Digital Overs in Limiter 6GE "True Peak" Mode?

[Winstontaneous]

Today I rendered 2 tracks in Ableton Live 10.0.5 using Limiter 6GE (64 Bit AU v.1.1.0) at the end of the chain, with Ableton master fader at 0.0. Though I had the L6GE Output module set to -0.3 ceiling in True Peak mode, when I ran Amplitude Statistics in Adobe Audition CC I was getting overs up to +0.5. I was rendering to MP3 CBR320, Normalize off, directly in Ableton if that makes a difference from WAV/AIFF.

I don't have any reason to doubt the accuracy of Audition (the Amp Stats for 13 tracks I got professionally mastered matched with the ME's figures 1:1), and I don't have such issues with ToneBoosters Barricade 4, Ableton Limiter, or Voxengo Elephant 4. With these, what I set is what I get when running Amplitude Statistics.

Bug? User error? Thanks!

[sleepcircle]

mp3s are notoriously bad for overshoots.

idk if that has anything to do with it, but i know some mastering engineers who say -1db of headroom is a lot safer for mp3 audio.

[heavymetalmixer]

The reason why masterin engineers recommend never passing over -1dB True Peak on the WAV/AIFF/FLAC is because of two reasons:

1) Those peaks do distort in real life.

2) After converting those files to lossy (like MP3) the peaks go higher, so they always clip.

[FabienTDR]

MP3 is a lossy format, it changes peaks in a rather unpredictable manner (it adds distortion, i.e. stuff that wasn't there before). lossy encoding/decoding usually needs additional safety room, around 1dB (which is also what EBU r128 recommends).

EDIT: sorry, didn't update the thread before answering. Great answers above of course!

[Winstontaneous]

Thanks y'all, that's what I thought. I typically render to WAV/AIFF unless sharing works in progress with my bandmate, which I was doing today.

So in that case, normalizing the rendered file would mean that the clipped waveform is maintained, just shifted downwards by however much I specify? In my case (making MP3 quick tracks to share) should I just click on the "Normalize" option?

Thanks, I have a pretty good grasp of digital audio fundamentals, but there's always more to learn!

[heavymetalmixer]

Thanks y'all, that's what I thought. I typically render to WAV/AIFF unless sharing works in progress with my bandmate, which I was doing today.

So in that case, normalizing the rendered file would mean that the clipped waveform is maintained, just shifted downwards by however much I specify? In my case (making MP3 quick tracks to share) should I just click on the "Normalize" option?

Thanks, I have a pretty good grasp of digital audio fundamentals, but there's always more to learn!

I don't know what the "Normalize" function in that program does, but yeah, "normalizing" or even just turning down a rendered file won't take away the clipping if it was rendered that way.

[sleepcircle]

Well, if it WASN'T rendered that way, and is merely a result of the mp3 decoding processing putting a non-clipped segment a little too close to -0db, you might actually be able to get away with it. I'm not absolutely sure, of course, but it's worth a shot.

[mcbpete]

Thanks y'all, that's what I thought. I typically render to WAV/AIFF unless sharing works in progress with my bandmate, which I was doing today.

So in that case, normalizing the rendered file would mean that the clipped waveform is maintained, just shifted downwards by however much I specify? In my case (making MP3 quick tracks to share) should I just click on the "Normalize" option?

If you're going to convert to mp3 to share the file I'd first normalise the WAV/AIFF to -1db peak, then convert to mp3.

[Hermetech Mastering]

Or just run the MP3 through MP3Gain after conversion, I've been doing that the last fifteen years.

http://mp3gain.sourceforge.net/