EAC peak values no guarantee of identical mastering

Since a lot of folks here use the peak values from EAC as a way to compare disc issues in an effort to determine whether they are from the same master, I thought a recent experience might be of interest.

In the past few days, I’ve completed mastering for a project I’ve been working on. As I always do, I keep notes which include the peak values as well as RMS values for each track, at each stage of the process (e.g. originals, after EQ, after level adjustments, etc.).

The mixes for this project were done in such a way that I found a number of tracks needing gain (level) adjustment within the track, as opposed to of the whole track as it fits into the album. There were songs where the one of the verses just sounded a bit low (or a bit loud or a bit different EQ-wise, etc.) compared to the rest of the song. In a number of these, I had to make changes that proved significant when listening to the album.

What some may find interesting is that I could make an appreciable level change to one verse in the song -or part of a verse - and if this part of the track did not contain the max peak (which might have occurred during a chorus), the resulting peak level was exactly the same (to 2 decimal places, that’s a hundredth of a dB) as the version without the level change.

Similarly, I made EQ changes to some sections that were different from the EQ applied to the song as a whole. As long as these didn’t occur in the part of the track where the max peak occurred, the resulting peak level was again, exactly the same (to the same 2 decimal places) as the version without the EQ change in that section.

I hope it is clear from this that while different EAC peak values seem a good indicator of different masters, identical EAC peak values are no guarantee at all the two files were sourced from the same master. The best they suggest is a definite maybe.

The only way I know to truly prove the data is identical is a well performed null test, where the result is a true null (i.e. dead silence all the way down to the lowest levels).

***

I should add that while peak levels remained exactly the same, RMS levels did show a change, sometimes as small as a hundredth of a dB but a difference nonetheless.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

Good point, Barry, and of course it makes sense, considering what the EAC peak levels are! That being said, it's hard to think of a situation where, when the peak levels match for an entire commercially pressed disc (as opposed to the experiment you conducted), that you'd have anything but an identical copy. In fact, I'd almost say such a scenario, if verified (exactly matching peak levels across two discs, but different masterings) would be worthy of a prize!

Also, apart from EAC peak levels, you can also simply compare the files bit-for-bit. Rip two tracks with identical settings, from their respective discs, and compare the size. The number of samples in a track can also be a give-away.

 

I have always assumed it was possible to tinker with EQ and not impact peak values as measured by EAC. Good (or may be bad) to see this confirmed.

 

Hi Modern_Mannequin,

The values of the pressed discs will not be different from the values on the master sent to the replication facility.

What my post says is that I can make changes to the files that result in appreciable level and/or EQ differences without impacting the peak levels at all.

In fact, if peak levels were the criterion for making the determination, I could find the peak in each track and radically alter the whole track except for the tiny fraction of a second that includes the peak. Those using EAC peak levels to make their determination would quite erroneously declare the two identical.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

Well...
http://stevehoffman.tv/forums/showthread.php?p=1251749&highlight=eac+mastering#post1251749
http://stevehoffman.tv/forums/showthread.php?p=3178377#post3178377
http://stevehoffman.tv/forums/showthread.php?p=4576476#post4576476

 

Thank you Barry! I am very far from being the expert that you & others on this forum are. But that said, I've always believed that EAC peak values are nothing more than an indication of the highest peak in the recording...and that sometimes (not always) we place too much importance on these numbers when comparing masterings.

Dexter

 

Perhaps there have been a few, but I don’t see anyone claiming that peak values are a guarantee. Peak values is simply a good tool since we can’t swap discs through the forum (and we don’t have access to the masters) and the likely hood of 2 titles with the same peak values not being the same master is remote at best. Of course, I'm talking about discs which have a range of values, not those with all the peaks smashed at the limit.

You don’t have to use a null comparison to compare discs. Using CRC values to compare discs is an even better means and is far, far more accurate than peak values. This is what the AccurateRip database uses to verify ripped discs.

As an example, CUETools can provide CRC values with and without null to compare different pressings of the same title and even rips with bad offset. Here’s an example below where my disc didn’t match up with what’s in the AccurateRip database, but it is obviously still the exact same mastering because the CRC w/o null is identical. Only the offset is different which can be corrected with CUETools if one wants to.

CSN&Y “American Dream”
(I abbreviated the post to just the 1st 3 tracks)
[Verification date: 1/21/2010 7:13:51 PM]
[Disc ID: 001d5776-0139180b-c50d690e]
Track [ CRC ] Status
01 [d15ab818] (00/26) No matches
02 [d9e908ad] (00/26) No matches
03 [8822c497] (00/26) No matches

Offsetted by -123:
01 [72bf2894] (26/26) Accurately ripped
02 [49f6feaf] (26/26) Accurately ripped
03 [9de919cd] (26/26) Accurately ripped

Track [ CRC32 ] [W/O NULL] [ LOG ]
[6755ECDB] [61D39FB4] W/O NULL
01 [6D2E7340] [9A70AAB7]
02 [B7AE5E9E] [FFAB4356]
03 [C39A7A45] [EE504042]


I wish everyone would use CRC values w/o null in this forum, but I’m guessing that’s a level of detail/accuracy that most folks don’t want to bother with.

 

Hi Scott,

A lot of folks here use EAC peak values in an attempt to make the determination.

"more accurate than peak values" is an understatement since peak values can at best, as I said, provide a definite maybe. If I wanted to create a "secret remaster" I could change night to day and folks checking those peak numbers would never know.

Personally, I'll go with the null test to make determinations of equivalence (or not).

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

Pretty sure there are a couple out there from me as well...

 

And, it's also worth noting that the inverse it true as well - 2 discs can have drastically different peak levels but be 100% identical sans a few handfuls of samples.

 

+1
People should really just post the entire EAC scan-log file, instead of the EAC peak numbers alone. I think the more informed among us always realized the peak numbers were just a lazy shortcut, and should never be used as the final word.

 

I raised this question some years ago when many of us were searching for the elusive Hoffman-mastered version of Aja. Steve said that he made a few EQ adjustments relative to the Nichols mastering. At the time, we were testing a bunch of early pressings and finding them to all be indentical by EAC, yet there was the occasional opinion that one of these pressings was better than the others.

 

Makes sense.

 

"identical by EAC" meaning peak levels, or that the WAVs actually matched when doing a WAV comparison? While peak levels don't tell the whole story, a WAV comparison doesn't lie. Two WAVs can't be identical if the EQ is different.

It seems more likely the "occasional opinion" was due to pressing differences rather than mastering differences.

 

Brilliant as usual Barry I think for myself and maybe some others, we use it more as a tool to show differences in masterings instead of ones that are identical. One thing that is a fact is if two cd's have different EAC peak levels, then they are definitely a different mastering. In the future I will investigate further before I state definitively that two discs with the same EAC peak levels are the same. Great thread Barry!

 

It depends exactly how you want to define "a different mastering".

Like I said, you could adjust a single sample from each track to get different peak levels. While no longer 100% identical, do a handful of different samples across a disc constitute a "different mastering"?

 

It would no longer be identical so yeah in my book it would be different. Really though, practically speaking, do you feel this ever happens, or do you have any evidence of this (your example above)? Most of the time it's "hey I got 2 cd's here that look identical, but this one peaks at 70 and this one peaks at 100". Since we live all over the country and all over the world for that matter, it's a decent way to start a discussion on mastering on individual titles. I'm gonna go out on a limb and say most forum members could listen and hear the difference between two masterings anyway, I know I can. Furthermore, I can honestly say I have never encountered two cd's with identical peak levels on a commercial cd that weren't the same mastering. If they are out there, that should be a thread all on it's own. Interesting thoughts though friend!

 

Also I don't wanna put words in Barry's mouth, but I seem to recall him saying once that whenever he sent a master to a pressing plant to be used for a cd pressing and the plant "tweaked" it, he no longer considered it to be his work. Barry, feel free to correct me on this if I'm mistaken.

 

Peak levels. Sorry for being vague, but I was responding to earlier posts about peak levels.

The "occasional opinions" may have based on nothing real, either.

 

While that's fair enough, the opposite is not true. Disks can have different checksums and EAC levels, yet still the 'same' mastering.


My case in point was Bjork's Homogenic. I always thought the US Elektra was brighter and less 'boomy' than my UK . When I compared carefully, there was still a difference but it wasn't so pronounced as I thought. Very careful level matching gave me a genuine null, and the same number of samples .. when the waveforms were both time aligned. It took a while, but with changes in level and start I showed that the numbers were the same for these two disks.

EAC reports difference in levels, Checksums different, tracks different times.. yet the music is in fact the same.

Doing the same thing on two genuinely different masterings is more instructive. Often you can show straight away that the analog master was running at a different speed (I did two copies of AJA like that once. Kind of trippy hearing them diverge when played together...) If the differences are put in digitally, then typically the number of samples stays the same and the different discs stay time aligned to the last sample of the track, start and end permitting.

 

Hi steeler1979,

As I see it, you have it right.
If something is not exactly the same, by definition, it is different.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

Hi MikeyH,

I would argue that different peak levels constitute a different mastering.
Even if the same source is used for both, to my mind, changing the levels on one makes it different. I certainly wouldn't want my name on a mastering if the levels I sent out were altered along the way; such would no longer represent my work. (If a chef prepares a meal and someone along the way to your table decides to add some extra salt, I wouldn't consider the result the chef's work either. They might have used his work as the source but what is delivered is not what he created.)

The music of course is the same. The original source might even be the same. But what you have with even a single sample changed, is in my view, a different master.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

Very good analogy and I'm with you 100%

 

It's the wrong analogy.

Adding extra salt is like applying an eq change which changes the tonality. Changing the volume is like taking the meal to the oven and warming it up. You have to be careful not to burn the meal (= clipping) or melt the cheese (=peak limiting), but the taste (=sound) doesn't really change. Would the chef disown his meal if the waiter carefully warmed it up?

 

Well, it seems likely that whatever method is employed to shift levels would leave some sort of sonic signature when listened to carefully on good equipment by someone with well-trained ears. And I would certainly defer to Barry as the chef who is having his creation f@#@'d with But as a practical matter, I am sure the vast majority of listeners listening on the vast majority of systems would be unable to hear any difference after adjusting the volume to offset the level change.