Hopefully Steve gets to do Aja for the new Audio Fidelity gold series; that would make these threads unnecessary.
Here's the log Mal posted: Seems like track 6 of Mal's disc is 86.2 instead of the 86.1 posted in the graphical chart earlier in the thread. Bizarrely, everything else matches...
Really? I wasn't aware of that. Well, then maybe someone can explain to me how two different engineers, on two separate occasions, but using the same analog tapes came up with two masterings that are digitally in-synch. The odds of this occuring...well, astronomical would maybe just about cover it.
I had just assumed that the two versions were from the same digital submaster (or whatever it is called) but that glass masters for pressing the CD were made on two separate occasions and that the error correction from trasferring off of the tape wasn't perfect, if that makes any sense.
I'm not sure I understand you here swoosh. Are you saying that even if they do not cancel out at a perfect 100% they are still the same mastering?
I hate to raise this thread from the dead as a zombie again and maybe stir up some controversy, but I was able to borrow a copy of the confirmed SH mastering of Aja (the UK presssing) and decided for my own curiosity to compare with my own "almost" copy (Matrix: CI 5650-1). I looked at the song Black Cow since it has a difference in EAC values. I trimmed them both to be in sync (they are a little off from each other) and looked at the first 30 seconds averaged in frequency space. I figured that this is the easiest way to see mastering differences since different masterings should have different equalization. In the graph below I have the two samples compared. The red is my "almost" CD and the green is the confirmed SH version. I made the green 50% transparent on the graph so that it would be dark green where they overlap. It looks to me like the main difference between the two is that they are filtered slightly differently at the high end (and if you look carefully, the low end too). The other differences are on the order of hundredths of a db and if you plot the difference there seem to be some regular, but extremely tiny peaks like maybe a filter or some electronics are ringing slightly. Anyway, it looks like the same mastering to me, but there is some difference in the way they were transferred. As I comparison, I show my "almost" copy, the "bright" DIDX 55 version and the remaster in the same kind of plot. The differences are obvious. Anyway, the two versions of the CD aren't digitally identical and I am not going to say whether anyone has good enough ears to tell the difference between the two (I don't), but they appear to be the same mastering to me. Maybe someone else has another opinion.
I never mind the topic coming up because this is probably the single greatest mystery of the forum, and new information is always interesting, IMO. Thanks putting the effort into those graphs and posting them. Maybe someday someone will get those two CDs to Steve and get his opinion on how they compare.
I just bought a copy of the original Aja 1984 pressing (have no idea who mastered it) and compared it to the 1999 remaster. I know I am late to the party on this because of all the discussions that have already taken place but I just had to post and say man, what a difference! The 1999 remaster is way too bright and shrill!
The Citizen box has noise reduction added. Not as much as a lot of remasters from the same era (Paul McCartney, Genesis, Elvis Costello), but compared to the 1999 CDs or the mid 80s original masterings it's quite obvious. Unless you can't live without the "Everyone's gone to the movies" demo I wouldn't bother with this one if you already have the superior 1999 remasters. The other bonus material (FM, Here at the western world, Bodhisattva live) can be found much cheaper on the Steely Dan Gold CD.
Nice job - can you zoom right in on the 200-1000Hz region (on both if you like) - that will tell us more. Thanks
I can try to do that when I have a break over lunch if you are curious. Until then I'll just say that the biggest difference over the entire 20-20K Hz spectrum is at around 1388Hz with a whopping 0.08 dB difference. Thats why I say its almost certainly different treatments of the same mastering. I'm not sure of what would cause the differences, though.
Sorry for a useless post, but it seems I can't edit my own posts. I just wanted to say that the graph will have to wait until at least tomorrow since the air conditioning is out in the office and its currently over 34C in here. I had to shut down all of our workstations that are not currently running calculations and that data happens to be on one of them.
I'm not exactly sure of what you are looking for, or otherwise I would try to make it easier to see. Here is the area from 200 to 1500 Hz. And here is the difference between the two from 20-20K.
"That's close enough for jazz" as they say Same mastering - for sure. The two discs are clearly made from a different production master but the audio data is originally sourced from the same original A/D transfer (ie "mastering"). Whether the distortion to the original master (Steve's 1630) seen above is audible is the question. To me it seems entirely plausible. That second graph should be a flat line if the audio data matched - but it isn't. The "almost" copy (Matrix: CI 5650-1) isn't the same as the UK version that Steve said was his. My feeling is that the differences up to a maximum of one tenth of a decibel all the way through each song (presumably - have you tested any other songs, jsternbe?) could make an audible change in the "character" of the sound. Digital audio is an exact science - get the sample amplitudes slighlty wrong and the result will be a less accurate representation of the input signal and, particularly in the case of A/D -> D/A, a less convincingly analogue sound at the end of the chain. The lower graph shows that the errors in the sample values on the "almost" CD are largest in the midrange - the very range we are most sensitive too as listeners.... Thanks for the fine analysis, jsternbe
My thoughts are that differences of 0.1dB could be subtley audible, but that these differences are comparable or even small compared to the other components of the playback chain (unless you have a very high-end system that is set up to be as analytic as possible). These difference are also tiny compared to how "flat" speakers reproduce sound. I guess what I am saying is that the only way to really be sure you are hearing the CD with all the right "character" is to play the UK version Steve Hoffman's own system. Anything else is going to raise or lower some frequencies by a fraction of a dB or more compared to what he is hearing. For our own homes, just play whichever one you prefer. They are probably equally close to sounding the way they are supposed to and they both sound really good.
The point is - one will give you the more accurate representation of the "A" from the original A/D on D/A than the other. The one that is closer is the one you want to listen to for the "truth"...
This is true, but you have to be able to reproduce "truth" in the first place. I know that my system isn't nearly analytic enough to make either of the versions sound the way it did when it was mastered. My guess is that when it was mastered it was done on equipment that SH was really familliar with and was meant to sound good on a wide variety of systems. The system it was mastered on is probably not analytic at all and is most likely not colored the way I would prefer anyway. We also can't be sure of which is the more accurate of the two except to go by faith. After all, its the "almost" CD that has a lower noise floor. I do wonder how we got two different versions, though. The best theory I can come up with is that the original digital master was copied. I've never seen a 1630, but I am guessing that it recorded PCM on either beta or Umatic onto its video signal. A tape could either be copied in the analog domain by just using two tape decks, or it can be digitally copied. If its an analog copy there can be some distortion of the pulses which are recorded onto the tape and some bits could be wrong sometimes. If it was digitally copied by decoding and then re-encoding and there isn't some sort of digital time base corrector between the decks then there could be jitter problems between the transmitter and the reciever. I think both would make errors of the order of magnitude we are looking at.
