Why do records sound so much better than digital?

[Robin L]

There is one recording where the (overpriced) 20/44 "re-master" eliminates the hiss at the cost of completely murking the stage of what was an amazing presentation of a classical orchestra on a fantastic stage. Lesson learned - I tend to stick with the originals.

Please tell us which one.

I've heard Decca and Philips do some early attempts at digital noise reduction that ruined the recordings [however, there's a Bruckner 4th led by Furtwängler where the transfer engineer hand-penciled out all the coughs, a heroic early effort], also the stupid EMI de-noising of Arthur Schnabel's Beethoven Piano Sonata set [since rectified, FWIW]. But there's also examples of first-rate work, like the BMG series of RCA's Living Stereo series, also Philip's transfers of the Mercury Living Presence series. Sony's digital restorations have done wonders for George Szell/Cleveland recordings from Epic.

The kinds of things LPs do wrong bother me more than a little compression---at least the pitches will be right, there's no wow and flutter, no disc eccentricity, no gradually devolving styli and discs, no IGD, no surface noise.

 

[pablolie]

Please tell us which one.

Karl Munchinger. A recording that includes his Pachelbel kanon, which in the original is one of my favs despite some hiss. The re-master was disaster. I think the remaster was short-lived. I could not find it right now to provide a link...

 

[Robin L]

Karl Munchinger. A recording that includes his Pachelbel kanon, which in the original is one of my favs despite some hiss. The re-master was disaster. I think the remaster was short-lived. I could not find it right now to provide a link...

Funny, a lot of the early "Weekend Classics" were flat transfers. That was one of the most popular versions of the Kanon [my favorite is the Speedmetal Baroque rendition by Musica Antiqua Koln], and I recall seeing it in numerous remasters. And after checking different YouTube videos I recall that Munchinger recorded the work more than once. I found one needledrop of the original German LP issue, seems to be sharp relative to the other recordings, one is noticeably longer. I remember having that recording on numerous LP issues, also this Weekend Classics early CD transfer:

 

[mocenigo]

Three things come to mind.

1. Better mastering; no loudness wars for vinyl.
2. Euphoric distortion.
3. Psychoacoustics. When I am going to focus on music with intent, I usually pour a drink and put a big wholesome record onto my gorgeous analog system, after taking pride in ownership of the physical medium and spending a moment looking at the artwork in my hand. It sounds way better to my ears than the digital I stream while working.

I love euphoric distortion more than the euphonic type!

 

[krabapple]

Well looking for another thread I ended here, maybe a simple answer is tonality in vinyl. From America ("to each his own" track) a comparison between vinyl first take (AT VM-540 SL-1200) and hd tracksView attachment 190975
Digital provide extra 3-4 dB below 100 Hz, analog was almost 10 dB high at 10 kHz cymbals details are impressive. Same peak level (normalized) and same loudness range (untouched) due to same analog master source, digital sounds dull at best

You aren't comparing just vinyl to digital here. You are comparing 1) masterings and 2) a flat FR playback technology to a nonflat FR playback tech. You can't tease out the contributions of 1 vs 2 from just these data.

 

[Rock Rabbit]

You aren't comparing just vinyl to digital here. You are comparing 1) masterings and 2) a flat FR playback technology to a nonflat FR playback tech. You can't tease out the contributions of 1 vs 2 from just these data.

Both comes from same old master mix and have very similar loudness and RMS statistics. But there's an MM capsule that add a non linear response. So I take another good album from Tracy Chapman (1988) to make same comparison between CD and Vinyl (clean, no clicks) and curiously all loudness (BS1770) statistics are equal (-1 dB track normalized), so we can better compare frequency response

Upper picture is CD and almost identical to vinyl FR so it seems same master to me.
Now it's clear that HR digital track from America is completely wrong in high frequency, probably old backed tape or tape recorder with issues on read head.
I have repeated this experiment with many old records and always the digital version is seriously compromised if not completely mutilated by digital processing, only exception is a hi res Grand Funk recording that fortunately was transferred with a good tape machine and without any DSP, it's possible to see on FFT real tape capabilities without Dolby or brutal compression with bass information that don't fit on the restricted groove modulation of vinyl.

 

[dlaloum]

The

Both comes from same old master mix and have very similar loudness and RMS statistics. But there's an MM capsule that add a non linear response. So I take another good album from Tracy Chapman (1988) to make same comparison between CD and Vinyl (clean, no clicks) and curiously all loudness (BS1770) statistics are equal (-1 dB track normalized), so we can better compare frequency response
View attachment 191329
View attachment 191331
Upper picture is CD and almost identical to vinyl FR so it seems same master to me.
Now it's clear that HR digital track from America is completely wrong in high frequency, probably old backed tape or tape recorder with issues on read head.
I have repeated this experiment with many old records and always the digital version is seriously compromised if not completely mutilated by digital processing, only exception is a hi res Grand Funk recording that fortunately was transferred with a good tape machine and without any DSP, it's possible to see on FFT real tape capabilities without Dolby or brutal compression with bass information that don't fit on the restricted groove modulation of vinyl.

The FR of any cartridge is driven first by the effective tip mass of the stylus - which is mostly driven by the cantilever mass - this will in turn generate a resonance - with this type of stylus, most likely around 12 to 14kHz - so the bare bones response before any processing, coming from the cartridge will have a bell shaped resonance curve around 13kHz or thereabouts.

Then the circuit formed by Cartridge inductance, and load impedance and capacitance, will adjust this, and provide its own EQ - this is typically used to control the cantilever resonance and achieve a flat F/R in combination with the RIAA EQ.

The problem with all this is - you need to know what the cartridge and stylus are, and then whether the precise parameters required for a flat F/R have been met. How many people actually know their effective C load?... it's not just the phono stage, but the cabling in the arm, and the cabling leading to the phono stage + whatever load is applied within the phono stage.

Those running with MC cartridges can pretty much disregard capacitance issues, as due to the tiny inductance of the coils on MC's capacitance has completely negligible impact (yeah you can run 1000pf and it doesn't care). - The downside being, that the raw cantilever resonance cannot be "cancelled out" - and tends to get exposed.... (why do so many MC's have a "bright", Rising, high end... making them sound more "detailed"...)

But lots of excellent turntables are then ruined by people putting heavy, audiophile looking interconnects on them... unfortunately most of the heavy audiophile interconnects have high to very high capacitance - not unusual to see 450pf.... whereas the thin and ordinary looking cables connected to many 1980's turntables, have a total capacitance of under 100pf.

Specifications for the AT VM540L are that it should be loaded with between 100-200pf (and by the way that range would imply 2 to 3db of potential difference in high frequency response at least... ie: between 100pf and 200pf there can be quite a difference).

If you have proper low C cabling - the cabling alone will take you to 100pf - lots of phono pre's default to around 200pf internal C load.... which puts the poor cartridge at 300pf and well outside of its specified loading.

What I am trying to say - if you want to make this sort of comparison, you need to know in detail the setup that was used for the Vinyl replay, and it needs to have been measured and confirmed for a flat Frequency Response - otherwise it is not meaningful.

With a properly measured setup, digital EQ can potentially also be applied to adjust as needed to achieve the optimal flat F/R... (Technics new TOTL integrated, provides exactly that function, with a test record provided with the integrated for testing and automated EQ... - but for the rest of us, this is a very manual process)

 

[Brinkman]

Both sides of a record are sequenced for technical reasons and well a how well they flow together. Album artwork is often better when reproduced at its original intended scale. I know this is essentially LARPing but I enjoy experiencing music from the vinyl era on vinyl for the those reasons. Lots of music originally issued on compact disc works better in that format for similar reasons. If I really love a release, I will often purchase it in both formats just to experience it in various ways.

 

[dougi]

The

The FR of any cartridge is driven first by the effective tip mass of the stylus - which is mostly driven by the cantilever mass - this will in turn generate a resonance - with this type of stylus, most likely around 12 to 14kHz - so the bare bones response before any processing, coming from the cartridge will have a bell shaped resonance curve around 13kHz or thereabouts.

Then the circuit formed by Cartridge inductance, and load impedance and capacitance, will adjust this, and provide its own EQ - this is typically used to control the cantilever resonance and achieve a flat F/R in combination with the RIAA EQ.

The problem with all this is - you need to know what the cartridge and stylus are, and then whether the precise parameters required for a flat F/R have been met. How many people actually know their effective C load?... it's not just the phono stage, but the cabling in the arm, and the cabling leading to the phono stage + whatever load is applied within the phono stage.

Those running with MC cartridges can pretty much disregard capacitance issues, as due to the tiny inductance of the coils on MC's capacitance has completely negligible impact (yeah you can run 1000pf and it doesn't care). - The downside being, that the raw cantilever resonance cannot be "cancelled out" - and tends to get exposed.... (why do so many MC's have a "bright", Rising, high end... making them sound more "detailed"...)

But lots of excellent turntables are then ruined by people putting heavy, audiophile looking interconnects on them... unfortunately most of the heavy audiophile interconnects have high to very high capacitance - not unusual to see 450pf.... whereas the thin and ordinary looking cables connected to many 1980's turntables, have a total capacitance of under 100pf.

Specifications for the AT VM540L are that it should be loaded with between 100-200pf (and by the way that range would imply 2 to 3db of potential difference in high frequency response at least... ie: between 100pf and 200pf there can be quite a difference).

If you have proper low C cabling - the cabling alone will take you to 100pf - lots of phono pre's default to around 200pf internal C load.... which puts the poor cartridge at 300pf and well outside of its specified loading.

What I am trying to say - if you want to make this sort of comparison, you need to know in detail the setup that was used for the Vinyl replay, and it needs to have been measured and confirmed for a flat Frequency Response - otherwise it is not meaningful.

With a properly measured setup, digital EQ can potentially also be applied to adjust as needed to achieve the optimal flat F/R... (Technics new TOTL integrated, provides exactly that function, with a test record provided with the integrated for testing and automated EQ... - but for the rest of us, this is a very manual process)

Yes I have the VM540ML and run it into a phono pre with variable loading (50pF lowest) and use the lowest C cables I could easily find. (RG6-ish) type from local electronics store for cheap. From my measurements the lower the C (including the arm wire) the better for HF flatness. I still then add digital EQ with the RME ADI-2 pro.

 

[Rock Rabbit]

The

The FR of any cartridge is driven first by the effective tip mass of the stylus - which is mostly driven by the cantilever mass - this will in turn generate a resonance - with this type of stylus, most likely around 12 to 14kHz - so the bare bones response before any processing, coming from the cartridge will have a bell shaped resonance curve around 13kHz or thereabouts.

Then the circuit formed by Cartridge inductance, and load impedance and capacitance, will adjust this, and provide its own EQ - this is typically used to control the cantilever resonance and achieve a flat F/R in combination with the RIAA EQ.

The problem with all this is - you need to know what the cartridge and stylus are, and then whether the precise parameters required for a flat F/R have been met. How many people actually know their effective C load?... it's not just the phono stage, but the cabling in the arm, and the cabling leading to the phono stage + whatever load is applied within the phono stage.

Those running with MC cartridges can pretty much disregard capacitance issues, as due to the tiny inductance of the coils on MC's capacitance has completely negligible impact (yeah you can run 1000pf and it doesn't care). - The downside being, that the raw cantilever resonance cannot be "cancelled out" - and tends to get exposed.... (why do so many MC's have a "bright", Rising, high end... making them sound more "detailed"...)

But lots of excellent turntables are then ruined by people putting heavy, audiophile looking interconnects on them... unfortunately most of the heavy audiophile interconnects have high to very high capacitance - not unusual to see 450pf.... whereas the thin and ordinary looking cables connected to many 1980's turntables, have a total capacitance of under 100pf.

Specifications for the AT VM540L are that it should be loaded with between 100-200pf (and by the way that range would imply 2 to 3db of potential difference in high frequency response at least... ie: between 100pf and 200pf there can be quite a difference).

If you have proper low C cabling - the cabling alone will take you to 100pf - lots of phono pre's default to around 200pf internal C load.... which puts the poor cartridge at 300pf and well outside of its specified loading.

What I am trying to say - if you want to make this sort of comparison, you need to know in detail the setup that was used for the Vinyl replay, and it needs to have been measured and confirmed for a flat Frequency Response - otherwise it is not meaningful.

With a properly measured setup, digital EQ can potentially also be applied to adjust as needed to achieve the optimal flat F/R... (Technics new TOTL integrated, provides exactly that function, with a test record provided with the integrated for testing and automated EQ... - but for the rest of us, this is a very manual process)

Thanks for the info, the 540 is probably loaded by 47k and some 150 pF (2' Sommer cable), I compared with 5' cable and no difference in FR, adding extra 150 pF and start to loose HF response. But I don't try to make lab measurements ...I only use a 0.000x thd interface and that's enough for comparative analysis , for FR compliance is the Chapman album comparison that yield same FR (between 2/3 dB) and same loudness statistics (a clean album don't rise DR). That album demonstrate that digital and analog versions could be the same thing at ear except for noise floor (min RMS -50 dB on vinyl). But when the loudness (BS1770) is -16 to -12 dB and loudness range is 6 dB or less music "fits" very well on vinyl media and it seems that old audio engineers have very good ear to make this happen (btw streaming use same loudness target).
Obviously in the digital domain the numbers could be impressive but we are still using our limited ears where dynamic range over 16 dB or separation over 40 dB, or noise floor below 70 dB is absolutely meaningless (at normal 70-80 dBSPL).

Obs.: from Amir (Cambridge duo review), ear sensitivity at 120 dB SPL!...we are not better than this curve

 

[Pogre]

Do they really tho..?

 

[Snoopy]

Do they really tho..?

They Dont, but this topic would be dead long time without asking the question like that.

 

[krabapple]

Thanks for the info, the 540 is probably loaded by 47k and some 150 pF (2' Sommer cable), I compared with 5' cable and no difference in FR, adding extra 150 pF and start to loose HF response. But I don't try to make lab measurements ...I only use a 0.000x thd interface and that's enough for comparative analysis , for FR compliance is the Chapman album comparison that yield same FR (between 2/3 dB) and same loudness statistics (a clean album don't rise DR). That album demonstrate that digital and analog versions could be the same thing at ear except for noise floor (min RMS -50 dB on vinyl). But when the loudness (BS1770) is -16 to -12 dB and loudness range is 6 dB or less music "fits" very well on vinyl media and it seems that old audio engineers have very good ear to make this happen (btw streaming use same loudness target).
Obviously in the digital domain the numbers could be impressive but we are still using our limited ears where dynamic range over 16 dB or separation over 40 dB, or noise floor below 70 dB is absolutely meaningless (at normal 70-80 dBSPL).

Obs.: from Amir (Cambridge duo review), ear sensitivity at 120 dB SPL!...we are not better than this curve

The digital audio revolution was driven initially by recording engineers of 'classical' music works, where dynamic range can top 96dB. Whatever you think of the adequacy of vinyl, *they* were certainly not satisfied with it.

Btw, the Tracy Chapman album is considered an 'audiophile' reference recording, often trotted out to test hi-end gear in the usual silly magazines . I would hardly take it as typical. And you have no sure knowledge of the provenance of any of these masterings. You don't know what was done or not done to them to achieve the final measured performance. And NB, differences of as little as 0.5 dB in the mid-treble can be audible. Add to that the deficiencies of vinyl that your graphs do NOT show - including crosstalk, rumble, speed inaccuracy, distortion that varies with radius, and of course the noise floor -- your case seems strained. Vinyl is amazing for what it can do given what it is, but it's not as high-fidelity as digital. That's simply a fact. Some people prefer its sound. That's also a fact.

Let me refer you to a thread about this very topic from years ago on Hydrogenaudio (another science-based audio forum). Here is some straight talk about vinyl distortion that could actually be considered sympathetic. From JJ (James Johnston, whom you should consider authoritative), post #49

Surface noise is required by physical processes. You can get rid of the gross noise due to damage, but surface noise is unavoidable.

Vinyl, furthermore, can easily be shown to have a particular set of distortion mechanisms that increase the loudness (using the term correctly) by much more than one would expect by the increase in intensity.

What is more, Vinyl almost always (system dependent but only usually on how much) will add L-R signal in midrange, and will also add L-R rumble.

The distortion and noise mechanisms in vinyl can actually sound better than the clean signal, sometimes, for very clean vinyl and playback.

I didn't say "accurate" I said "sounds better" as in "many people prefer".

The background noise on vinyl also, of course, often functions as "comfort noise". Very few concert halls are that quiet, I know, I've measured one or two myself...

and later, post #101

These issues surrounding perceived dynamic range in no way endorse or validate the claims of LP as 'more accurate' or "having more information" both of which are measureable and testable, and which are simply clearly not true. Any kind of analytic accuracy issue is easily resolved, CD is better. LP is worse. There is no dispute on this, assuming we accept the usual 20-20kHz bandwidth for human audio perception, which I think is reasonable and completely justified in the case of teens or adults. The ONLY possible analytic dispute is in bandwidth, and I, like most here, question its relevance.

The issue is purely in the realms of acoustics (in terms of missing information in a stereo signal) and psychoacoustics in the human auditory system. Some people do in fact prefer LP, and that is not something that can be said to be right or wrong, unless and until somebody comes along and starts with the "more accurate", etc, claims. Such claims can be very, very wrong. On the other hand, L-R enhancement from stylus flexure, hinging, etc, varying distortion mechanisms, etc, are documented and well understood, and do result in perceptual effects. Certainly, they are inaccurate, but if someone prefers them, so be it.

 

[Frgirard]

The 96db was a demand of the Sony President wife.
Decca in the 70's recorded in 18bit.
The first dac as Linn was in 14 bit.

 

[sq225917]

If you want another matched cd and vinyl master the 1st press of American iv, Johnny cash is as close as I've heard between sources.

 

[dlaloum]

Thanks for the info, the 540 is probably loaded by 47k and some 150 pF (2' Sommer cable), I compared with 5' cable and no difference in FR, adding extra 150 pF and start to loose HF response. But I don't try to make lab measurements ...I only use a 0.000x thd interface and that's enough for comparative analysis , for FR compliance is the Chapman album comparison that yield same FR (between 2/3 dB) and same loudness statistics (a clean album don't rise DR). That album demonstrate that digital and analog versions could be the same thing at ear except for noise floor (min RMS -50 dB on vinyl). But when the loudness (BS1770) is -16 to -12 dB and loudness range is 6 dB or less music "fits" very well on vinyl media and it seems that old audio engineers have very good ear to make this happen (btw streaming use same loudness target).
Obviously in the digital domain the numbers could be impressive but we are still using our limited ears where dynamic range over 16 dB or separation over 40 dB, or noise floor below 70 dB is absolutely meaningless (at normal 70-80 dBSPL).
View attachment 191400
Obs.: from Amir (Cambridge duo review), ear sensitivity at 120 dB SPL!...we are not better than this curve

Yes, it seems the better and easier the audio tools have become, the more careless much of the mastering has become - many of the records in their original untampered (NOT remastered) state, sound superior to any of the later "remasterings" using gee whiz digital tech...

 

[dlaloum]

The 96db was a demand of the Sony President wife.
Decca in the 70's recorded in 18bit.
The first dac as Linn was in 14 bit.

The first releases from Philips - a leader back in the day - were 14 bit... I had a 14 bit philips chip Revox CD player, which around 1984/85, I had back to back compared to all the other high end players I could find - it was superior to most (all! the ones that I heard...) 16 bit implementations for years....

I sold it to a new home about 5 or 6 years back - and even then it was holding its own

Nothing wrong with well implemented 14 bit.. (quad oversampling.... etc...)

At the same time, Sony was time splitting a single dac between two channels, yeah it was 16 bit, but cheaping out like that did not do good things for the sound.

 

[Frgirard]

Someones succeed abx between 14 and 16 bit. The Linn GM has failed it.

 

[Rock Rabbit]

The digital audio revolution was driven initially by recording engineers of 'classical' music works, where dynamic range can top 96dB. Whatever you think of the adequacy of vinyl, *they* were certainly not satisfied with it.

Btw, the Tracy Chapman album is considered an 'audiophile' reference recording, often trotted out to test hi-end gear in the usual silly magazines . I would hardly take it as typical. And you have no sure knowledge of the provenance of any of these masterings. You don't know what was done or not done to them to achieve the final measured performance. And NB, differences of as little as 0.5 dB in the mid-treble can be audible. Add to that the deficiencies of vinyl that your graphs do NOT show - including crosstalk, rumble, speed inaccuracy, distortion that varies with radius, and of course the noise floor -- your case seems strained. Vinyl is amazing for what it can do given what it is, but it's not as high-fidelity as digital. That's simply a fact. Some people prefer its sound. That's also a fact.

Let me refer you to a thread about this very topic from years ago on Hydrogenaudio (another science-based audio forum). Here is some straight talk about vinyl distortion that could actually be considered sympathetic. From JJ (James Johnston, whom you should consider authoritative), post #49



and later, post #101

Yes T Chapman (1988) is a reference album, using Neumann (1950s!) microphone, and double microphone technique in several tracks, details of the recording are in several magazines. It sounds impressive even with out of phase and comb effect impossible to avoid and all the non linear effects of valve electronics. So yes! my comparison is completely strained by the media and technology and is only intended to show FR response and tonal difference between digital and vinyl. The Chapman album is an exceptional example of both media versions, with same FR and same dynamic ( same as in Dr Loudness webpage!), meaning that my vinyl rip is perfectly good for the intended tonal comparison at normal listening loudness where the noise threshold of the ear get noise out of the equation most of the time. So I hope this could demonstrate that there are good digital editions...and an impressive high number of absolutely flawed ones, like the America HR track example where a difference of 10 dB at 10 kHz couldn't be explained by problems in the turntable chain. And is not that both are good but with different EQ, good tracks have a typical defined EQ profile, a slope tendency to pink noise, the HR version is bad. There are examples of good sound in digital albums like Dire Straits and bad things like Tracy Chapman GH where DR 8! make irrelevant any other measurements...and is not an isolated case.
I don't have any arguments for people that find DR 8 a masterpiece. Time by I will look for some examples.

 

[krabapple]

Yes T Chapman (1988) is a reference album, using Neumann (1950s!) microphone, and double microphone technique in several tracks, details of the recording are in several magazines. It sounds impressive even with out of phase and comb effect impossible to avoid and all the non linear effects of valve electronics. So yes! my comparison is completely strained by the media and technology and is only intended to show FR response and tonal difference between digital and vinyl. The Chapman album is an exceptional example of both media versions, with same FR and same dynamic ( same as in Dr Loudness webpage!), meaning that my vinyl rip is perfectly good for the intended tonal comparison at normal listening loudness where the noise threshold of the ear get noise out of the equation most of the time. So I hope this could demonstrate that there are good digital editions...and an impressive high number of absolutely flawed ones, like the America HR track example where a difference of 10 dB at 10 kHz couldn't be explained by problems in the turntable chain.

It could be explained by a number of things other than simplistic 'vinyl vs digital'. You simply don't know what was done during the sourcing and mastering of each release.