Why, oh why, does vinyl continue to blow away digital?

[Lagonda]

Dave is a Mastering Engineer and not a Recording Engineer. A better question would be which commercial release sounds most like your mastering.

FYI - there is so much that happens between a live feed and a commercial release on the recording and production process that they are so far removed from each other, even after the recording engineers hands it off to the mixing engineer, who then hands it off to the mastering engineer.

And then the mastering engineer tries to polish the digital turd all shinny and listenable !

 

[Mcsnare]

The absolute accuracy is not really open for discussion. A signal is preserved orders or magnitude better going through a digital process than an analogue one. Consider the following:

Digital:
sound waves -> voltage at mic membrane -> preamps and manipulation -> high resolution ADC -> bit perfect storage -> high resolution DAC -> voltage at preamp input to feed your system.

Analog:
sound waves -> voltage at mic membrane -> preamps and manipulation -> to storage:

a) tape: voltage induces magnetic encoding in a magnetizable support. Timing is encoded by motor motion pulling a tape trough the place encoding takes place (the head)
b) direct to metal vinyl: voltage induces magnetic fluctuations that move a hard needle across a metal or lacquer support. Timing information is again entrusted to a motor speed. Encoding is performed by either active or passive elements to apply an eq curve to the signal (usually RIAA)
c) tape to vinyl: sum the two processes before and add the tape read

-> storage in material media directly -> to read

a) tape: feed again the tape, run motor and read magnetic fluctuations on tape to induce voltage coming out of a read head -> perform any corrections for used curves via passive or active electronics
b) vinyl: run motor and use a stylus to create voltage coming from shaking a magnet or coil, amplify that by orders of magnitude (40 to 80db?), re-apply equalization to undo the previous one that was applied in recording by passive or active components -> voltage at preamp input to feed your system.

It should be clear that the digital step is infinitely more accurate, there are no moving parts, very few parts at all, and they are all quite quantifiable, trackable and easy to account for.
Every part of the analog process is difficult and adds errors: signal goes though a incredible number of magnetic reconstructions, speed relies on motors all the way, everything has inertia (stylus, platters, bearings,...). There is no such thing as a 'pure' analog recording any more that a pure digital recording. The signal is tortured all the way from the mic to the final medium, and then tortured back. It is rebuilt so many times on the other side of a transformer, equalized in lossy processes and then back again to be produced, it is absolutely comparable to a (haphazard) digitalization process, just in analog media. There is nothing magical about it.

ADC is effectively transparent. DAC can be made effectively transparent as well. This is trivial to measure and quantify, don't get confused by people that say otherwise. Digital processes run the world and their resolution is not something we should trivialize with annecdote. 24 bits gives you just short of 17 million discrete intervals to categorize something.

Just a bit of fun, those 24 bits, how do they compare to the potential of vinyl? the smallest groove is about 0.04 mm, so half that divided by 24bits is ~1.2e-11. That's the information size, in meters, you'd need to get to reproduce 24bits on a record. So do we get it? A small molecule is about ~1e-9 meters. We're in about two orders of magnitude off, in favor of a simple 24bit recording. It is, at minimum, 100x more resolving than analog. This is the reason why people digitize tape to DSD and 'it sounds the same'. The tape signal literally fits within the digital signal, with headroom to spare, and that is only possible with the ubiquitous high quality ADCs and DACs we have.

So, digital is more accurate, in absolute terms.

Now here comes the kicker, after all this blabber: does it matter? Are we even making the right questions? Is absolute accuracy relevant or is there something else?
It is more relevant if we are considering just 'information'. But what about 'music', and more generally, sound?

Remember this all starts at the microphone and ends in our ears (assuming acoustic music of course). And that's important, they are the gatekeeper for all of this. A mic is a deeply flawed device: quite frequency selective, saturates quickly, directional, has internal resonances and so on, and those are always there, independent of the downstream being digital or analog. Our ears are laden with eons of evolutionary pressure to be also very selective, typically at the expense of linearity in most processes. There are a number of psychoacoustic clues as to why we prefer a given type of euphonic presentation over another, as to why a given type of noise renders details more present and trackable, and so on. Some of these things are natively (by coincidence) present in analog chains, and that seems to be a reason why they score high in the preference score, even trying to discount the cultural and ritualistic effects innate to using the medium. So it could be that analog is more accurate to the music under certain conditions.

As a user of both, I generally don't share that opinion. I can get digital to sound analogue but not the other way around, and that tells me something. I certainly entertain these ideas and they inform my design choices. But I'm not sure we are debating a meritorious problem, at least in it's current typical format. There is nothing inherent about any of the two processes that elevates it above the other, in relative terms. There is too many music recorded natively in each of the processes to disregard, and I have to agree with Mike, above a, certainly diffuse but nonetheless real, threshold of quality, the differences are difficult to pin down, highly release dependent and quickly left at the door and you just enjoy the outcome.

What he said.

 

[Carlos269]

Great post.

It seems to me, the quality of the final product that we have available to listen to, whether in an analog or digital form, is far more dependent on all the myriad production decisions, and the skills of those involved, from recording to mastering and beyond, than the format the product is delivered in or the gear we use to play it back on.

Everything we spend so much time kvetching over — particularly with regards to the so called digital/analog debate- is just the icing on the cake.

The Mastering is the most dominant factor of resultant sound quality. Nothing else, including source components, speakers, and the room contributes as much the Mastering.

 

[mtemur]

And then the mastering engineer tries to polish the digital turd all shinny and listenable !

You can’t polish a turd

 

[Mcsnare]

The Mastering is the most dominant factor of resultant sound quality. Nothing else, including source components, speakers, and the room contributes as much the Mastering.

I’m not sure I agree with that. The recording, mixing, and production decisions are far more important to a good sonic outcome than the mastering IMO.

 

[Carlos269]

You are right. Contractors, designers and developers of military systems have little knowledge about how to use them in the operational environment. They know a little about operational or tactical usage because of keeping certain things classified. That’s normal because their function is to design the product up to specification. Beyond that is above their head. it’s the procuring military’s job. Actually they’re not too different than developers working for other industries in this respect. And when it comes to detecting submarines with a filter similar to HQPlayer’s…Well, good luck with that.

If it has been true that those filters play an important role for Finnish Navy to detect Russian submarines would it be announced publicly? Would they let it be shared here? Maybe he did something for the Navy but catching submarines with the help of a filter is a very long shot.

No.one has claimed that submarines in Finland are being detected with HQPLAYER. Wow to think of such a thing is telling. The dynamic filtering and techniques can be ported over from those types of developments. I have not seen any clear evidence so it is not spoken in the open, but it makes good sense who know about dynamic filtering and detecting data beneath the noise floor, which is inline with what HQPLAYER does. You need to use logic and common sense here.

 

[Carlos269]

I’m not sure I agree with that. The recording, mixing, and production decisions are far more important to a good sonic outcome than the mastering IMO.

Here is where also having a good understanding of electronics comes in Dave. You need to understand that consumer electronics as opposed to professional audio studio electronics work at different voltage levels and deviate from neutral to different degrees, in dB. There is nothing that you can do by connecting random consumer devices, which are designed for neutrality, that is anywhere to the control over the resultant sound quality that mastering engineers have at their fingertips. I’m speaking about things that the end user can control. The recording and mixing are a done deal, but consumers have a choice of mastering versions, in both analog and digital releases.

 

[PeterA]

So it could be that analog is more accurate to the music under certain conditions.

I think this is an interesting comment as it gets to the heart of what the actual goal is. Accurate to what? Do we want what we experience at the listening seat to remind us of what we experience when listening to live music, or do we want something else, simple enjoyment of art, the original performance interpreted and altered by those in the production side post performance for some "predictable result"?

Is what we hear at the listening seat in our rooms still accurate to the music? Carlos asked a good question: Are we expecting something that we can not have because of the processes involved in the making of the recording?

 

[Mike Lavigne]

I attended an AES, NYC chapter demonstration of this very nature.
A person with the magic box varied levels of jitter and distortion while the music played. Y’all would be very surprised how much harmonic distortion it takes before it becomes audible. Increasing jitter (while it takes far less than harmonic distortion) is also surprisingly hard to detect up to a larger than I expected amount.
In 2024, digital when done with well designed gear, and using best practices, is a much closer representation to what is sampled than the same signal sampled by analog-regardless of the analog format.
I would rather cut lacquers off a well made digital copy of the 1st generation mix tape, than use a 2nd gen tape safety copy.
To the person that preferred the sound of their voice when recorded to lacquer because it sounded more real, your preference is totally valid but your logic is off. Something can’t be less faithful to the source (analog) and also be more real. It might SEEM more real but it’s just the pleasing quality to the distortion. I get it!
When you add in all the analog steps an analog recording has to go though to get to the end user, the differences become even more magnified as the distortion adds up. Digital doesn’t have this problem, well it kind of does in certain cases like during mixing when something is eq’d or otherwise processed, The DAW apps are run at a higher bit rate to perform the math to a high precision - then round up to a usable bit depth.
To the people that say how can you even know that what is fed to the A to D is what comes out the other side - it’s easy in a studio when a person or group is recording and you hit a button in the control room to compare the analog feed of the console to the DAW output (post digital)
I’ve done it many times in my mastering studio, but not using a live feed - rather I used a 1/2 “ or 1/4” 2 trk of final mixes.
Unless something is really off, it usually sounds damn near identical.
A tape copy, not so much - but the tape copy may pick up in one area and lose in another so in some cases the copy may have something special about the sound. But that would NOT be as a result of accuracy.

I still like the distortions inherent in analog so much, that I like listening to vinyl much more than digital. Especially if the vinyl version is the ONLY analog step in the recording process.

Come at me. I ain’t afraid.

ok.

other prominent recording engineers have a different perspective about digital and analog. Bernie Grundman talks about (1) what can be measured, (2) whether digital can capture everything that analog can and also (3) generational losses from digital. Bernie is just one opinion. but he likely works in analog as much as anyone. so certainly has the hands on experience. and has a history of great sounding results.

this one at the 1 minute point talks about what can be measured, and from 1:17 to 1:27 he speaks about generational losses from digital;

"things are missing"

this one around the 24 minute mark he talks about high rez digital not capturing as much of the tape source as analog.

 

[sparkie]

It makes no scientific sense. And I am nothing if not a diehard scientist with a PhD in computer science to boot, having spent a large part of my career teaching in one of the world’s greatest PhD programs in CS, and now working at one of the world”s leading software firms in the Bay Area. I’m no technological Luddite. I’ve invested heavily in digital with top solid state and tube DACs.

But each time I play my vinyl on one of my turntables, man oh man, why do I always feel that digital sucks? I’m listening to a garden variety pop album — Fleetwood Mac — on an SME 20/12 table with the V12 arm with a Koetsu Onyx Platinum feeding the ARC Ref 3SE phono stage. The sound is organic, so compelling to listen to that my left brain analytical mind that whispers this is all distortion sounding pleasant is quashed by my right brain saying who cares when it sounds so good?

Am I delusional?

No, your not delusional. As an electronics engineer, I can tell you the imperfect things about digital, even though it looks like you invested into a DAC which is most of that media's issue. Since I left Dolby Labs, people tell me off and on I should go on my own and make better converters which I might do once the non-disclosure agreement laws are canceled here in the US.

As far as mixes and the styling, its only changed for some. Because the mastering is the stage where they over crank it to overcome DAC losses.

This video is one of the few examples out there that shows the over cranked mastering method. Even though there is a few other methods I observed in studios.

 

[Rensselaer]

The absolute accuracy is not really open for discussion. A signal is preserved orders or magnitude better going through a digital process than an analogue one. Consider the following:

Digital:
sound waves -> voltage at mic membrane -> preamps and manipulation -> high resolution ADC -> bit perfect storage -> high resolution DAC -> voltage at preamp input to feed your system.

Analog:
sound waves -> voltage at mic membrane -> preamps and manipulation -> to storage:

a) tape: voltage induces magnetic encoding in a magnetizable support. Timing is encoded by motor motion pulling a tape trough the place encoding takes place (the head)
b) direct to metal vinyl: voltage induces magnetic fluctuations that move a hard needle across a metal or lacquer support. Timing information is again entrusted to a motor speed. Encoding is performed by either active or passive elements to apply an eq curve to the signal (usually RIAA)
c) tape to vinyl: sum the two processes before and add the tape read

-> storage in material media directly -> to read

a) tape: feed again the tape, run motor and read magnetic fluctuations on tape to induce voltage coming out of a read head -> perform any corrections for used curves via passive or active electronics
b) vinyl: run motor and use a stylus to create voltage coming from shaking a magnet or coil, amplify that by orders of magnitude (40 to 80db?), re-apply equalization to undo the previous one that was applied in recording by passive or active components -> voltage at preamp input to feed your system.

It should be clear that the digital step is infinitely more accurate, there are no moving parts, very few parts at all, and they are all quite quantifiable, trackable and easy to account for.
Every part of the analog process is difficult and adds errors: signal goes though a incredible number of magnetic reconstructions, speed relies on motors all the way, everything has inertia (stylus, platters, bearings,...). There is no such thing as a 'pure' analog recording any more that a pure digital recording. The signal is tortured all the way from the mic to the final medium, and then tortured back. It is rebuilt so many times on the other side of a transformer, equalized in lossy processes and then back again to be produced, it is absolutely comparable to a (haphazard) digitalization process, just in analog media. There is nothing magical about it.

ADC is effectively transparent. DAC can be made effectively transparent as well. This is trivial to measure and quantify, don't get confused by people that say otherwise. Digital processes run the world and their resolution is not something we should trivialize with annecdote. 24 bits gives you just short of 17 million discrete intervals to categorize something.

Just a bit of fun, those 24 bits, how do they compare to the potential of vinyl? the smallest groove is about 0.04 mm, so half that divided by 24bits is ~1.2e-11. That's the information size, in meters, you'd need to get to reproduce 24bits on a record. So do we get it? A small molecule is about ~1e-9 meters. We're in about two orders of magnitude off, in favor of a simple 24bit recording. It is, at minimum, 100x more resolving than analog. This is the reason why people digitize tape to DSD and 'it sounds the same'. The tape signal literally fits within the digital signal, with headroom to spare, and that is only possible with the ubiquitous high quality ADCs and DACs we have.

So, digital is more accurate, in absolute terms.

Now here comes the kicker, after all this blabber: does it matter? Are we even making the right questions? Is absolute accuracy relevant or is there something else?
It is more relevant if we are considering just 'information'. But what about 'music', and more generally, sound?

Remember this all starts at the microphone and ends in our ears (assuming acoustic music of course). And that's important, they are the gatekeeper for all of this. A mic is a deeply flawed device: quite frequency selective, saturates quickly, directional, has internal resonances and so on, and those are always there, independent of the downstream being digital or analog. Our ears are laden with eons of evolutionary pressure to be also very selective, typically at the expense of linearity in most processes. There are a number of psychoacoustic clues as to why we prefer a given type of euphonic presentation over another, as to why a given type of noise renders details more present and trackable, and so on. Some of these things are natively (by coincidence) present in analog chains, and that seems to be a reason why they score high in the preference score, even trying to discount the cultural and ritualistic effects innate to using the medium. So it could be that analog is more accurate to the music under certain conditions.

As a user of both, I generally don't share that opinion. I can get digital to sound analogue but not the other way around, and that tells me something. I certainly entertain these ideas and they inform my design choices. But I'm not sure we are debating a meritorious problem, at least in it's current typical format. There is nothing inherent about any of the two processes that elevates it above the other, in relative terms. There is too many music recorded natively in each of the processes to disregard, and I have to agree with Mike, above a, certainly diffuse but nonetheless real, threshold of quality, the differences are difficult to pin down, highly release dependent and quickly left at the door and you just enjoy the outcome.

Just one point, in an earlier post I mentioned how Sony music has bought up a large number of music companies in order to own the music. They allegedly do not release the master tapes but instead make DSD copies of the analogue tape then sell copies of that DSD master to record companies (like MoFi).

So, in those cases where the music being converted to digital is coming off the analogue tape, the digital DSD files record all the distortions from mic input to reel2reel output into the A2D device. That should erase those variables you mentioned in front of tape output when comparing accuracy of the two, no?

So, what comes off the analogue master tape gets cut directly to lacquer, vs what comes off the analogue master tape being converted by A2D to digital and stored as DSD, then that DSD data going through a DAC to return to some sort of analogue and that being cut to a lacquer. Which is more accurate?

 

[Mcsnare]

ok.

other prominent recording engineers have a different perspective about digital and analog. Bernie Grundman talks about (1) what can be measured, (2) whether digital can capture everything that analog can and also (3) generational losses from digital. Bernie is just one opinion. but he likely works in analog as much as anyone. so certainly has the hands on experience. and has a history of great sounding results.

this one at the 1 minute point talks about what can be measured, and from 1:17 to 1:27 he speaks about generational losses from digital;

"things are missing"

this one around the 24 minute mark he talks about high rez digital not capturing as much of the tape source as analog.

Bernie is great and beyond reproach, however some of his opions in this area are not shared by any other mastering engineers I know off.

 

[RCanelas]

Just one point, in an earlier post I mentioned how Sony music has bought up a large number of music companies in order to own the music. They allegedly do not release the master tapes but instead make DSD copies of the analogue tape then sell copies of that DSD master to record companies (like MoFi).

So, in those cases where the music being converted to digital is coming off the analogue tape, the digital DSD files record all the distortions from mic input to reel2reel output into the A2D device. That should erase those variables you mentioned in front of tape output when comparing accuracy of the two, no?

Well, the tape need to be read by the r2r machine in any case, introducing its own set of wow and flutter, head defects and other processes. All of that is just captured by a high enough quality of a digital capture. Correcting for those requires:

a) a control signal, so you can make the differential - can't even imagine how that would look like given everything the signal has gone though
b) a lot of faith and an average, ensemble, educated or best guess understanding of what the signal was subjected to, so you can reverse it differentially
c) a certain amount of desire self harm for no gain. The tape and the capture are going to sound absolutely apex as is (if the tape and recording are good), so you don't need to waste time and can just enjoy the music

 

[Mike Lavigne]

Bernie is great and beyond reproach, however some of his opions in this area are not shared by any other mastering engineers I know off.

in your post you said "Come at me. i ain't afraid" and the best you can do is to dismiss Bernie as wrong.

that is pretty lame. how can we take you seriously?

 

[bonzo75]

Bernie is great and beyond reproach, however some of his opions in this area are not shared by any other mastering engineers I know off.

Maybe the difference between professionals and amateurs

 

[Mcsnare]

in your post you said "Come at me. i ain't afraid" and the best you can do is to dismiss Bernie as wrong.

that is pretty lame. how can we take you seriously?

I’m not dissing Bernie or anyone else. I have my opinions and you have yours. Nothing more or less. I don’t care if anyone takes me seriously or not. I am only sharing my experiences that have informed my opinions.

 

[Mcsnare]

Maybe the difference between professionals and amateurs

Not quite.

 

[Mike Lavigne]

I’m not dissing Bernie or anyone else. I have my opinions and you have yours. Nothing more or less. I don’t care if anyone takes me seriously or not. I am only sharing my experiences that have informed my opinions.

fair enough. i respect you have your opinions.

but the thing is; what i hear in my experiences, exactly mirror what Bernie is saying. so there is that.

 

[sparkie]

Maybe the difference between professionals and amateurs

Its hard to tell.
But there is a lot of 'quirks' or 'idiosyncrasies' when designing the ADC circuit much less the ADC IC chip itself. Because it has a different non-linear losses in level and harmonic response compared to other recording media. Analog recording also impart small Fourier harmonics that are not present in digital as well. But no one really talks about that except chip engineers and that sector of people. But the guy was correct that it is difficult to measure some things with lab equipment than to hear it. Especially analyzing odd and even harmonic behavior in a circuit.

 

[PeterA]

I never listen to suspend disbelief. More or less real is not on my wish list. I simply connect with the music better when it has the distortions inherent in the cutting and playback process of vinyl.

@Ron Resnick , do you see this? Perhaps this is another goal that should be added to your list.

I understand wanting connection to the music, but not caring if it is more or less realistic/convincing is an approach that seems a bit unusual. Very interesting in light of the discussion about accuracy.