Elton John: Goodbye Yellow Brick Road
Universal (Japan) UIGY-9613
Stereo Single Layer
"Goodbye Yellow Brick Road"
Features the 2010 DSD mastering based on Japanese original analog tape. Reissue features the high-fidelity SHM-SACD format (fully compatible with standard SACD player, but it does not play on standard CD players). Comes with a description and lyrics.
Support this site by purchasing from these vendors using the links provided below.
As an Amazon Associate HRAudio.net earns from qualifying purchases.
Click here to report errors or omissions in the music details.
Comment by Brian VanPelt - May 14, 2022 (1 of 17)
Not bad, but it does not hold a candle to the Speakers Corner 1999 vinyl release. The vinyl is more open, with much sharper focus on everything. To listen to the SACD, and then the record, is like going from one version of a hi def TV to the subsequent version. The contest is not really close. Even though I recommend the SACD, the Speakers Corner vinyl is substantially better.
Comment by Paul Hannah - May 15, 2022 (2 of 17)
OMG......anybody care to explain scientifically just how vinyl can sound better ? ? ?
Comment by Mark Werlin - May 15, 2022 (3 of 17)
It wouldn't surprise me if the Speakers Corner LP sounds better than the Universal Japan SACD. Their outreach is to different markets, so you'd expect the releases to sound different. Moreover, Speakers Corner masters from best tape sources. I don't know what the term "Japanese original analog tape" means; it could be a 2nd or 3rd generation tape that was intended for LP production.
Lots of articles can be found online that discuss and compare the real, measurable differences between analogue recording/playback vs. digital.
The author asserts, on scientific measurement principles, that to equal the properties of vinyl playback, a digital signal would need to be sampled and delivered to the listener at 16 bit, 730MHz. That's a very high sampling rate!
See also: https://audiouniversityonline.com/analog-vs-digital-audio/
It comes down to subjective preferences. Some of my favorite analogue-era rock and jazz LPs don't sound as good to me in their newest CD, hi-res PCM or SACD versions as the original vinyl. I don't think that's because of inherent deficiencies in hi-res digital mastering processes, but because some mastering engineers modify the sound of tapes with added EQ, compression/limiting, and normalization. Most of the Universal Japan SACDs I've heard are worthwhile, but Traffic's "John Barleycorn Must Die", for example, sounds much better to me on the original vinyl.
Comment by Ben Leggett - May 20, 2022 (4 of 17)
"The author asserts, on scientific measurement principles, that to equal the properties of vinyl playback, a digital signal would need to be sampled and delivered to the listener at 16 bit, 730MHz. That's a very high sampling rate!"
The author asserts that, but it's nonsense, and not science at all - it ignores Nyquist entirely. CD quality PCM far, far exceeds the dynamic range available to vinyl, and has a much much lower noise floor. CD quality PCM can capture all the detail and range possible with vinyl and then some - it's not even a contest.
Now, sure, some LPs have different *masterings* which people may like better (in fact usually masterings for vinyl have to be compressed and altered in specific ways to account for failings and quirks of the medium - people might prefer that compression or altering, but it's a stretch to say it's more accurate, or higher fidelity), but that has nothing whatsoever to do with the format, and every possible vinyl mastering can be replicated exactly by CD-quality audio with ample headroom.
Comment by Adrian Quanjer - May 23, 2022 (5 of 17)
Now that we are on the subject: By nature, sound is analogue. If we could reproduce it on a carrier ‘as is’, that would be the best. However, as Ben pointed out correctly: “masterings for vinyl have to be compressed and altered in specific ways to account for failings and quirks of the medium”. So, what one gets on vinyl is not the same sound picture. I won’t talk here about wear&tear and needle drop&skid accidents, but what Ben says is my view as well.
I once owned an old-fashioned Mercury vinyl recording with Antal Dorati conducting the Minneapolis Symphony in Strauss’ Symphonic Poems amongst which Don Juan. The engineers, trying to maintain as much as possible the original dynamics resulted in my needle jumping out of the groove on several occasions. The amplitude was too much! Probably because of my modest Japanese turntable. But to get sufficient material pressed on each side (more grooves, less amplitude) led to the commercial necessity of compressing files.
The CD format solved some of the above deficiencies, but the digitalized sound did not please all of us. To mine and many other ears, it failed to give the warmth of the analogue sound. Whatever the technicalities of remastering old master tapes (of which I know little indeed, and it would seem that some do it better or worse than others), for me, newly recorded pure DSD SACD’s -preferably in multi-channel- do recreate much of the warmth of the analogue sound.
Comment by Athenaeus - May 24, 2022 (6 of 17)
My experience roughly accords with what Mark Werlin wrote. I've had the opportunity to compare the same recordings on vinyl and optical disc a few times. I've found that vinyl generally sounds better when the recording is jazz or rock; but when it's classical music the optical disc usually sounds better. It seems to have something to do with the typical acoustic properties of each musical genre versus the sonic advantages and drawbacks of each format. (These are generalizations and, of course, there are exceptions!) I realize that the vinyl and optical disc often use different masterings, which does make comparisons difficult. I have tried to take that into account. In any case, I shouldn't be hearing a genre-based difference if it came down to mastering differences. I have a feeling I'm not the only person who, consciously or unconsciously, is hearing a similar difference. I don't have any numbers to back this up but I've noticed that jazz-lovers tend to gravitate towards vinyl, whereas classical folks lean towards digital.
In audio, I listen to the science but I do trust my ears more than the science. The history of audio has shown that's the wise thing to do. Remember how, in the early days of digital audio, differences that people were hearing but that shouldn't be there were dismissed with the phrase "bits is bits"? After some research, it was discovered that audiophiles weren't hallucinating; the differences they were hearing were due to jitter. There are still things we don't understand about digital audio; and audio in general. One shouldn't forget that.
But I don't want to get involved in a technical discussion. I have no expertise whatsoever. I'm just a guy who likes to listen to music, preferably in high-fidelity sound.
It's funny but the other day I bought the SACD version of the Beach Boys album "Surfin' USA". I already had Analogue Productions' vinyl re-release of the mono version of this album. But I wanted AP's SACD version because digital is more practical for me. When I listened to the SACD version, I was disappointed. I'm used to the vinyl version, which has a clearer, fuller, more lifelike sound. The SACD sounds good but not as good, unfortunately.
In this case, it's important to note that SACD and vinyl are using the same mastering! I was listening to the mono tracks on the SACD. Plus, my digital system is superior to my analogue system. I don't even have a dedicated phono stage. I plug my trusty Rega turntable straight into my integrated amplifier. In other words, my analogue system doesn't have a qualitative advantage. Unless the engineers purposefully made the digital version sound poorer than the vinyl pressing, at least in this case, vinyl beats SACD. According to my humble ears, that is. It so happens this is a rock album.
Comment by John Broggio - May 31, 2022 (7 of 17)
I'm with Adrian.
I'd summarise my position as thinking that genuinely hi-res media are wonderful at revealing all the glories of a recording but also any shortcomings.
Other media that are not hi-res (and I would include vinyl in this statement) may be incapable of being as accurate to the same source material and so unable to reveal those shortcomings or glories properly, which can be (mis)interpreted as sounding "better" in the first case and as worse in the second.
Comment by hiredfox - June 9, 2022 (8 of 17)
The sounds I hear in the concert hall every week are by far best reproduced on vinyl despite the obvious limitations, deficiencies and annoyances of the format. That is all there is to it really. No digital format has been able to deliver the throatiness of string/wood instruments, the ring of brass or the honk of woodwind. DSD SACD comes closest but not close enough although I love SACD for its airiness and separation. Sit with me and you will agree.
Comment by Brian VanPelt - June 15, 2022 (9 of 17)
I made a 128k DSD copy of the Speakers Corner Record. Is it allowed to be shared legally, in part or whole? I would love other ears to hear it.
Comment by Mark Werlin - June 16, 2022 (10 of 17)
Brian VanPelt: please email me at mwerlin@HRaudio.net. There are HRAudio users (myself included) who have DSD128-capable DACs and computer audio servers, and who would be interested in hearing your transfer of the LP.
Comment by Brian VanPelt - June 17, 2022 (11 of 17)
You all might be interested in the Misty for Direct Cutting recordings. Tsuyoshi Yamamoto and his group performed this last year, but they also recorded simultaneously in digital and analog. The digital was recorded at 11.2 MHz DSD, with no PCM anywhere in sight. The analog was recorded direct to disc (actually, the lacquer was cut while the band played).
To me, 11.2 MHz vs direct to disc seems like an ultimate battle of analog vs digital. I have both and I say that they are just about perfect, but they are different.
In line with others in this thread, the all digital one sounded light and airy, and it is incredibly pleasant. Additionally, the tone and weight of the piano sounded lifelike; I know because I have sat and listened to pianos close up in Jazz joints and such.
On the other hand, the direct to disc felt warmer (surface noise, albeit was almost non-existent, surely adds to that weighted feel). The tone and weight of the piano sounded lifelike.
They are equal, but different.
Comment by Mark Werlin - June 19, 2022 (12 of 17)
Having listened to tracks from Brian Van Pelt's DSD128 transfer of the Speakers Corner LP in comparison to the same tracks from the SHM-SACD, I agree with Brian's assessment. The Speakers Corner mastering produces a more enjoyable listening experience for me than the DSD mastering.
I appreciate the artistry that engineers bring to the remastering process. If you listen closely to a jazz or rock reissue on SACD and the same title in a different mastering in hi-res PCM or 180-gram LP, you can hear the different approaches. The engineers at Speakers Corner and at Universal Japan took different routes. Some may prefer the SHM-SACD, but that doesn't devalue the experiences of those of us who prefer the LP.
Comment by Brian VanPelt - June 20, 2022 (13 of 17)
I'm glad this has spawned such a discussion. Let me add some things. First, I'll start with Nyquist's Theorem.
Nyquist's Theorem involves any generic function of period p. You have to get 2 independent measurements in one period to reproduce the function. So it it with the sine function. Given 2 independent samples, or two independent points, within a period, and the function (or graph) can be completely reconstructed. No problems there. In fact, given a sine function
f(t) = a sin (bt),
you can see that I need 2 independent t values to determine both a and b. Thus, with 2 t values, any tone f(t) = a sin (bt) can be perfectly reconstructed. However, the curve a computer draws is always, and I mean always, an approximation of the actual curve in real life. Suppose you have a complex event where multiple tones are registered at the same time. Each tone has the form f(t) = a sin (bt), but the composite of those tones is a Fourier series. No matter what, the computer must approximate with a finite Fourier series, and each function within is also approximated.
Check this out. While coding for an RSA encryption algorithm, I made use of the program's "mod" function (short for modular arithmetic). When I wrote my program, it kept crapping out because the numbers it encountered were too large for the program to run - even though I set those variables to the maximum amount of accuracy possible (I dimmed them as Long). Now, I was ablto to make my program run simply by writing the long division algorithm myself (jettisoning the built-in mod function). All of the sudden, the program worked perfectly when I, conceviably, made no difference in the program at all - if the computer worked perfectly, that mod function would have worked as needed, but it didn't.
Furthermore, in pure mathematics, there is the concept of measure, and it deals with how many items are in a set. On the real line, the measure of the interval (0, 2pi) is 2pi, equal to its length. This matches the idea of continuity, which means that there cannot possible be any gaps in the real number line. The measure of the interval [0, 5] is 5, the measure of [5, 7) is 2, the measure of (a, b] is b - a; you get the idea. All of those sets had positive measure. However, any discrete set, no matter how many items it has, has measure 0.
So, mathematically, there is a guargantuan difference between the interval 0 Hz to 20,000 Hz, and any amount of approximations between 0 Hz and 20,000 Hz, no matter how accurate a computer, will never captured absolutely everything heard. But a tape will.
Comment by Ben Leggett - July 1, 2022 (14 of 17)
"I'm glad this has spawned such a discussion. Let me add some things. First, I'll start with Nyquist's Theorem."
"So, mathematically, there is a guargantuan difference between the interval 0 Hz to 20,000 Hz, and any amount of approximations between 0 Hz and 20,000 Hz, no matter how accurate a computer, will never captured absolutely everything heard. But a tape will."
Your conclusion here is simply wrong, and exactly backwards from what the Nyquist Theorem actually indicates.
"The Nyquist theorem specifies that a sinuisoidal function in time or distance can be regenerated with *no loss of information* as long as it is sampled at a frequency greater than or equal to twice per cycle." (emphasis mine)
Literally, the Nyquist theorem tells us that digital music *absolutely can* capture *absolutely every* ounce of information on an analog tape and reproduce it *absolutely perfectly*, well past the human audible range into the extreme ultrasonics, if the sample rate is high enough. There's an exceptionally clear explainer on the math here: https://www.youtube.com/watch?v=cIQ9IXSUzuM that categorically debunks the myth that digital "can't capture or reproduce analog signals perfectly".
This is, in fact, why storing digital music is possible at all in the first place. If this was not what Nyquist meant, none of your CDs or FLACs would work.
I'll say it again, so there's no confusion. The Nyquist theorem gives us a *mathematical proof* that digital music *absolutely can* capture *absolutely every* ounce of information on an analog tape and reproduce it *absolutely perfectly*, well past the human audible range into the extreme ultrasonics, if the sample rate is high enough.
(And for the curious, the question of "what sample rate is high enough to perfectly capture and reproduce everything on analog tape up to the limit of human hearing, excluding reproduction of ultrasonics which we can't hear", the answer that the Nyquist formula spits out if you plug in the numbers is 44.1Khz, which is (not coincidentally) exactly the RBCD sampling rate)
Comment by Brian VanPelt - July 1, 2022 (15 of 17)
I think the last commenter misunderstood my conclusion. I did say that the Nyquist Theorem is correct, that's why we call it a theorem! I hope we are clear on this point. I clearly indicated this when I said a single tone can be represented as a sine curve f(t) = a sin bt , where a, b are constants that need to be determine. You'll never guess how many pieces of data that takes - 2 pieces of data inside of one cycle. See, 2 data pieces gives 2 equations, and you can only solve for a, b reliably when you have 2 pieces of data; exactly as Nyquist's Theorem indicates. Did I say something mathematically that was incorrect there? If so, what did I say mathematically that was wrong?
Before you snip, I believe every single thing the Nyquist Theorem says. I am not arguing the Nyquist Theorem. Are we clear on this point?
All of my questions, every single one of them, are whether the computer can handle the data fed to it acuarately. I tried to explain that when the same computer file is read in more than one way, you can expect more than one output. This is why I believe I can hear the difference in a DSD 64 vs an 88.2 kHz, 24 bit download. It's also why I believe that a DSD 64 file sounds more true to the source (to me, not you) than an 88.2 KHz, 24 bit hi-res download does. One file is streamed at 1 bit, and the other is more arrayed; and I believe that these two file types .dsf, .flac sound differnt and it lies in how the computer porcesses each file type. How a computer processes a file is something I've seen a little, but I am no expert there. That's what has driven me to find out why I prefer .dsf over .flac.
Then there are different masterings, which I'm sure that Nyquist's Theorem handles, so please tell me how it accounts for different masterings and why they sound different - or do they all sound the same? Because, you know, Nyquist.
You know what else, there currently is no digital Goodbye Yellow Brick Road that touches the record I referenced above - noy even close and you would agree if you heard it.
Comment by John Bacon-Shone - July 3, 2022 (16 of 17)
Mastering is a completely different issue, in this case, the frequency response is usually different and there may be different levels of compression, so this is a red herring.
Even standard DSD is equivalent to a much higher frequency limit than 88k PCM, so again Nyquist is not directly relevant.
Vinyl may sound better because it adds 2nd harmonic distortion, which is often pleasing to the ear, so this has nothing to do with Nyquist.
Digital may sound worse because of jitter, which breaks a key assumption of Nyquist.
We all have different hearing thanks to variation in ears, eardrums etc. (and different deterioration with age), so we may each prefer different things.
I think most people on this website prefer higher resolution and more channels, all else remaining equal (especially good quality recording and mastering). I am very grateful to all the musicians, engineers and companies that support this.
Comment by Mark Werlin - July 4, 2022 (17 of 17)
The recent post by John Bacon-Shone adds welcome clarity to a number of points raised this discussion thread.
Because Brian Van Pelt's post #13 dated June 20 raises a mathematical problem that I am not qualified to asses, I forwarded it to a mathematician friend of mine. Here is my friend's response:
"Essentially [Brian Van Pelt] is making the case that continuous capture of data cannot be replaced by sampled discrete data, no matter how frequently data is sampled. By truncating infinite sums of continuous values, the approximation can be distorted.
The following analogy is a bit of a digression, but e.g. you can ask does the square root of 2 really exist? It would necessitate an infinite string of decimal digits to identify this, i.e you can't come up with a number in finite time that when you square it yields exactly 2. So then no such number exists. But that would prove troublesome because there would be a gap in the real number line which would then sabotage the process of arithmetic and analysis. So the solution is to make it an axiom: yes, such a number exists although we can't identify it numerically (essentially ALL irrational numbers, those that can't be expressed as a fraction or finitely repeating decimal e.g. 4.675675675 ... and so forth) cannot be identified numerically as such a decimal.
The analogy to Nyquist is that it is assumed that a sufficient quantity of sampled data can represent a continuous/infinite collection of data. This then is similar to the square root of 2 business: mandate or insist that the discrete sum of all data actually = the continuous sum, or otherwise accept that a sufficient sampling is a valid approximation of the continuous sum within a tolerance of accuracy that is acceptable."
I've posted this communication in the hope that some readers will find it useful. I find mathematical concepts are often easier to grasp through analogy.
Those who contribute reviews and comments to this website, as John Bacon-Shone points out, are here because we appreciate the quality of high-resolution and DSD recordings. I am very grateful for the expertise and artistry that go into fine audio recording and audio remastering.