All Things Must Pass -- really in 24/96?

regnaD kciN · « **on:** 22 Aug 2011, 06:28 pm »

I notice there's a free download of Wah-Wah on the GeorgeHarrison.com website, which is claimed to be from the 24/96 version, and is a large-enough WAV -- 185 MB -- to suggest it really is the high-res version. I checked it out in Audacity (1.3+, which I understand to be the non-buggy release). Here's the screenshot:

Considering the complete lack of energy above 20 kHz, it sure looks to me like this is upsampled Redbook. Does anyone here have the complete album, and have you checked to see if other songs on the album display the same phenomenon?

ted_b · « **Reply #1 on:** 22 Aug 2011, 07:02 pm »

Yes, they look the same. The album doesn't sound all that great, either, so I am concluding it ain't worth it. It's likely either upsampled redbook or a fractionally better 24/44 or 24/48 master...but the addtl bandwidth/dynamic range of 24 bit does it no real good to my ears IMO. YMMV. The only real advantage might be for DACs that play better at 24/96k than 16/44, but that's a stretch.

regnaD kciN · « **Reply #2 on:** 22 Aug 2011, 07:06 pm »

Thanks -- it looks like you just saved me $29.95!

srb · « **Reply #3 on:** 22 Aug 2011, 07:15 pm »

My audition conclusion is that it sounds worse than any of my small collection of 128kb/s MP3s. Really quite awful.

Steve

regnaD kciN · « **Reply #4 on:** 22 Aug 2011, 08:09 pm »

The more I think about this, the more troubling it gets. Wasn't this re-master done by the same Apple team that did the 09.09.09 Beatles re-releases? You'd think they'd have better ethics than to put out upsampled Redbook as 24/96 hi-res. And it thus makes me wonder about Sir Paul's reissues...are they genuine 24/96, or faux hi-res like this?

Phil A · « **Reply #5 on:** 22 Aug 2011, 09:35 pm »

I just rewired the network cables yesterday and listened to the Squeezebox Touch, including bits of All Things Must Pass. It is not the best recording in the world, especially comparing it to Band on the Run, although also not the best I've ever heard, but good (and that cost me $9.99).

firedog · « **Reply #6 on:** 23 Aug 2011, 07:52 am »

My spectrum plot looks different, isn't brickwalled, and has substantial material above 20K. It may depend on what part of the track you sample. Or the download from the site isn't the same version as the one you get when you buy. Or the version people are buying isn't the same one that was being sold when it first came out. Wouldn't be the first time something like that was screwed up on a web site. Or I don't understand the graph - but I think my graph is pretty clear.

The above is from Wah-wah 24/96 in wav format. It's the first 109.2 seconds of the track. I got the same plot when I tried it on the flac version and with different "function" windows in the result. I used Audacity 1.3.

Audacity 1.2 has bugs and will give inconsistent and/or incorrect results plotting spectrograms, especially with hi-res flac.

The album sounded bad from the get go - even in the original. It wasn't so noticeable in 1970 because it's less obvious with the limitations of vinyl, and most of us didn't listen then using audiophile level systems.

Producing with the Phil Spector wall of sound (multiple instruments of each type, background singers, strings ) in those days necessitated using heavy compression in order to get all the tracks mixed onto a master. High background noise, low SNR, etc. In the notes to the 2000 remaster George wrote that he was very tempted to remix the whole thing and eliminate some of the production. A shame he didn't .

I have the 30th anniversary edition (2000) which improves over the original CD. The hi-res version sounds different. Whether you like it or not - a matter of taste. I copied two tracks of the 24/96 - a friend bought the download and I tested them to be sure they were hi-res, but after listening I decided the sound wasn't worth the cash.

rbbert · « **Reply #7 on:** 23 Aug 2011, 01:47 pm »

This album always sounded bad. On the original LP's, even listening in 1970, the hiss level in the audio portion, compared to the "dead space" between tracks, was remarkable. Also as noted, the "Wall of Sound" production was anything but a model of clarity.

I downloaded this early, and that version (checked at the time, although I no longer have those files on my HD) had a spectral analysis like the last one shown here, and Izotope Ozone indicated true 24-bit recording depth. I like the sound better than the 30th Anniversary edition (which now sounds rather harsh and trebly to me), but others may feel differently.

ted_b · « **Reply #8 on:** 23 Aug 2011, 02:42 pm »

Although I normally use Blackman-Harris, the Hanning looks more like regnad's as well. Prolly 24/48 master, but this just isn't a great hirez album IMO.

*Scotty* · « **Reply #9 on:** 25 Aug 2011, 02:53 am »

If you could sample part of a track where a cymbal is struck,depending on the size of the cymbal you should see a strong fundamental around 500Hz and a harmonic at 15kHz that is almost as loud as the fundamental tone. There should also be overtones above 20kHz that are well above -100dB in level in a 24/96 recording.
Another thing that is missing is tape hiss between 8kHz and 12kHz. I think there should be some kind tape hiss noise floor that comes up from the bottom to meet the program signal level. When I have heard tape hiss on a CD I don't think it was at -60dB down from the program as it was about the same loudness level as the music.
Of course some sort of digital magic may have made the tape hiss disappear,in which case ignore my comments about it.
40kHz content at -120dB could be noise that got digitized,I kind of doubt that it is related to musical content in the file.
Scotty

firedog · « **Reply #10 on:** 25 Aug 2011, 08:45 am »

Quote from: *Scotty* on 25 Aug 2011, 02:53 am

Of course some sort of digital magic may have made the tape hiss disappear,in which case ignore my comments about it.
40kHz content at -120dB could be noise that got digitized,I kind of doubt that it is related to musical content in the file.
Scotty

Sure, but that wasn't my point. My point was that my sample isn't brickwalled at 20K and doesn't look like fake hi-res.

rbbert · « **Reply #11 on:** 25 Aug 2011, 12:40 pm »

Many classical and jazz recordings from the '50's and '60's have noticeable high-frequency content up to about 25 kHz. Generally speaking, those had at most 2 analog tape generations on 2 or 3 track reel recorders. "Pop" and rock recordings of the late '60's, however, frequently involved multiple analog tape generations on multi-track recorders, which had poorer frequency response to begin with and much poorer after a few generations. Transferring A>D at 24/96 puts aliasing well outside the audible range, but won't put true high-frequency information where there isn't any to begin with.

*Scotty* · « **Reply #12 on:** 25 Aug 2011, 08:08 pm »

rbbert

Quote

"Pop" and rock recordings of the late '60's, however, frequently involved multiple analog tape generations on multi-track recorders, which had poorer frequency response to begin with and much poorer after a few generations.

Crap there goes any hope of high frequency air and life in any cymbals on the recording.
I guess I was hoping that if I got a good transfer of the analogue master tape into the Hi-Rez digital medium I would hear things that were previously impossible to hear from the old black disc.
In failing to allow for the impact of over-dubbing on the frequency response possible from the final version of the analogue master tape I guess I had worked up a good set of unrealistic expectations.
The effect of over-dubbing would also be a potential explanation for reports of no differences heard between 16/44.1 versions and 24/96 versions of some recordings.
The GIGO rule in force with a vengeance.
Oh Well.
Scotty

regnaD kciN · « **Reply #13 on:** 27 Aug 2011, 04:13 am »

Quote from: firedog on 23 Aug 2011, 07:52 am

My spectrum plot looks different, isn't brickwalled, and has substantial material above 20K. It may depend on what part of the track you sample.

Interestingly, the main difference between your graph and mine is that your plot goes all the way down to -126 dB, while mine stops at -90dB. (How do you get Audacity to generate that window? I can't figure out how to get it below -90.) I notice that, if you cut your graph off at -90, it looks virtually identical to mine. What it all signifies, I can't tell, except that I'd question the notion of "substantial material above 20K," if all said material appears to be at -94 dB or softer.

firedog · « **Reply #14 on:** 27 Aug 2011, 08:54 am »

Quote from: regnaD kciN on 27 Aug 2011, 04:13 am

Interestingly, the main difference between your graph and mine is that your plot goes all the way down to -126 dB, while mine stops at -90dB. (How do you get Audacity to generate that window? I can't figure out how to get it below -90.) I notice that, if you cut your graph of at -90, it looks virtually identical to mine. What it all signifies, I can't tell, except that I'd question the notion of "substantial material above 20K," if all said material appears to be at -94 dB or softer.

Okay, "substantial material up to 25K" if that makes you feel better. But again, hi-res isn't all about frequency response. The greater dynamics also mean that sounds and audio clues (e.g., small differences such as instrument overtones that enable us to tell different instruments apart, or hear the spatial characteristics of the recording venue) that are impossible or difficult to hear with standard resolution are audible with hi-res.