MP3 True Peak higher than Wav on bounce

[Jazzvibe]

Greetings: After bouncing stereo files from 88.2 khz, 24 bit to 44.1 khz, 16 bit I've noticed that the true peak of the wav files is .9 (Oxford limiter set at -1) but the mp3 files bounced simultaneously read .1 true peak (.2 in some cases) according to Youlean Loudness Meter.

Any idea why the MP3 true peak would be higher than the Wav file bounced at the same time? Shouldn't they read the same?

I'll be submitting the Wav files for CD but I'm wondering if there will be a problem when the CD is converted and sent to streaming services. Should it be corrected and if so how?

Protools version 23.6.0.110
2019 Intel 16" MacbookPro Monterey

[Sardi]

Yes. This happens with MP3 encoding. It’s well documented. Lots of info out there about it.

If you’re worried about inter sample peaks and clipping, master to TP -1dBFS.

FYI - if you did an analysis on 99% of modern music, you’d see inter sample peaks exceeding well over 0dBFS.


Sent from my iPhone using Tapatalk

[jeam25]

Originally Posted by Sardi

FYI - if you did an analysis on 99% of modern music, you’d see inter sample peaks exceeding well over 0dBFS.


Sent from my iPhone using Tapatalk

Indeed !

[Ray Lyon]

Helpful info, good question...

[Jazzvibe]

Sardi, thanks for the info. Are you saying to master down even farther for the MP3s to hit -1 True Peak?

I do have the limiter set at -1 and the Wav files are sitting comfortably at .9 True Peak -11 LUFS.

Your FYI suggests I shouldn't worry too much about the MP3s hitting .1 True Peak. Is that correct?

[BScout]

Conversion will change TP (unless it is lossless.)

For a very longtime I've done TP = -1dBFs (on the lossless, source files) as that's what's liked by Apple for their digital masters. Apple asks for that because their conversion does the same thing for peaks and they expect to change their algorithms for compression in the future. That will mean files bought in the future from the store may have a different TP than today's converted files sold.

No point trying to chase a moving target like the TP on the converted file for that reason. Just give enough margin on the source files.

[Jazzvibe]

Originally Posted by BScout

Conversion will change TP (unless it is lossless.)

For a very longtime I've done TP = -1dBFs (on the lossless, source files) as that's what's liked by Apple for their digital masters. Apple asks for that because their conversion does the same thing for peaks and they expect to change their algorithms for compression in the future. That will mean files bought in the future from the store may have a different TP than today's converted files sold.

No point trying to chase a moving target like the TP on the converted file for that reason. Just give enough margin on the source files.

Thanks that's helpful!

[Sardi]

As BScout mentions, there’s no point in trying to get 1dBFS headroom on a lossy file. You’ll just be going back and forth. If you want to avoid inter sample peaks on a lossy encode, use a TP limiter and set the ceiling to at least -1dB. Some go even lower.

To be brutally honest, the whole inter sample peak thing is blow a bit out of proportion. I do a lot of mastering, and sometimes I don’t use a TP limiter as it sounds better with a regular limiter that isn’t TP or I switch the TP off. I will invariably get a few overs, but they don’t really result in distortion that you hear as digital clipping. Granted I’m using high end conversion and monitoring, but even referencing on a myriad of systems with sub standard playback, I’ve yet to go ‘woah, that sounds clipped.’

The other thing to think about is the distribution of said files. If it’s primarily on streaming sites, they’re all loudness normalised on playback. Whilst they have different targets, I’d wager that unless you have an incredibly dynamic mix/master, your song is going to be turned down and thus never played back at its theoretical limit.


Sent from my iPhone using Tapatalk

[Jazzvibe]

Originally Posted by Sardi

As BScout mentions, there’s no point in trying to get 1dBFS headroom on a lossy file. You’ll just be going back and forth. If you want to avoid inter sample peaks on a lossy encode, use a TP limiter and set the ceiling to at least -1dB. Some go even lower.

The other thing to think about is the distribution of said files. If it’s primarily on streaming sites, they’re all loudness normalised on playback. Whilst they have different targets, I’d wager that unless you have an incredibly dynamic mix/master, your song is going to be turned down and thus never played back at its theoretical limit.

Sent from my iPhone using Tapatalk

Part of the confusion I (and probably others) have is that the wav files I'm creating are targeted for CD manufacture. But those files are then used by digital distribution services (i.e. CD baby) to create files that are subsequently sent to all the streaming services. I'm not sure what process they use to create the files they send out, if over samples will be created at that point, and or if it's a problem to be concerned about. It only came up in my mind because I noticed that the MP3 files were peaking .8 db higher than my wav files when bouncing down. Thanks again for the info.

[BScout]

Originally Posted by Jazzvibe

Part of the confusion I (and probably others) have is that the wav files I'm creating are targeted for CD manufacture. But those files are then used by digital distribution services (i.e. CD baby) to create files that are subsequently sent to all the streaming services. I'm not sure what process they use to create the files they send out, if over samples will be created at that point, and or if it's a problem to be concerned about. It only came up in my mind because I noticed that the MP3 files were peaking .8 db higher than my wav files when bouncing down. Thanks again for the info.

Most online stores only accept uncompressed waves. The stores themselves make the compressed version. So a distributor like CD baby has no idea/control.
Apple (with their "Digital Masters" and all their other previous names for that group) has a qualification process for mastering engineers (at least they used to; I've been one from the beginning at the request of Sony Music) where there are specific tools to use to check. They also give a descriptor for high resolution files and what their future plans are (at least by phone call.)
Their compression codec has a TP limiter built-in. But it's still better not to hit that (and it is Apple's request) so -1dBFs for TP.
In the grand old days of CDs, plenty of really famous CDs clipped above TP (MJ's Thriller is a rather famous one.) Sony and their oversampling players tried to help that. But being above TP is nothing new. (And you'll find plenty of tracks in online stores that are above TP limit.)