Global Analysis Reporting Peaks below 0.0 dbfs

Could Global Analysis include a threshold of 0.0 dB or +0.0001 dB for Peaks to report (Digital and True Peaks). And the time between peaks (currently 100ms minimum) be made able to reduce to 0.0ms ?

It’s not possible to even roughly equate the number of peaks that Wavelab finds for files to the number of peaks found by the Apple Tools for MFiT, because Wavelab is reporting Peaks at and below 0.0 dbfs (in the case of ISP, “peaks” sometimes 10dB down).

Also, Apple afclip reports every peak above 0.0 db, no matter how close together, but Wavelab is limiting this to 100ms minimum between peaks.

I know it wouldn’t be exactly the same because of the different codecs, but it would be much closer than it is now if a threshold was made available. Right now the counts aren’t even close, so they don’t relate at all to what Apple is counting.

The pictures below are from an actual MFiT song from the iTunes Store. Just one song. Wavelab shows 1190 left channel “peaks”. Apple afclip shows 240 left channel “clip” peaks.


threshold of 0.0 dB or +0.0001 dB for Peaks to report

What do you mean with that?

I’m mean the option to limit the report to only peaks above 0.0 dbfs like the apple tool. As it is “peaks” below 0.0 are also reported and counted in Wavelab.

I guess WaveLab is oversampling and thus reporting reconstructed overs where the samples each side (which are what you can see) are below 0dBfs.

I guess WaveLab is oversampling

Only for the True Peak report.

I’m mean the option to limit the report to only peaks above 0.0 dbfs like the apple tool. As it is “peaks” below 0.0 are also reported and counted in Wavelab.

Indeed, this option would make sense.

But I am more sceptic about a limit-less report. That could lead to memory overloading.
Concerning the gap, I would agree to reduce it, but not till 0.

Thanks for considering this P.G. that’s great.

But, as an example, the Apple tool reports clips in channel 2 at the following seconds locations in the song:
205.421616 secs.
205.421621 secs.
205.421627 secs.

That has the distance between peaks down to at least 0.000005, well under 1 ms.

The Apple peak analysis takes longer (about 5 seconds) than the Wavelab peak-only analysis (about 0.5 seconds), but it’s more complete. And that’s the way they do it, and how they count the number of clips. Sonnox must do it also if they claim complete MFiT analysis compliance. But then again Sonnox is probably just calling the same built-in Mac routines as Apple, including afclip. That’s the only way Sonnox can do it and be current, is to use all the same Mac built in codec, src, and commands as Apple. How to do that is all publically available in the AppleScripts in the MFiT droplets, and in the commands outilined in the MFiT PDF.

205.421616 secs.
205.421621 secs.
205.421627 secs.

A bit hard to read and grasp. What’s the real interest of getting this degree of detail?

There’s no practical use for that degree of detail that I know of, except to get closer to being on the same page as Apple. But that might be worth something. Apple chose to document every clip level when making afclip. I think Audacity does the same. I think it makes sense to make that abililty available. Rather than unavaiable by design.

Audacity does the same, counting every clip no matter how close together, even the next sample. Reading Apple indicates they do the same, counting every clip, and requiring that analysis.