What Constitutes a "Different" Mastering?

How do they sound? Do they sound the same? I ask because I compared two different CDs (digitally identical except for a level change) recently and found them to sound different. Not a drastic difference, mind you, but the louder one was less warm and sounded a bit more digital to me.

 

Yes they do sound the same. They also look the same in a DAW. If you load in the older track and RMS normalize to match the louder one, they're very close.

 

I assume that normalizing does not affect the sound?

I should also ask that when folks refer to normalizing, do they mean amplifying the quieter one and if so, would doing so using audacity on a MAC be a smart way to do this cleanly?

 

Hi George,

It depends if it is done at 16-bits or longer.
And it depends on the software "engine" doing the processing. I've found different software sounds different when performing the same process.

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

 

Thanks Barry.

So normalizing done at 16 bit affects the sound?

 

Hi Stefan,

Much of this discussion came about in the thread about Target CDs, which were all done in the days of 16-bits (that is, 16 if you're lucky).

IF the example you cite was done on a workstation (hopefully one with better capability than 32-bits, which is nowadays more descriptive of consumer audio software than professional applications, which will do 48, with some processes occurring with 80-bit precision) then there won't be any truncation.

However, there will still be the results of whatever dithering/noise shaping algorithm was used. In addition, my experience is that different software "engines" each create a different sounding result, even when performing the exact same process. The mastering engineer selects his/her tools carefully and for good reasons that will produce the results they are looking for. Other tools produce different results.

I see there are different perspectives on whether this would still constitute the same mastering (in the current marketing or any other sense). To my mind, this is because folks tend to think of mastering as involving EQ and perhaps compression and/or limiting -and a host of other processes in the mastering engineer's bag of tricks.

What many don't seem to consider -and perhaps their thinking is correct for certain mastering engineers- is that level considerations, down to a fraction of a decibel carry equal weight among the decisions the mastering engineer will make. (So can a decision to leave a mix completely as is, though some feel the result is "unmastered", as if mastering means knobs MUST be turned.) When I change a track by a half dB, sometimes less in level, there is a purpose related to making an album a cohesive whole. Sometimes a fraction of a dB level difference can change the apparent timbre. As one mastering engineer I know said, "level is EQ".

So even without considering the losses engendered by plants making level alterations in the old days, changing the level is negating a decision made by the mastering engineer, perhaps several decisions. This is a production call and not a manufacturing call. Any manufacturer that takes it upon themselves to make such a change deserves, in my view, to go out of business.

You can call it the same mastering if you like and refer to the current marketer's definition of what mastering is.

As a listener, if someone changed a master from a given engineer, even in the tiniest way, knowing from personal daily experience just how many decisions are involved in mastering a record properly, I would no longer deem the end product to be that engineer's work.

Put another way, if I wanted Steve's mastering of a given album and I found out a certain release had been tampered with in any fashion after it left his hands (and the approval of his ears), I would look elsewhere, knowing Steve's decisions had been overridden without his knowledge, permission, approval, etc. I want to hear Steve's work, not someone's alteration of Steve's work.

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

 

As Barry mentioned, different software may have different effects on the sound. However, in principle, normalizing done in 32-bit floating point should not "affect" the sound other than to scale the level (up or down). The actual level changing itself shouldn't differ much or at all among various applications. A level change is simply a scaling of a value up or down - 2*2 will always equal 4 no matter what software is doing it. Likewise, properly implemented 32-bit floating point math should be the same no matter what the application or platform (PC or Mac for example). It's what ELSE is done that affects the sound more. How is the sound converted back to a usable format? Hardware plays back audio in 24- or 16-bit integer format. CDs are limited to 16-bit and the conversion to that word length and the dither chosen can have a significant impact on the sound.

 

Do you (or anyone else ) know if audacity does this in 16 bit?

 

Hi George,

It depends on what you mean by "cancel out". Some folks see partial cancellation as cancellation. I don't. To me, when something cancels, nothing is left. Absolutely nothing.

If the definition is partial = cancellation, I would but surprised if a difference was NOT heard. (Partial for cancellation is like a "little pregnant" or as George Piros used to say "a little Dolby". Like the digital code we're talking about, these are binary states. They either is or they ain't.)

If the definition is cancellation = cancellation (i.e. it either cancels or it doesn't), I would expect the files to sound identical. (Note I said "files", referring to listening to extracted files, not the disks they come from.)

Now, all that said, I can see how some folks (and some systems) will not detect a difference if the partial "cancellation" is sufficiently deep. But this is not the same as a difference not existing for other folks and other systems.

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

 

Barry I fully agree with everything you're saying here. My perspective on this and the discussion I was referencing had to do with the discussion of the Led Zeppelin SHM-CDs just released in Japan. I can't recall if it was here or on a LZ mailing list where someone said they couldn't understand why the Japanese went for the 1994 remasters as sources (according to one web site vendor anyway) and not the "2005 Led Zeppelin remasters." There were none in the sense mosy "lay people" perceive "digitally remastered from the original master tapes." In fact, a more accurate sticker would have read "levels digitally altered from previous digital remasters of the original master tapes" or else "digital remasters of digital remasters of the original master tapes." Is this a different mastering that's going to sound noticeably different the way George Marino's sounded noticeably different from your and Joe Sidore's original CDs? I don't think so.

of a new round of digging out the master tapes, sitting down and making all the decisions involved in a major remastering and then

 

Hi Stefan,

Based on some recent experiments I conducted, where I took an original recording and imported it into four different editing/mastering applications, then made a simple level change and saved the results, I would disagree.

There were no edits, no EQ, no fades or trims. Nothing but a level change in each of the programs. Then each file was saved.

Interestingly, when I compared the files using one of the programs for playback, it was very difficult to ascertain any sonic difference. I had distinct impressions but did not feel they were 100% consistent.

However, when I compared the files in one of the other programs, the differences were clearly revealed (small though they were, they were consistent and 100% repeatably identifiable in blind comparison).

What this tells me is the "engine" within the application makes a sonic difference in the results. Even for something so simple as a gain adjustment.

When more processes are involved, such as using my favorite 64-bit EQ plug-in, the differences between the software packages is highlighted even more.

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

 

You'll have to check around the sourceforge pages for technical details, but I believe Audacity processes internally at 32-bit resolution. I know it supports VST plugins (albeit with a patch) and they most definitely process 32-bit data internally. Whether this is converted in Audacity I can't say.

It may be like Cool Edit/Adobe Audition in that the word length of the file determines whether internal conversion occurs. In thes progs, if the currently open file is 16-bit then you have the option of dithering the results of transforms every time they are done as a way of preserving more details. By converting the file to 32-bit floating upon opening (either manually or with the automatic setting), you can ensure the there are no practical losses and no build up of layer upon layer of dither noise.

 

Hi Stefan,

"Noticeably different" in this case, I would say depends on the listener and what they're listening on. I'll agree they won't sound radically different.

But is this George Marino's work? To me, absolutely not.

I agree it would be a "remastering of a remastering", hence, to me at least, not at all the same mastering.

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

 

That's interesting because from the bit od dabbling I did with audio programming, it seems that the actual routines for scaling the 32-bit samples are pretty much static. Yes, how different programs deal with the truncation of data beyond the precision limits of 32-bit math may differ but seriously we're talking about a sound level of -192dB well below the measured limit of -120dB (as I understand it). Yes if the errors accumulated perhaps they'd push up above the audible limits of hearing but one level change means simply taking each sample and multiplying it once by a scaling factor.

It could be explained if the four files sounded the same in each program but different from the same file played in a different program.

You saved these all as 32-bit floating point files?

 

Hi Stefan,

The four files sounded different from each other, regardless of the program I chose to play all four in. However, some programs revealed the differences much more than the others did. But these happen to be the better sounding programs.

To be sure, the differences are not "night and day" but they are there to be heard and are of sufficient magnitude that after the comparison, I changed the tools I use in my work.

This is important for consumer applications, some of which save their temp files at the word length of the original (a good reason to do a "Save As" to 32 float before doing anything else in some applications). It is also true in some higher level apps but not all.

After running several tests, the program I find the best sounding, which processes internally at 48-bits, still sounded better than the others when given a 24-bit file to work with.

By the way, in order to hear the "engine" in each of the apps, with no other influences, I did not apply dither to the results (since I had no need to put them onto a CD). I listened to the full, high-res files at their native 24/96 resolution. (I started a thread about this long ago. If you're interested, I'll place two of the files on my web site -temporarily- so you can compare them and see if you hear a difference with your software/system/ears. Since they are not edited and the piece is if I recall correctly, about seven minutes in length, they are quite large.)

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

 

This can explain why I feel that some of my 87/88 Beatles CD's sound lousy even though I have the same CD's that were either made at a different plant or a different plant and country of origin. Yet I've always been told they couldn't sound differrent because they use the same glass masters world round.
Thank you, thank you for this or these explanations that I would have never of thought to ask about. Although I have asked Steve in the past about the sound of my Japan made Beatles discs. I don't want to quote him so I won't, however if I recall something was said that many Japan discs be it Vinyl or CD had a bit more high end on them, to which I can agree with.

 

The use the same digital masters, but not necessarily the same glass masters. This thread is about CDs that aren't digitally identical, which isn't the case with the Beatles' CDs.

 

Hi Beatle Terr,

I don't know the source of that "information" but I would tend to doubt any company would have a piece of glass sent around the world when they could just send clones of the digital CD master.

There would be no advantage to sending the glass and there would be plenty of risks to its safety. Also, keep in mind, better than 99.99% of record folks will tell you digital "can't" change and all the other "perfect" silliness invented by the marketers.

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

 

My apoligies for using the incorrect term as in (glass master) when I really meant to type (CD master).

 

As I said above, though, this thread has no bearing on that. To my knowledge, various Beatles CDs have *not* had any level changes.

 

Then any differences you hear might be attributable to the processes that occurred in manufacturing: glass cutting (both the particular LBR and the speed of the cut), injection molding cycle (timing), etc.

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

 

Maybe he's referring to pressing differences? Still, that is not the topic of the thread, but it could account for the differences heard.

 

Hey guys,

Interesting thread. I thought this might be fun. I took a song (Baker Street from the DCC gold CD), lowered it 15dB in Pro Tools (with AudioSuite) and then raised it back 15dB to its original level. I just wanted to hear what that would do. I made a clip:

http://www.sendspace.com/file/27n45a

 

Hi Jamie,

I don't have the original.
Did you do this at 16-bits?
I'm assuming you saved after the first level change as well as the second, yes?
If so, I'm curious as to what *you* hear.

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

 

Yeah, 16 bit session and when you AudioSuite something it bounces the changes to the disc so it was saved back to the session twice... three if you count the final bounce to disc.

I hear a big difference in the low end, the top end, the imaging and the overall grittiness of the sound. With the help of my assistant I was able to pick the original nine out of ten times. Pretty drastic changes but a 30dB change is pretty drastic too.