Led Zeppelin

Just what the name says.
User avatar
krabapple
Posts: 1615
Joined: Wed Apr 16, 2003 4:19 pm

Postby krabapple » Sun Aug 14, 2005 2:37 pm

Phil Elliott wrote:Either that, or it's me ... I'm the only one that can hear Keith Moon laughing on the "Who Collection" mix of WGFA.

Must be me :)


Surely any true dropout is visible on the waveform?
"I recommend that you delete the Rancid Snakepit" - Grant

User avatar
krabapple
Posts: 1615
Joined: Wed Apr 16, 2003 4:19 pm

Postby krabapple » Sun Aug 14, 2005 2:55 pm

Andreas wrote:
krabapple wrote:For IV I've only looked at Four Sticks, but the fade there looks actually a little bit longer on the remaster

That's true in this case,but the difference is only about one second. On the other hand, Going To California is four seconds shorter on the remaster, and I mean four seconds of music.


Interesting. I'll have to give that a look-see, as well as look for 'dropouts' that Phil mentioned in various tracks.

You are right....what does it mean? Why would they filter at 21 KHz?


Don't know...an early AD transfer, where they weren't confident of the
'brickwall' filtering at 22 kHz, maybe? I'm just talking out of my ass here.

D y'all think it's better to compare files *after* they've been normalized, or does it make more sense to compare them at 'native' levels?

After normalization, of course. But the average volume or the volumes of some sections can still be different.


Which is why I'm tending now towards using level matching based on replaygain numbers, rather than normalization. Replaygain is intended to make the level of tracks perceptually similar.

Even so, in a remaster that has had both band-limited boost or cut applied (i.e., EQ) *and* global boost or cut (e.g., normalization) how could one differentiate the contributions of each? Assuming of course that even more confusing changes weren't made -- like a 'global' boost/cut of just one section of a track.

The only situation I can think of where one can confidently assert that EQ changes weren't contributing to a remaster, is when the ONLY difference in a frequency profile is a consistent level difference across the audible frequency spectrum. This shows up as a straight line above or or below '0 dB' in the sorts of graphs I make.


One thing that was interesting was how 'smooth' the graph is for Communication Breakdown -- that is, the new vs old differences from
one frequency bin to the next are minor, resulting in a smooth curve -- whereas Four Sticks, like many remasters, is quite 'jagged' in its difference graph.
"I recommend that you delete the Rancid Snakepit" - Grant

Andreas
Posts: 535
Joined: Wed Jan 07, 2004 2:41 am

Postby Andreas » Sun Aug 14, 2005 3:32 pm

krabapple,

could you elaborate what kind of graphs you are creating, how you do it, and could you post one of those here? (I do not have any "remastering" software, just primitve wave editors).

Phil Elliott
Posts: 128
Joined: Fri Aug 08, 2003 3:26 pm

Postby Phil Elliott » Sun Aug 14, 2005 3:54 pm

krabapple wrote:
Phil Elliott wrote:Either that, or it's me ... I'm the only one that can hear Keith Moon laughing on the "Who Collection" mix of WGFA.

Must be me :)


Surely any true dropout is visible on the waveform?


Maybe what I class as a drop out is better refered to as "tape defect". You never lose the sound altogether, but levels keep drifting, treble keeps dipping, that kind of thing. I guess that's not going to show up on a waveform...
"If you knew what I was thinking you'd BE me."

User avatar
lukpac
Top Dog and Sellout
Posts: 4591
Joined: Wed Apr 02, 2003 11:51 pm
Location: Madison, WI
Contact:

Postby lukpac » Sun Aug 14, 2005 9:25 pm

Phil Elliott wrote:After the channels were swapped, I discovered that there was a one sample offset between the two channels. I brought the right hand channel forward by one sample on my disc. After doing this, with the exception of 3 errors, everything else was identical.


Interesting...another to add to the list. Same with the SH Buddy Holly disc (some copies). And some original mix Quadrophenia CDs have a one sample offset, but the channels aren't reversed.

The TEAC CD-ROM drive I used to use originally had this problem. It read the wrong channel first or something, so the channels would be reversed. A firmware upgrade fixed that.
"I know because it is impossible for a tape to hold the compression levels of these treble boosted MFSL's like Something/Anything. The metal particulate on the tape would shatter and all you'd hear is distortion if even that." - VD

Andreas
Posts: 535
Joined: Wed Jan 07, 2004 2:41 am

Postby Andreas » Mon Aug 15, 2005 6:17 am

Phil Elliott wrote:Maybe what I class as a drop out is better refered to as "tape defect". You never lose the sound altogether, but levels keep drifting, treble keeps dipping, that kind of thing. I guess that's not going to show up on a waveform...

That raises the question if those "anomalies" are already on the master tape. They could have "fixed" it on the remaster.

Phil Elliott
Posts: 128
Joined: Fri Aug 08, 2003 3:26 pm

Postby Phil Elliott » Mon Aug 15, 2005 11:03 am

Hmm ... given the often flaky stereo imaging of the old disc, with one channel being brighter than the other in many places, that gives them a lot of work to do in remastering.

Why would the original have all these defects? It really sounds to me like a lot of head clogging is going on...
"If you knew what I was thinking you'd BE me."

User avatar
krabapple
Posts: 1615
Joined: Wed Apr 16, 2003 4:19 pm

Postby krabapple » Mon Aug 15, 2005 1:14 pm

Andreas wrote:krabapple,

could you elaborate what kind of graphs you are creating, how you do it, and could you post one of those here? (I do not have any "remastering" software, just primitve wave editors).


For an example of a graph, see this link I posted for the Communication Breakdown comparison.

http://www.m-ideas.com/sullivan/audition/CB.doc

To make them, I have Cool Edit/Audition run a 'frequency analysis' of the track, at default settings (which I think analyses the level every 40 Hz or so from 0 Hz to 22 kHz but don't quote me on that as I 'm not at my home computer). Then 'copy to clipboard' , which captures the numerical level values (in dB below full scale) as three columns of data: frequency/ left channel/right channel. Paste to Excel. Do this for both tracks being compared. Then for each frequency (and for each channel) , calculate the difference between the levels on the new and old versions of the track. Convert this set of difference values (two columns, one for each channel) into a graph with standard Excel tools, with the frequencies as the x-axis labels.

The result is a graph with two curves, one for each channel. The curves show the *difference in level* betweem the two versions of the CD track (calculated as 'newer' minus 'older') for each channel, at each of the frequencies sampled. A solid straight line at 0 dB would mean there is no EQ or level difference between the two versions in either channel.
Last edited by krabapple on Mon Aug 15, 2005 1:18 pm, edited 1 time in total.
"I recommend that you delete the Rancid Snakepit" - Grant

User avatar
krabapple
Posts: 1615
Joined: Wed Apr 16, 2003 4:19 pm

Postby krabapple » Mon Aug 15, 2005 1:16 pm

Phil Elliott wrote:
krabapple wrote:
Phil Elliott wrote:Either that, or it's me ... I'm the only one that can hear Keith Moon laughing on the "Who Collection" mix of WGFA.

Must be me :)


Surely any true dropout is visible on the waveform?


Maybe what I class as a drop out is better refered to as "tape defect". You never lose the sound altogether, but levels keep drifting, treble keeps dipping, that kind of thing. I guess that's not going to show up on a waveform...


It would definitely show up in a waveform, though you would more likely be able to spot it in a spectral analysis.
"I recommend that you delete the Rancid Snakepit" - Grant

Andreas
Posts: 535
Joined: Wed Jan 07, 2004 2:41 am

Postby Andreas » Mon Aug 15, 2005 1:34 pm

Re: Drop-outs on the old LZ IV:

I haven't seen anything in the waveform or spectral analysis, although I don't exactly know what I should look for.

User avatar
lukpac
Top Dog and Sellout
Posts: 4591
Joined: Wed Apr 02, 2003 11:51 pm
Location: Madison, WI
Contact:

Postby lukpac » Mon Aug 15, 2005 1:35 pm

krabapple wrote:It would definitely show up in a waveform, though you would more likely be able to spot it in a spectral analysis.


A dropout, yes. A treble "dip" - probably not. Certainly not easily.
"I know because it is impossible for a tape to hold the compression levels of these treble boosted MFSL's like Something/Anything. The metal particulate on the tape would shatter and all you'd hear is distortion if even that." - VD

User avatar
krabapple
Posts: 1615
Joined: Wed Apr 16, 2003 4:19 pm

Postby krabapple » Mon Aug 15, 2005 10:22 pm

lukpac wrote:
krabapple wrote:It would definitely show up in a waveform, though you would more likely be able to spot it in a spectral analysis.


A dropout, yes. A treble "dip" - probably not. Certainly not easily.


Depends on the depth of the 'dip'. Certainly a marked change in high-frequency energy can be visible in a spectral view. If it's very short, you'd have to know roughly where to look.

Andreas,
A dropout looks like a 'notch' in a spectral view, and like a contraction or flattening of the waveform, in wav view.
"I recommend that you delete the Rancid Snakepit" - Grant

User avatar
krabapple
Posts: 1615
Joined: Wed Apr 16, 2003 4:19 pm

Postby krabapple » Mon Aug 15, 2005 11:15 pm

Phil Elliott wrote:Channels are REVERSED on my LZ IV disc, harmonica on right channel in "When The Levee Breaks", etc ...


Definitely in the left channel on my copy of the 'nonremaster' (Atlantic A2-19129)



Dropouts:


Stairway To Heaven:

5.28 - left channel drops out briefly


Good call! There is a high-frequncy (5 kHz on up) 'dropout' readily visible in the left channel at 5:28:06 , using spectral view. Not a full dropout since it doesn't go down to silence, nor does it span the whole spectrum, but definitely a drop in levels up there beyond 5k for a bit less than a second. The right channel looks normal in comparison. Here's a screen cap of the area:


Image


Going To California
1.54 - left channel drops briefly


Not sure I see anything truly dropout-like in the left channel. I do see what may be a tape edit...the view brielfly becomes weird-looking in both channels at 1:53:25 and most of the track goes quiet. Are you talking about a moment *just before* Plant sings 'throw me a line..." ? Or do you mean something *during* his vox? The strange looking bit happens before his vocals come in. There's something possibly dropout-like soon after, but it is actually happening in both channels, and then another soona fter that where the drop is morein the left than right (at the word 'if'). Hard to tell if this is in the mix, or a flaw. At relatively low resolution, the weird area (center) looks like this:


Image
"I recommend that you delete the Rancid Snakepit" - Grant

User avatar
krabapple
Posts: 1615
Joined: Wed Apr 16, 2003 4:19 pm

Postby krabapple » Tue Aug 16, 2005 12:10 am

btw, Stairway looks phase-inverted in both channels compared to the remaster. The >5kHz dropout is still there at the same place in the left channel in the remaster, masked perhaps a little since levels are about 3 dB higher overall.


For Going to Cali on the remaster, both channels phase-inverted again; left and right channels look more alike at 'if' (because the channels have been balanced more closely than on the original issue), though the streak is still there. The funny curvy bit is also still there.

For remastered Levee, harmonica still in the left channel, channels both phase inverted again.
"I recommend that you delete the Rancid Snakepit" - Grant

Phil Elliott
Posts: 128
Joined: Fri Aug 08, 2003 3:26 pm

Postby Phil Elliott » Tue Aug 16, 2005 11:59 am

krabapple wrote:
Phil Elliott wrote:Dropouts:


Stairway To Heaven:

5.28 - left channel drops out briefly


Good call! There is a high-frequncy (5 kHz on up) 'dropout' readily visible in the left channel at 5:28:06 , using spectral view.


For a moment there I thought you were looking at something else - my post says it is on the left, which would be your right (with me so far?). I made a mistake back there - it is actually on my right.

krabapple wrote:
Going To California
1.54 - left channel drops briefly


Not sure I see anything truly dropout-like in the left channel. I do see what may be a tape edit...the view brielfly becomes weird-looking in both channels at 1:53:25 and most of the track goes quiet. Are you talking about a moment *just before* Plant sings 'throw me a line..." ? Or do you mean something *during* his vox?


During the vox, on the word "line". It's subtle, but it's the kind of thing I seem to notice without even trying :roll:

krabapple wrote:The strange looking bit happens before his vocals come in. There's something possibly dropout-like soon after, but it is actually happening in both channels, and then another soona fter that where the drop is morein the left than right (at the word 'if'). Hard to tell if this is in the mix, or a flaw. At relatively low resolution, the weird area (center) looks like this:


Image


I loaded that into Cool Edit so I could get the same spectral view - while it looks the same to me (albeit with the channels swapped), all I'm hearing at that point is guitar strings being damped and some fret sqeak, and a mandolin note sailing in over the top. That's probably enough to draw some strange patterns though ...
"If you knew what I was thinking you'd BE me."