Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Curious find: mastering "affects one channel much more" (Read 1753 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Curious find: mastering "affects one channel much more"

First: please no "you cannot hear the difference", that is not the question. Rather: does anyone know any tools that would lead to this phenomenon, and why?

Back in the day I bought a few CD collections and it so happens that I have more than one of the same release ripped. And every now and then I ran a bitcompare. Sometimes the differences are way more than you would expect from redithering, and sometimes they differ systematically across channels. Take this - figures after offset alignment, track by track on the same CD:
Code: [Select]
Channel difference peaks: 0.027008 (-31.37 dBTP) 0.198303 (-14.05 dBTP)
Channel difference peaks: 0.009918 (-40.07 dBTP) 0.013458 (-37.42 dBTP)
Channel difference peaks: 0.000153 (-76.33 dBTP) 0.161407 (-15.84 dBTP)
Channel difference peaks: 0.017609 (-35.09 dBTP) 0.094757 (-20.47 dBTP)
Channel difference peaks: 0.005737 (-44.83 dBTP) 0.030212 (-30.40 dBTP)
Channel difference peaks: 0.015350 (-36.28 dBTP) 0.167877 (-15.50 dBTP)
Channel difference peaks: 0.051544 (-25.76 dBTP) 0.051575 (-25.75 dBTP)
Channel difference peaks: 0.012756 (-37.89 dBTP) 0.299591 (-10.47 dBTP)
Channel difference peaks: 0.034271 (-29.30 dBTP) 0.071747 (-22.88 dBTP)
Channel difference peaks: 0.020752 (-33.66 dBTP) 0.100250 (-19.98 dBTP)
Channel difference peaks: 0.008881 (-41.03 dBTP) 0.036041 (-28.86 dBTP)
Difference always greater in the left channel. All tracks, although for one (track 7) only by the minimal amount - and up to a sizeable -76 vs -15.
(Lazily I sanity-checked them with the whatever-it-is-worth Tau Analyzer, which differs in assessment for two tracks: One is track 7 where the difference shows up near-equal over channels, so maybe that is part of the explanation; then for the second to last track, Tau cannot assess more than one of them.)

Any tool that is known to do things like this upon mastering?
Uneducated guesses: Some "enhance stereo separation" filter activated but set to "minimum" which isn't completely the same as switched off? Or, deciding on dither shape by an analysis of only one channel?

Re: Curious find: mastering "affects one channel much more"

Reply #1 –
Sorry, but I don't understand what the numbers are showing...

If you have two differently-mastered copies of the same recording there's no reason for the left & right to be altered/mastered differently unless it's some kind of synthesized "fake stereo" or some kind of stereo widening.  

That kind of stereo processing would be rare with re-mastering and different mastering engineers would do it differently so the differences should be random over a large number of recordings.

Quote
And every now and then I ran a bitcompare.
With a different master virtually EVERY byte/sample will be different.  If you just change the volume, all of the samples change.   If it's a limited number of different samples, that's probably errors in the data.

Re: Curious find: mastering "affects one channel much more"

Reply #2 –
Figures are from the output of fb2k's bitcompare function, difference between pressing #1 and pressing #2 per track.
"Channel difference peaks" means, for example, that for track 1, the biggest difference between pressing1track1left and pressing2track1left is at -31.37 dB, while comparing the right channels the largest difference is only -14.05 dB.

There is no noticeable "remaster" going on in the sense of changes in volume or dynamic range, but it is very common indeed that different pressings do differ - they are not from exactly the same file, so some kind of processing has happened at the "UK" pressing plant and at the "NL" pressing plant. Sometimes for an entire album you have no sample differing in more than -90 dB, sometimes the differences are way bigger. Sometimes, some "repairs" have quietly been done without much else being different (like if track boundaries were off).
 

Re: Curious find: mastering "affects one channel much more"

Reply #3 –
It would be interesting and informative to see how these differences look on spectrogram when one file is subtracted from another.

Re: Curious find: mastering "affects one channel much more"

Reply #4 –
Quote from: Porcus on
"Channel difference peaks" means, for example, that for track 1, the biggest difference between pressing1track1left and pressing2track1left is at -31.37 dB, while comparing the right channels the largest difference is only -14.05 dB.
The difference peak is much higher for the right channel, not lower.

Quote from: Porcus on
There is no noticeable "remaster" going on in the sense of changes in volume or dynamic range,
Are these different transfers of the same analog master? Tape speed slowly drifting over time could explain what you're seeing.

Re: Curious find: mastering "affects one channel much more"

Reply #5 –
Quote from: Octocontrabass on
Are these different transfers of the same analog master? Tape speed slowly drifting over time could explain what you're seeing.
Likely transfers from the same master. This particular album is from 1991. Is it even realistic that peaks - I mean the worst possible sample - are so low then? They seem reasonably peak-normalized, peaking at 0.940735 to 1 (always the same over the two rips). 
And there being a difference between channels (yes thx, of course the other channel around) would then follow, nothing says the two reading heads are (aligned) the same.

Re: Curious find: mastering "affects one channel much more"

Reply #6 –
Quote from: Porcus on
Is it even realistic that peaks - I mean the worst possible sample - are so low then?
At least in my limited experience, the equipment engineers use to digitize master tapes is very good at maintaining a consistent speed. I could see two different transfers of the same master tape being responsible for the small differences.

If tape speed is responsible, you could hear it by subtracting one copy of the CD from the other after sample-alignment. It'll slowly get quieter as the two recordings drift into sync, and louder as they drift apart.