Clipping in au lab, not in WaveLab

WaveLab
#1

Who has used the Mastering for iTunes droplets and au lab on a Mac?

I’m seeing clipping when I play back my 24 bit or 16 bit master using the AU Round Trip AAC plugin, but I don’t see clipping in WaveLab. What’s the deal here? Which application is lying to me?

I understand that no clipping can occur if one is submitting tracks using the Mastered for iTunes process - but if WaveLab is telling me that no clipping is happening, how can I make sure I’m doing this right??

Thanks for any tips/advice on this process.

Johnny

#2

I think more info is needed. It’s normal for peak levels to increase when converting a WAV to AAC. If you don’t leave enough headroom on a mastered WAV file, you might see clipping on the resulting AAC file. It’s normal. It’s very possible for WaveLab to not report any clipping (because there is no clipping), but still to have clipping after the WAV is converted to AAC. This is entire reason to have these testing tools. Peak levels increase on mp3/ACC encoding as well as sample rate conversions. You need to find an output ceiling level that survives the encoding process.

Some people set the output ceiling as low as -1.0dBFS and their master WAV files, and it can be helpful to use a true peak limiter rather than just a digital peak limiter.

The funny thing about Mastered For iTunes if you read the manual is Apple says this:

"iTunes won’t reject a master file based on the number of clips the file contains. This tool is provided so you can make an informed decision about whether to submit an audio file or go back to the drawing board and make adjustments. It’s a creative decision— one that’s entirely up to you."

So while they recommend that your master WAV files don’t result in clipped AAC files, they still allow it.

I think the Apple supplied tools are garbage though, especially the one that runs live like a plugin. You’ll get differing peak results on each playback because the playback is not in sync with the codec buffer.

To get accurate results, you must do it in an offline process like the Sonnox Pro Codec Manager app does. Offline testes will give you repeatable results. Live playback will not.

With the OS X Terminal I believe you can do some offline tests but I have never tried it. There is more info in the MFiT manual about it:
https://images.apple.com/itunes/mastered-for-itunes/docs/mastered_for_itunes.pdf

The main answer I think is that if you’re seeing clipping on the AAC files, you need to lower the output ceiling of the final limiter in WaveLab to leave some headroom.

#3

Hmm. Ok - I did set my UAD Limiter to -0.2 dBFS, but perhaps I need to up that.

If I decode the AAC file, and analyze that in WaveLab, shouldn’t that give me truth?

#4

It all depends on the material, and how loud it’s pushed but -0.2dBFS with the UAD Limiter which to my memory isn’t true-peak aware is probably not low enough.

There are a lot of variables in the question “If I decode the AAC file, and analyze that in WaveLab, shouldn’t that give me truth?”.

Tools like Sonnox Pro Codec can encode the precise iTunes+ AAC codec that is used for Apple Music/iTunes. I would suggest sticking to the provided tools or Sonnox Pro Codec app. WaveLab will decode to 32-bit floating point and too many variables are involved.

Sonnox Pro Codec is a great tool, and the Apple stuff is just OK. Check the OS X Terminal method for true offline encoding/testing if you don’t want to invest in Sonnox Pro Codec.

#5

Thanks Justin -

This helps a ton.

Johnny

#6

No problem. WaveLab also has a built-in “Encoder Checker” plugin but I haven’t really used it. It has a 256 AAC setting but I was at 96k and kept getting errors about sample rate concerns which sort of defeats the purpose of MFiT checking since MFiT allows you to submit 24-bit/96k WAV masters.

#7

It might depend if your monitor so-called “true peaks” or digital peaks.
true peaks are monitored by reconstructing the waveform between samples (this is a kind of up-resampling). The resampling method depends on each application. There could be minor differences while digital peaks (the sample values themselves, don’t involve any calculation).
Anyway, it’s better to be in the safe side, and keep a margin, and never try “to normalize to 0 dB”.

#8

That’s not true. That’s definitely the safest route and what Apple prefers, but officially it’s up to the mastering engineer to decide whether clipping levels are acceptable to be sent to Apple for MFiT. That’s in the official documentation, and it’s borne out by analyzing MFiT tracks from the iTunes store in afclip. There are definitely clipping levels on many MFiT tracks.

But it’s safest to send few or no clip levels, because it’s what Apple prefers, and they won’t question whether the clip levels are causing audible distortion or not.

#9

Sorry, just saw that Justin already said this, “apple will not reject…”. But Apple will probably question if there are thousands and thousands of clip levels.

#10

Cool thanks y’all - I’ve gone back through and fixed everything.

The Normalizer in WaveLab can look at True Peaks as well as digital peaks. I lowered my target output on the limiter to -0.5dBFS, which brought all but a couple of tracks in the right place - and other two, with only a couple of true peaks, I just used the Normalizer and brought them down manually about .2 dBFS.

Seems to have worked.