wildly.

 

[mjbphotos]

SRC=sample rate conversion.

It could simply be a difference in your meters. An "over" is actually just several full scale samples in a row. The number of samples needed to register on over differs from meter to meter.

The answer to the problem is to have the guy complaining about the over turn that song down by 0.3db and then both of you can get on with your life.

That's the ticket!

I kind of thought that I could get away with abreviating a term that I had used earlier in the paragraph...

You really do need to know what is happening and why if you're ever going to render anything again. One of my biggest pet peeves is limiters that don't actually limit! If I can't know exactly where my loudest peaks are going to be then I'm fucked and the "limiter" sucks.

Reaper's peak meters are dead-nuts-on by default. True, there's only one decimal place precision, but that can't possibly explain this issue. Now, if you've messed with the meter preferences... And the meters indicate exactly what the levels in the rendered file will be as long as you actually render what the meters are indicating. If the Master meter hits -0.3, and you render the Master Mix to the same sample rate as the project and go to an uncompressed format, the resulting file will peak at -0.3. Well, ok, somewhere between -0.25 and -0.34... Even if you can't trust the limiter, you can trust the meters.

...unless you've got a bunch of "live" randomization running on the mix. "Humanized" drum plugins, random oscillators, or whatever are kind of not supposed to do the same thing twice. Many kind of actually do play the same "random" sequence every time you rewind and hit play, but you can't always count on them. So, if you haven't previously rendered those things to stems, you can't really know for sure what might happen when you render it. Course the limiter should catch that, and the render meter should show it.

It is possible for a file that peaks at -0.3dbfs to push "inter-sample peaks" beyond 0, but we can't ever know for sure what that means until it hits the DAC. We can run various models to try to estimate, and sometimes even come close, but "true peak" metering just isn't fair.

I suspect dude did something stupid to your file and that made it go over, but we don't actually have enough info to do more than speculate wildly.

Yeah, I figured Reaper meters should be half decent, even if only 1 decimal place. Nothing 'live' going on in the mix - it's a rendered-to-stereo file. I 'suspect' the same thing, as it was the only song that 'seemed' to have that problem.

 

[Farview]

I just thought of something... If you have mastered the tracks and limited them up to 0.3dbfs, what is this guy's job? If he is trying to master a file that you have already limited, that's a problem.

If he is just a duplicator, not being red book compliant shouldn't be a big deal for anyone using equipment less that 12 years old. Any CD over 74 minutes isn't red book compliant, but I have a couple dozen of those... I've probably got a hundred commercial CDs that are clipped to death, it obviously doesn't stop the replication process.

I'm not sure what the actual problem is.