Does Vinyl Have Wider Dynamic Range Than CDs? Here's Some Math

While many of us seem to hear wider dynamic range coming from vinyl, the numbers would seem to point in the opposite direction. Or do they?

Please download and read the attached PDF file written back in 1996 by Ron Bauman, who has a degree in Electrical Engineering from Lehigh University. He argues that vinyl actually has wider dynamics and he's got the numbers to prove it (unfortunately a few pages "went missing" over the years).

In addition he makes some impressively prescient observations and posits that time domain "..may be a more natural frame of reference for audio instrumentation development" (than frequency domain).

The article originally appeared in the April 1996 issue of the now defunct Audio Electronics magazine.

I am attempting to contact Mr. Bauman to get his current thinking on the subject.

(The image, used without permission, is of an Audi display ad at Victoria Station in the U.K. using 45s to recreate the Audi ring logo.)

tnargs's picture

Let me get this straight, Michael. You found someone with an electrical eng. degree who backs your anti-CD views with a paper that is an embarrassment. Meantime, all the elec engs who disagree with you are cloth-eared philistines? If you have any interest in the truth of a topic, cherry-picking a small minority of favourable papers is exactly the wrong way to do it. It doesn't look good.

The reasons LP can sound better than CD are well known and conventional science: [1] 'loudness war' mastering ruining some CD's, and [2] the psychology of unconscious bias in many listeners making for a 'favourable receptivity' for LP in their minds that is driven by non-sonic factors and have sufficient strength to over-ride the actual sound wave differences. For example, Michael, you have such a tragic case of [2] that no CD has a chance under audition with you, no matter what its sonics.

So please, stick to reviews, and stick to LP reviews.

Joe Crowe's picture

I offer this in Michaels defense (not that he needs it). The theoretical dynamic range of a CD is always given as ~95 dB, calculated as the difference between 16 zeroes and 16 1's. This ignores several facts tnargs and others may not even be aware of. First off 2 bits are for error correction (bit errors not musical ones sadly). Next, maximum level is dropped several dB as computers just don't like all 1's. Also about 10 - 15 dB of dither is added at the bottom to counter the extinction effect observed at low frequencies. Add it all up and you get a usable DR of about 60 - 65 dB, still impressive but certainly not the ungodly 95 many digiphiles swear by. Incidentally the Sheffield Track Record, admittedly no more than a demo, was measured at greater than 70 dB. It's unfair to accuse MF of cherry picking papers. Lots exist, he picked one that was mostly readable. This is the advantage of having been a practicing audiophile during the digital wars of the 70's and 80's. If red book audio truly was the be all and end all of music reproduction I don't think its' advocates would protest so vocally but would rather let the product speak for itself.

Dr. AIX's picture

CDs with dithering can achieve 93 dB of signal to noise ratio. Let's start with the right information before you claim CDs do about 60-65...that's more like the dynamic range of a first generation analog tape. And most vinyl LPs were created from third generation analog tapes back when I was a mastering engineer.

The whole notion that vinyl LPs have more dynamic range than a well produced CD spec PCM digital recording is a myth.

pt11055's picture

Having actually engineered and mastered CDs (and vinyl) I can assure you there are some errors here. In 16/44, the first two bits are not for error correction, that's entirely separate, and structured depending on the storage medium. There's no 10-15dB of dither, it's more like 3dB, and carefully shaped if done properly. The maximum is not dropped several dB because computers, or anything else, "doesn't like all 1's. What is avoided is full scale clipping, or signals where inter-sample clipping could occur. The adjustment isn't several dB.

Vinyl's dynamic range is dependant on frequency, because vinyl doesn't have a flat maximum output curve. PCM of any flavor has a flat response to FS.

When I cut a master for vinyl and a CD master from the same digital master tape, they sounded pretty much the same except for the noise floor. Yes, vinyl was noisier. By a lot.

tnargs's picture

Fortunately others have sorted you out -- only for you to call them all 'sooooo touchy', and acknowledge none of your vast erroneous assumptions -- and I'm half expecting to see Mr Fremer quote your post in the future.

The 'mostly readable' paper, that you applaud as being one of many that claim vinyl has a DR of 110+ dB, assumes that the vinyl itself is 'infinite' and the only DR limitation is in the playback electronics. What a crazy assumption.

The only thing I am still wondering about, is whether Michael didn't think it through, or did he realize it but chose to treat his readers as gullible?

Michael Fremer's picture
1) You don't have this "straight" at all. You are the one saying "all electronics engineers who disagree with me are cloth-eared philistines". I never said that nor is that implied in anything here. 2) Calling what he wrote an "embarrassment" is hardly a refutation of what he wrote. 3) I posted an interesting viewpoint written by an electrical engineer and posed the headline as a question or does the meaning of ? escape you? (? means a question).
Michael Fremer's picture
LPs can sound better than CDs because they have more extended frequency response on bottom and top. They don't require brick wall filters. Instead the high frequencies that can extend upwards of 40kHz gently roll off. So there's no ringing, filter caused time domain issues to which the ear is sensitive and there's no "approximation" of the original waveform. LPs usually sound better and more like real music simply because THEY DO. I don't have a tragic case of anything. In fact I'd say you have a tragic case of believing what measurements say rather than what your ears tell you. But whatever.
pt11055's picture

No ringing, huh? Clearly you've never run a square wave test record and scoped the output of the phono preamp.

Michael Fremer's picture
I do not have "anti-CD" views. I have anti-unpleasant sound views. You claim "no CD has a chance" because of my anti-CD bias. I 100% reject that. I play plenty of CDs and do my best to sit through them but after a short time my body tenses up and my senses turn off. That has been the case since I first started listening to CDs at a time when my mind and body were primed to LOVE them.

You claim the loudness wars are responsible for why LPs can sound better. That is 100% ridiculous. CDs have sounded unpleasant since long before the "loudness wars".

Ortofan's picture

Assume that your listening room is as quiet as a library [] with a background noise level of 40dB. Add to that the 96dB dynamic range and the peak output level from your speakers would need to be 136dB to experience the full dynamic range from a CD. A sound level of 136dB is jet engine loud.

Now, assume that your speakers are relatively efficient and can produce a sound level of 96dB with an input power of 1 watt. How much power is then required for those speakers to produce a sound level of 136dB (40dB higher than the level at 1 watt input)? 10,000 watts would be needed to achieve that 136dB level []. Can your amplifier output 10,000 watts? Can your speakers handle 10,000 watts? Can you tolerate a sound level of 136dB? If not, then the dynamic range issue is a moot point.

Joe Crowe's picture

Thanks for clearly stating what I was clumsily attempting earlier. The "theoretical" dynamic range of CDs is just not realized in practice. Probably never will be outside of an outdoor rock concert setting.

audiof001's picture

MF wrote: "He argues that vinyl actually has wider dynamics and he's got the numbers to prove it"

No where does Michael demand that this report be taken as gospel, nor is he calling people 'cloth-eared philistines.' He merely suggested we read the article. No need to get your back up against the wall. Enjoy your CD's - I'm sticking with vinyl - the difference I hear I prefer.

audiof001's picture

When I first heard a CD in the 1980's, I noted its gain was louder than my phono section. At the time I was running a low output MC thru a head amp and into a highly regarded preamp. The bass was fuller on CD's because bass had been systematically lowered on all vinyl for decades so that the records would play on inferior turntables snd CD;s didn't need this compensation. Given the choice, people most often prefer the louder source with more bass (Beats anyone?). But louder isn't always better, it's just louder. is still available if any CD lovers would like to purchase it and blog there. This is analog planet, uh, it's what we do.

Bassman59's picture

The bass was fuller on CD's because bass had been systematically lowered on all vinyl for decades so that the records would play on inferior turntables snd CD;s didn't need this compensation.

Well, that "systematic lowering" is called the RIAA Curve, and was implemented back in the Dark Ages so that the needle doesn't jump out of the record groove on low loud notes. Since there is no fear of program-related mechanical failure with CDs (or other digital playback), there is no need to master the recording through such equalization.

That producers choose to emphasize the low end in the program material is an aesthetic choice, not a technical one. There are plenty of CDs which do not have excessive low end.

pt11055's picture

The RIAA record characteristic and play characteristic are precisely complimentary, at least that's the intent. The response of the system is intended to be flat. There's no deliberate bass rolloff in the system. Extreme bass must be mastered panned center, though, because the maximum modulation of a vertical groove is much more limited than a lateral one. Bass modulation takes up a lot of land on a record, so lots of loud bass limits your play time. For those reasons, bass has sometimes been controlled during mastering, but you can also master completely flat and take the playing time hit. Bass reduction is a choice, a deliberate compromise. An example of this taken to extreme would be Todd Rundgren's "Initiation" from 1975, with a total play time of over 67 minutes...and no bass. In fact, the jacket included a disclaimer about that, and a suggestion that the record be dubbed to tape in an effort to restore the bass. That didn't work of course, but dub s to cassette did have less high end putting the thin bass a bit more in balance.

RIAA had nothing to do with it though.

Bassman59's picture

The theoretical dynamic range of a CD is always given as ~95 dB, calculated as the difference between 16 zeroes and 16 1's.

That's not how it's calculated. You need logarithms.

First off 2 bits are for error correction (bit errors not musical ones sadly)

None of the 16 audio data bits are used for error correction. The dual-interleaved Reed-Solomon error correction coding adds many extra bits which are used to detect and correct errors. And before you jump on me and say, "but the error correction changes the audio data," let me say: no, it doesn't. You install (or used to install) computer programs from data CDs and DVDs, and those media use the same sort of error-correction-coding and it works. How do we know? If it didn't, the programs you installed would crash your computer.

Next, maximum level is dropped several dB as computers just don't like all 1's

"All 1s" is called clipping, and if you ran your analog signal above the rail you get the same effect. So analog recording levels have to allow for sufficient headroom to avoid clipping, too.

Also about 10 - 15 dB of dither is added at the bottom to counter the extinction effect observed at low frequencies.

You don't understand what dither is, or why it is used. There is no such thing as "extinction effect." Dither has nothing to do with "low frequencies" vs any other frequencies.

. Add it all up and you get a usable DR of about 60 - 65 dB, still impressive but certainly not the ungodly 95 many digiphiles swear by.

If you add it up your way, you get "about 60 dB," but your math is wrong, as is your understanding of the entire process.

Joe Crowe's picture

Are digital defenders always sooo touchy? No you don't need logarithms, they are used to arrive at a number in dB but the loudest sound possible is all 1's and the lowest level possible is all zeroes this is self-explanatory. So touchy! Since you understand what clipping is you probably know that in the digital realm it causes more problems than in the analogue domain so engineers avoid the issue by dropping the max a few dB. I was deliberately trying to avoid a technical discussion when the point was simply that while the math gives 95 dB it's not all usable. I dare you to try and make a disk with a spell of 16 zeroes and another of 16 1's and try to play it. If you succeed and nothing blows up I'll buy you beer, hell a whole case. Technical gobbledygook can always be used to try and baffle people but I prefer to avoid it where I can. A simple point, 16 bit digital has ~95 dB SNR on paper but we can't record on paper. If we can't take advantage of the whole 95 then for all intents and purposes we don't have 95.

Bassman59's picture

I dare you to try and make a disk with a spell of 16 zeroes and another of 16 1's and try to play it. If you succeed and nothing blows up I'll buy you beer, hell a whole case. Technical gobbledygook can always be used to try and baffle people but I prefer to avoid it where I can.

So let's avoid technical gobbledegook. A file that has 16 zeros and 16 ones (which, more correctly, would be 16 -32768s followed by 16 32767s, but who's being technical, right), defines basically a 1500 Hz square wave (assuming 48 kHz sample rate and that you repeat the pattern).

Why would it blow up? You're simply playing back a full-tilt-boogie square wave. That sounds like shit (or EDM, your choice) if it was created digitally or only in analog. As long as you don't exceed your speakers' power ratings, why would anything be damaged?

And you miss the point about the ~95 dB dynamic range. You can sample a sine wave that "uses all of the bits," meaning that its one-sample max is 32767 and its one-sample min is -32768. Or you could sample a complex waveform where its one-sample max is 32767 and its one-sample min is -32768. Or you could sample a 1500 Hz square wave (16 samples 32767 and 16 samples -32768) and you'd still use "all of the bits."

Really, it is no different from analog clipping. Instead of your maximum and minimum voltages being defined by the power-supply rails (which is not always the case, since many amplifiers cannot swing to the rails), it is defined by the converter reference voltage.

Exceed the rails? You clip. Exceed the reference? You clip. It is the same thing.

Why are "Digital Defenders" so touchy? Because we constantly have to correct the absurdities you Analog Allies keep throwing out there.

Joe Crowe's picture

Never said it "had" to blow up. I said "if you succeed" so send me the disk the beer is waiting. Take something with barely audible content followed by something that is 95 dB louder and most systems will be strained to the breaking point. God bless any that aren't. Don't see myself as an analog defender so much as someone who was badly disappointed by early digital and hasn't fully recovered. Only thing I really know for sure is CD audio and digital recording was PERFECT in 1980 and has gotten a LOT better since. Haven't seen the "theory" explaining that one yet but it's a belief I don't dare question for fear of the backlash. Want to buy a bridge, I'm not using it.

pt11055's picture

Get out that beer. It's already been done, many times. Full scale square waves, and other FS signals, exist on several test CDs, which I promise you are guaranteed not to rust, bust, corrode or explode, caused anything else around them or connected to them to blow up, and have and are played just fine. It's been done since the early days of CDs and test CDs. I have one from 1982.

Make mine Spotted Cow please.

Michael Fremer's picture
The National Archives is busily archiving CDs that are slowly degrading over time for a variety of reasons.
zeus's picture

To me, the engineering of the record or CD makes all the difference. A Caddy for Daddy by Hank Mobley sounds great in ll formats. But Maiden Voyage by Herbie Hancock sounds lousy in all formats.

That said, when I first started playing LP's again after a 20 year lay-off, using a new $500 turntable, my wife, who does not listen to much music was amazed by the sound of the LP's I was playng.

I listen to & buy both LP's & CD's and enjoy both.

my new username's picture

"No doubt, we can more readily hear music embedded in thermal noise than in quantization noise. Why? Digital noise destroys information, whereas thermal noise adds innocu- ous, unrelated information to music."

Said another way, the absolute extremes between softest and loudness can be greater with 16/44 but the relevant changes in dynamics that occurs between the extremes is LESS.

Concluding "CDs have 96db S/N ratio and so that's that" is yet another example of measuring the wrong thing when labeling something as superior.

As a nice bonus, the author mentions how Nyquist is irrelevant ... not because Nyquist was wrong but that it's merely never addressing all the relevant aspects of music reproduction.

pt11055's picture

Proper dithering permits recording and reproducing signals at levels below the theoretical -96dBFS noise floor. 96dB is not that firm a limit, and with a little intelligent dither mixed in things don't disappear below -96. While dithering does add a bit of noise, it can function just fine even if it's been shaped, or moved to a less audible portion of the spectrum.

my new username's picture

Dither's a great band-aid for 16-bit files when done well. Something like MBIT+ gets closer to "pure" thermal noise for a better mask.

The point of the quoted passage however is that dither isn't necessary when you aim your efforts higher, with resulting benefits.

phuzzyday's picture

Some of you guys are pretty opinionated, you must have a lot of money invested one way or another...
Anyway, He talks a lot about Dynamic Range of the cart and preamp, etc... But then goes on to mention how, as soon as you put the stylus on a moving record, you lose a huge chunk of that range.

Could a record be created which had no surface noise, at all? Maybe someday. Meantime, folks, isn't the relevant thing the real world performance we actually DO get? Sure, CD's are more dynamic. But engineers are ruining them in mastering due to the loudness trend. If it wasn't for that, vinyls sound advantage would be pretty small.

Meantime, I'm buying vinyl 100% of the time, so don't get me wrong.

sdecker's picture

-- not numbers. The digital zealots will usually 'win' the numbers game on their own terms versus analog's numbers for the most conventional of measurements. Even digital's Nyquist theorem is fully accurate and relevant only for test tones, as the author here points out. But it will take an open-minded engineer with a working background in signal processing to deconstruct it for most of the folks getting into these lose-lose analog vs digital arguments. Most people are confused by dynamic compression (ie 'loudness wars') vs digital storage compression (ie mp3), so mainstream discourse of the Nyquist Theorem beyond half the sampling rate frequency limits are best not encouraged. So Mikey be careful of touting "MATH" in your headline, when the math in this paper is used primarily to establish a noise floor for a cartridge-preamp pairing.

This paper brings up two issues that don't seem to have had rigorous research recently. (Please correct me if I've missed a paper, as this isn't what I seek out for my reading pleasure!) The first is digital's time-domain performance with musical signals and appropriate measurements. The second is digital's noise floor definition and sonics with low-level music, and comparisons with an analog medium of choice using the same criteria. It underscores the quote that "Analog noise is separate from the music and can be listened through; digital noise is intertwined into the music and the brain cannot separate the two."

Both of these issues have improved in the past twenty years, but the 16/44.1 CD limitation is worth (re)investigating even as higher-rez digital makes it easier for designers to work around them. As we know, many still regard CD-resolution as the gold standard.

This paper is prescient for two decades ago. There have certainly been improvements in both the analog and digital recording and playback chains since then. I don't know if I'm convinced by his LP dynamic range analysis, but this kind of outside-the-box thinking is exactly what's needed if these tired old arguments (and measurements) are going to move forward. MF, I would be most interested if you succeed at contacting the author for an "update."

comfortablynick's picture

I'm sure this fellow is a fine EE, but anyone who has a basic understanding of sampling theory will recognize that many of his suppositions about digital audio are simply made up. The problem is that digital doesn't follow the same "logic" that analog does. I don't see any citations next to almost any of the "facts" that he states about digital. Just a few examples (I don't have time to list them all):

Quote: The effect of digital quantization noise is entirely different and nonlinear (The author compares this with "random analog noise, uncorrelated with the music.")

Hmm...I guess the author never read about DITHER. Dither is added before quantization and it replaces quantization noise entirely, whereupon said noise can become completely uncorrelated with the signal if desired. In other words, it's the same as the coveted analog noise (except superior, since we can control the level of dither and it's spectral distribution, which allows us to make it as unobtrusive as possible). This is such a basic error that I have little confidence the author has even the most rudimentary understanding of how digital sampling and quantization work.

Not to mention, this dither would already be WELL below the noise floor of any analog tape, so that would be self-dithering to some degree. At 24 bits, the dither would be well below thermal noise of any analog equipment used in the process.

Quote: The [Shannon-Nyquist] theorem fails to directly mention recovering the phase or amplitude of the signal.

I'm at a loss for why he states this. Why would sampling disregard the phase of the waveform? It seems like our complex musical signals would sound terrible if all the sine waves were not kept in proper phase! Also, the "amplitude of the signal" is certainly captured during the quantization step, which, the author refers to in other parts of the article but evidently does not understand.

Those are just two objections. I find issue with just about every statement this author makes about digital audio. It's also not well-sourced. I think it's a good idea to at least do one's research before trying to write a paper about something. This author simply repeats some standard audiophile anti-digital dogma and throws in some of his own "facts" which he surmises might be the way digital technology works.

Mr. Fremer, posting things like this can only undermine your arguments in favor of analog. Just like you, this author obviously prefers vinyl to CDs. That is perfectly valid and completely subjective. Who could argue that point? No one can argue against your subjective preferences. However, that does not mean that your subjective preferences can in any way translate to valid objective scientific facts. Why try to convince everyone that they do?

The author of your linked article is not secure enough in his preferences to simply state what they are. For some reason, he feels the need to prove that vinyl is technically superior to CD, contrary to all established (and extensive) scientific knowledge and research. He should be able to accept that he prefers vinyl even though it has less dynamic range and more noise than CD. I am perfectly okay with someone stating that they rely on their personal preferences rather than measurements, or that they believe that measurements can't tell us what sounds good. Fine. I won't argue that. But gross misinformation such as what appears in this article absolutely can, and should, be argued against. Giving incorrect "mathematical proofs" of why vinyl is better than CD is well outside the realm of subjective preference. It is an insult to those, like myself, who often prefer a given vinyl record to its CD counterpart but don't feel the constant need to claim that digital is therefore somehow "wrong."

tnargs's picture

Thank you Nick, I have to agree with your every point, including that this line of posting by Mr Fremer only serves to undermine his arguments in favour of analog recordings.

What also staggers me is some of the claims made by commentators on this article, rationalizing 'problems and limitations' with digital, which seem to have come out of a random assumption generating machine. Why doesn't Michael help these poor fellows with their education? At least they might believe him, because their ears are evidently closed to anyone who they think is on 'the wrong side of the war'. (BTW there is no war, there is only education and progress -- analog and digital.) One thing is certain: every time a vinyl digiphobe says he rejects digital because of some nonsensical claim, it looks to all the world like they are not rational or well-informed people. It makes me uncomfortable to admit how very much I enjoy vinyl, lest I be lumped in with them.

Michael Fremer's picture
"Quote:The [Shannon-Nyquist] theorem fails to directly mention recovering the phase or amplitude of the signal." I believe his point is that while in theory it results in 20Hz-20kHz response (which, by the way is far inferior to vinyl's theoretical frequency response and which does not rely upon steep, brick wall filters), the brick wall filters are notorious for having phase issues. If you want to argue that point, take it up with Meridian's Bob Stuart or Ayre's Charlie Hansen both of whom have been working on digital filter issues for decades.
nando's picture

In my view, attempts to demonstrate vinyl superiority by means of tech and math are still futile. What we need is a good pair of ears and a bit of attention.

To me,Analog is better in ways beyond all our math and science.

Kind of a "god given" magic if you will(of course, nothing wrong in trying to understand.....)

But musical appreciation is and will always be subjective, so, each person can have his or her preferences and they always will. That is something I respect.
To me, vinyl by far.

phuzzyday's picture

If a guy is going to try to make a paper as some kind of authority, perhaps he should have some idea of what the sampling rate of CD is. I believe he said 41 Khz.. twice. Talk about discrediting yourself.

And a 'God Given Magic' of vinyl? Dear me. It's definitely encroaching on the spiritual when a signal sent down a wire from two different sources shows, with more accuracy than the ear ever could, that digital is a lot closer to the original. Just have faith, I guess. Still, the poor engineering has me on the vinyl side. But you have to be real about the psychological factors here. It's the nostalgia. The Hands on process. The ritual of it all. It's more involving for the listener, and that appeals to us. Me too.

Michael Fremer's picture
If you really believe a seriously chopped signal at 20Kz and 20kHz is more "accurate" than one that extends from well below 20Hz and well about 20kHz and which doesn't rely upon steep brick wall filtering or a "reconstructed" waveform using added noise and mathematical tricks or that measurements are more accurate than what our ear/brain tells us! That one is positively ridiculous. How did we survive without first measuring where that bent twig sound came from?

Neither analog nor digital sound reproduction is perfect but the one that's perfect for me is the one that doesn't offend my senses. And it has nothing whatsoever to do with nostalgia!

Higher resolution digital audio is a major sonic and technical improvement yet the CD is perfect toads are fighting it. They are the ones who are truly "faith based".

timothy's picture

I've know for years the cassette, 8 track, reel to reel, record players are much more dynamic than digital sources. my ears say they do contrary to the speck sheets. 30 years ago Sony told me cd could reveal limitations of the analog source tape on the back of the cd long box. i remember reading all these lies. if cd sound is perfect why did sony invent super audio cd? sony is a pathalogical liar and would say anything thats not true. i play prerecorded cassettes in my car all the time. they sound great. cd sound and speck sheets are overated

nando's picture

I have to tell you a short story:
some ten years ago when I went back to vinyl, BTW using some very low end gear(Stanton cartridge on thecnics table),one day my father listened to some music I was playing not knowing what the format was (and don't particularly caring, he is not interested in audio)and he was not only amazed by the sound but also enjoying the music so much that he and my family asked me to show them my system and that wonderful sounding cd I was playing.
When I told them it was not a cd but an old LP record they couldn't believe me and I have to led them to the room my system was, for them to believe.
Being me a doctor in medicine, I am familiar with the scientific method. What we do that day(not planned of course) was a blind trial. So blind actually that it was totally not intended.
So, I wonder where is the nostalgia or psychological factor there?
What psychological delusion took over my entire family that afternoon?
Can anyone explain to me??????

MonetsChemist's picture

I think of those "sine waves" that show up in Mr. Atkinson's measurements of DACs. You know, the ones like this one

Astell&Kern AK240

at -90dB. Now clearly music at -90dB is, as someone pointed out above, going to be pretty much inaudible in any real-world listening room.

But the same principle applies.

16 bits of resolution at 0dB; 15 bits at -6dB; 14 at -12dB; 13 bits at -18dB...

And clearly compressed music tends to the 0dB end of things, not the -18dB end of things.

But how much distortion (THD, IMD, TIM, whatever) do we see in a 14bit DAC, which is what our 16bit DACs are when they are dealing with certain parts of our music?

And how does that compare with what comes out of an analogue chain in the same circumstances?

Let's face it we don't all listen to 0dB all the time.

And speaking of what we listen to, those brick wall filters... yikes! Look at at that same test link above. That ringing we see is around 9-10% of the total height of the wave form. That's 9-10% distortion... of some kind.

And this is the GOOD digital filter (no - or not much - pre-ringing).

So, I'm not saying digital's garbage; I'm not saying analogue is way better (or way worse). I'm just saying the digital testing done by Stereophile shows us the kind of distortion we can expect with digital music, and I'd like to see how the analogue gear relates to that. What is the impulse response of a Lyra Titan? How does its waveform appearance compare to a digital rendering of the same kind of impulse?

Maybe this kind of comparison could help us understand what the differences are...

pt11055's picture

"But how much distortion (THD, IMD, TIM, whatever) do we see in a 14bit DAC, which is what our 16bit DACs are when they are dealing with certain parts of our music?"
Not much understanding of distortion here. The THD of a 14 bit DAC isn't at all like the THD of a non-linear analog system, IMD is very low, but not the same spectrum as analog IMD, and there's not TIM.

A 16 bit DAC is never a 14 bit DAC. There are very few 16 bit DACS used today, the tech has moved forward.

And how does that compare with what comes out of an analogue chain in the same circumstances?

"And speaking of what we listen to, those brick wall filters... yikes! Look at at that same test link above. That ringing we see is around 9-10% of the total height of the wave form. That's 9-10% distortion... of some kind.

And this is the GOOD digital filter (no - or not much - pre-ringing)."

There's even more ringing in analog recording systems, you just never see them tested that way. Run a 1KHz square wave through a (modern!) 16 bit system, you get a 1KHz square wave out of it with a bit of overshoot and ring. Do the same thing with an analog tape machine, you get nothing like a square wave out of it, yet you still get overshoot and ring.

I'm surprised people are still griping about a brick wall filter when digital oversampling filters have been the norm for decades. Even if you had a perfect filter with no in-band phase shift there would be ringing, pre and post, by virtue of the removal of harmonics. That doesn't mean it's wrong, it means it's physics.

Really, just try to play a square wave test record and see what you get. It's a complete mess.

MonetsChemist's picture

"not much understanding of distortion here" - I think it's safe to the misunderstanding is actually on your end of things.

A 16 bit DAC only provides 16 bits of resolution at 0dB. It provides 15 bits of resolution at -6dB, 14 at -12dB and so on. We can see the end result of this in the -90dBFS graphs in Stereophile, which don't look much like a sine wave. I don't know what the average signal level of a classical performance might be, but it's not 0dB and it's not -90dB. Maybe -12dB or -15dB? Shaving a few bits off the discrete representation of the signal introduces a few doublings of error into the reconstituted signal.

Clearly the error introduced gets worse as the level of the signal decreases. So low-level signals are going to be more poorly represented by digital systems than high-level ones.

The abrupt filtering which you claim is gone in fact isn't. Take a look at the filter characteristics in the link I posted. The ringing which is a result (in the case I posted, converted to post-ringing by a so-called apodizing filter) is as you say because of physics. And it's a pretty decent example of distortion caused by digital sampling and reconstruction of an analogue signal, which particularly affects higher frequencies.

My point in mentioning these two things is that any kind of 1kHz test, especially a sine wave at 0dB, is a really poor indicator of digital system performance. So THD and IMD figures based on a 1khz sine wave at 0dB are not all that representative of digital performance.

You seem to think you are arguing with me when you say "there's even more ringing in analog recording systems". You aren't. I'm not trying to say one is better, one is worse.

The whole point of my post is to suggest that we take a look at some of the small signal characteristics of a decent turntable - cartridge combination and compare it to similar decent digital front ends to see what we can learn about their specific characteristics and maybe to begin to understand why some people like analogue front ends more than digital.

pt11055's picture

Your analysis of how quantization and reconstruction works is, well, incomplete. Remember, I've mentioned dither? It changes everything.

I suggest you ignore the so-called -90dBFS graph in Stereophile and absorb a bit of reality before we go on from here:

Jon's picture

That video is an extremely lucid and easy to understand explanation of the principals of digital audio. I look forward to watching his others. My concern as it relates to dither, however, is that in all the years I have been using dither algorithms (ever since I began my work in 24 bits), I have never found any dithering process that is actually sonically transparent (verified using blind testing). Yes, they all look great on paper and I am sure they would breeze through practical tests such as the type shown in the video you linked to, but when it comes to actual blind testing classical music, they all fail to a certain extent, particularly in terms of being able to retain the accurate timbre of stringed instruments, particularly violins. I have tried the very best dithering processes and they all degrade the sound - especially violin sound. The best I have ever come across is PSP X-Dither, but even that is a "sins of omission" process rather than a completely transparent one. In my experience I much prefer to live with whatever the original noise floor is based on the bit depth of the original recorder / recording. To me this is much more transparent, but then of course there are the practical advantages / necessities of working at higher bit depths during the original recording and subsequent mastering processes. Some of the very best (and most "transparent" digital recordings I have heard are actually the earliest ones - back in the days when classical music was originally recorded at either 48 kHz 16 bit, 44.1 kHz 16 bit or perhaps 48 kHz 18 bit. Whatever noise there is bothers me no more than it did in the analogue days (i.e not much). It seems to me (again based on experience) that the more sonically "high tech" recordings become (higher sample rates and higher bit depths), the worse the final result is when it ends up on a CD. I believe this is because the resampling and dithering processes are not sonically transparent, whereas in the older days of analogue such processes never occurred because the equipment was not capable of anything much more in the first instance. I have a CD released a few years ago that was an audiophile reissue of a Decca recording. Kyung-Wha Chung on violin with piano accompaniment recording by Simon Eadon using a humble DAT recorder in the late 1980s. It is one of the very best recordings I have ever heard. Never have I heard a modern "resampled and dithered for CD" digital recording that can hold a candle to it.

I know most will disagree with me and say that in their experience dithering is completely transparent so long as the reduced noise floor is below audible thresholds. As I say, this is has never been my experience (as much I would like it to be), however I am happy to become a believer if I ever hear a dithering process to 16 bit that is sonically transparent when it comes to reproducing orchestral music, particularly massed violin sound.

pt11055's picture

At the risk of going way off-topic, are you attempting to use a dither signal during the process of truncating to 24 bits to 16 ? Isn't there likely a sample rate change in there too? Hardly anyone recording in 24 bits is content to do so at 44.1, they will likely attempt far higher sampling rates. If so, you may be facing more than just the effects of dither, but sample rate conversion too. But I'm not sure why anyone would want to dither even a 24 bit recording even if converting to 16, unless the total system noise floor were below 15 bits, which is highly unlikely, especially with classical music.

Early digital classical recordings were done at 50KHz/16 bit, resampled to 44.1/16 for CD. Very little early digital was done at 48KHz, there really wasn't much gear that would do that, and unless the recorder were a multi-track and a mixdown were planned, digital sample rate conversion would have been required. The most common system was the Sony PCM 1610 thru 1630, which was 44.1/16 of necessity, and was the basis for most CD recordings. Telarc did their earliest work with Soundstream at 50KHz, but early sample rate conversion was inherently lossy, and added at least 3dB of...guess what...noise! DASH recorders could run at 44.1 or 48, but weren't introduced until 1982, and were typically mixed down to 2 track stereo through analog consoles to a PCM-1630 at 44.1. In other words, analog sample-rate conversion and analog dither. DAT recorders could be run at 44.1 or 48KHz, but anyone using them for CD release would most likely stick with 44.1 and avoid the conversion. Regardless, all digital recording systems were all dithered, if not purposefully, then by virtue of noise in the system. Any recording involving a microphone and preamp has been dithered accidentally with preamp and room noise which lands above the first couple of least significant bits. Even 24 bit systems are accidentally dithered because none of them have anywhere near real 24 bit noise floors, though if they manage to hit 20 bit noise floors (not just the A/D, but the entire recording...not very likely if mics and preamps are used), purposeful dither would be appropriately added for 16 bit conversion. The only instance I can think of that would actually require that would be a recording made at an artificially low level such than system noise is attenuated below 15 bits. That would be a very quiet recording.

Be cautious of two things when comparing dithering signals: first that you are only comparing dithering signals and not sample rate conversion too, and second, that you do a true double-blind test with adequate trials. Small difference comparisons are the most easily biased, even in what appears to be a blind or single-blind test.

Jon's picture

Thanks for the comments but I have been involved in recording and digital remastering and preparation of CD masters on a semi professional basis for 7 years now. Before this, I worked with analogue equipment for 30 years. I am well versed on the effects of resampling and dithering. I have literally spent several hundred hours over the last few years devoted to absolutely nothing except attempting to produce a sonically transparent CD master from a 24 bit source, yet have been doomed to failure (to my ears on my system). As I said in my post, the difference is in the dithering, not so much the resampling. I can resample with a reasonable degree of sonic transparency (again, it is a sin of omission process rather than an objectionable one), however when it comes to reducing word length, I have found it completely and utterly impossible to do so transparently unless the target word length is at least 20 bits. I cannot achieve a transparent result with 18 bits of less. My comparisons are done at a constant sample rate and ensuring all variables are completely eliminated bar the bit depth. Whilst it is my experience that at much higher sample rates (88 kHz and higher), a reduced bit rate has a far more benign impact - to the point of being transparent in many cases), there simply isn't the headroom at 44.1 kHz or even 48 kHz to work with.

So far as 48 kHz digital recordings are concerned, I was referring to Decca, who made a large number of early digital recordings with their proprietary 48 kHz digital recorder. 50 kHz recordings were actually the exception in terms of volume of commercial output, being the domain of Soundstream users such as Telarc.

pt11055's picture

That's an impressive career.

Not doubting what you've done, just asking, were your comparisons double-blind?

If I understand you correctly, you were converting from 24 bits to 16 bits, constant sampling frequency, (you would be changing the bit rate doing that however), with and without additional dither? And you found that adding additional dither resulted in less transparency vs not adding it?

And why exactly were you attempting to apply dither in the first place when it's likely already there?

BTW, sampling rate doesn't affect headroom, neither does bit depth provided reference levels remain constant. Perhaps you meant bandwidth. Bit rate is bit depth times channel count times sampling frequency, resulting in bits per second. Any of these conversions would result in a change in bit rate.

Jon's picture

I use Sound Forge Pro and Reaper on my DAW for editing and recording respectively, however I use the Foobar ABX comparator whenever I make any changes whatsoever to my hardware, software settings or mastering methodologies. That way I can compare my output prior to whatever changes I am proposing and afterwards so as to ensure I am not imagining things (assuming I believed I heard an improvement). I require myself to score 9 out of 10 minimum, the reason for the 1 point concession being that some changes are incredibly subtle (such as a less than 1% change in resampling steepness curve) and thus very fatiguing to consistently observe. Others such as dither algorithm settings are very easily observed. I do the testing using a Musical Fidelity M1 headphone amplifier and Fostex TH600 headphones.

As for the question regarding why I would dither when dithering is likely already there (your words, not mine), I don't. And I never said I did. I only dither at the very final stage of an entire project when preparing a CD master or for (whatever reason) the project requires a 16 bit output (such as 16/48 DAT). As would any professional audio engineer worth their salt. If I don't require a CD master the output remains at whatever sampling rate is required and at the original 24 bit depth as there is obviously no reason whatsoever to introduce dither as I am not reducing the word length.

I'm not sure you understand my comments regarding "headroom". What I am saying is that at low sample rates, the effects of noise shaping are quite obvious as the noise cannot easily be shifted to well out of the range of musical content. It is too close for comfort. If you were to listen to the output of my DAW on something like the extremely revealing Fostex TH600 headphones, you would easily be able to hear how the more aggressive the noise shaping curve is (and the more the noise is shifted into the higher but still audible frequencies), the coarser the sound becomes and it is particularly unfavourable to the first violin section in orchestras. On the other hand, conservative, flat dither is much more benign in it's effect on the music itself despite the theoretically higher noise floor. I am not the only person to think this way. Bob Katz does as well, for example. Incidentally, he is the reason I went to using the PSP-X Dither algorithm in favour of my previous Izotope, since the effects in my opinion are less destructive in the case of orchestral music. When you start using much higher sample rates, then if there is a conversion to 16 bit with noise shaping you can shift the noise to a much higher frequency to the point where the effect is well nigh close to inaudible. In other words, whilst I can easily distinguish between an original 24 bit file and a dithered 16 bit equivalent at low sample rates, I cannot easily (if at all) do so if the sample rate is much higher (say 88 kHz and beyond) but only so long as the resulting noise is shifted far beyond both audible and musical frequencies. Mind you, this is mainly a theoretical argument because most dither algorithms intend for the noise shaping to be optimised at 44.1 KHz which is the most common sample rate employed with 16 bit audio.

In the final analysis, I am happy to trust my trust my ears, not only because of my long experience and background as a violinist, but also because I am also one of the relatively few who blind ABX to make sure there really is a difference. There is no point throwing all this digital theory around when quite often it never comes together in actual practice when listing to actual musical content. There will always be people who say 16/44.1 is audibly transparent and "perfect" and good for them. If only I could not hear the difference as well, my life would be a lot simpler and the equipment would be far less demanding on my wallet. I would also have saved many thousands of dollars on high resolution downloads and all the audiophile vinyl that Michael reviews and gives his thumbs up to. Not only that, but I would also have saved a good chunk of my life over the last few years in my fruitless endeavour to create a sonically transparent 16/44.1 copy of either a vinyl LP or a 24/96 digital master (though I would still venture to suggest I have succeeded as well as the very best of them). I trust what my ears tell me. Not what other people (save for a small handful of professional engineers who I have great respect for, one of whom posts and reads this website regularly), text books and YouTube videos tell me.

pt11055's picture

My only comment here is that you may not have to dither at the final stage when transferring to 16 bit for final output. I am well aware that any engineer "worth his salt" does that, but you should check your system noise floor first, that's the noise floor with mics live, no music played. If it's anything less than -90dBFS, dithering is not required. If it's not required, then your dither/transparency issue goes away.

When you use Foobar to ABX you are not comparing just the difference in the two source files, you are including your system's digital path through the D/A converter, and including the converter's ability to handle changes in sampling frequency, which dictates a change in its reconstruction filter as well. An ABX score of 9 out of 10 is excellent, though the nit-pickers would point out that 10 trials is a few shy of the minimum statistical requirement.

I appreciate your sharing your observations, they are some of the most carefully done I've run across so far. You sit in a unique position of having original high-rate source material to experiment with. Thanks.

Jon's picture

I have to say that you are continuing to dispense advice that is not only bleedingly obvious, but completely un-necessary. I don't need it thank you. I am fully aware of noise floors and their relevance. In any event, we are going to have to agree to disagree about what processes may or may not be necessary based on the existing noise floor. I would rather use my ears than follow some dubious digital theory that is not borne out in practice when actually listening to music itself. I am not a beginner in digital audio - I've been hard at it in the nitty gritty of it for 7 years now.

As for your other comments, they are not correct. I am extremely painstaking and thorough in my ABXing to eliminate all variables except the specific one under test. If, for example I am testing the sonics of a new analogue interconnect between a CD player and pre-amplifier (just as an example), I will firstly do a "sighted" listening test to obtain a subjective opinion. If I believe there is an improvement (and only then), I will then record at 192/24 the output from my analogue pre-amp before and after installing the interconnect and then I blind ABX the two resulting output files untouched, except to ensure they both start at the exact same sample and end at the same sample. These two files will consist of a particular test suite of 60 musical tracks of around 1 minute each (orchestral) I have specifically chosen over the years to expose weaknesses in audio equipment. These tracks are the type of tracks that sound horrible on a less than perfect system but tolerable to very good on a well setup system. No point using tracks that sound great on everything. Anyway, the two output files are therefore identical in all possible aspects except that one was made using the new cable and one with the old.

As for differing sample rates in Foobar ABX, I never bother ABXing different sample rates these days because the differences are so obvious no blind testing is needed. And if I can comparing a 24 bit file to a 16 bit file, I make sure the sample rates are identical and that if a sample rate conversion was performed previously, precisely the same settings were employed.

Bottom line is I always make 100% certain I am testing only one change and that the change in question is the only variable between file "A" and "B". When everything else in the entire recording / playback chain is otherwise kept 100% the same, whatever components / settings used elsewhere become irrelevant so long as they are indeed consistent - which they are. Of course, it does matter that the entire chain is very revealing. If it isn't, I can't even tell the difference between 24 bit and 16 bit and ABXing becomes pointless. That is why I have set up my recording and playback chain to be as revealing as possible, even though it means that bad recordings sound bad rather than glossed over.

So please, if you are going to critique someone or give out "advise", please do so to people who actually need it and ask for it (i.e not me on either of those accounts).

Michael Fremer's picture
For your comments here. They are very helpful...
Michael Fremer's picture
blah blah blah.... I was challenged on amplifiers all sounding alike. Took an a/b/x test at an AES and got all identifications correct. The results overall though were statistically insignificant (even though one amp was a tube amp and one solid state that are EASILY audible). My result was tossed and I was declared a "lucky coin". Double blind testing is useful in some ways but a trap in others. It is an acquired skill set and when unskilled people who don't know what they are listening to take them, the results are often stupid and prove nothing. I don't like listening to 16 bit digital and never have. I don't need a double blind test to know that my ears and brain don't like it. And higher resolution digital though better still doesn't please my ears and that's what I be a fact.
Jon's picture

As I mentioned in much greater depth in this comments section, I am not a 16 bit fan either. In hundreds of hours of experimentation I simply threw up my arms and simply gave up. I can tolerate audio down to 20 bits and that is it. And even then I have to be exceptionally careful in the parameters I select in order to make a 20 bit file sound "OK". But once you drop to 16 bit (or even 18 bits), all bets are off. It is like trying to fully repack a previously opened army ration pack (i.e impossible, except for those unheralded geniuses at the clandestine facilities that packs those gastronomic enigmas in the first place :) Barry has said it numerous times - CD is the cassette of digital audio and as much as I wish it were otherwise, I sadly could not agree more.

Seriously, I wish CD had been delayed until we could have fitted 48/20 for an hour's worth on that 12 cm disc. It might not have made much of a difference 30 years ago but we'd sure be appreciating it now. It still might not sound as good as LP or open reel, etc, but it certainly would not actually be downright offensive.

Anyway, the argument is pointless since we have what we have. For my own part, the audio "fact" of the matter is whether I am reaching for the "off" button 20 minutes into a listening session or whether I am putting on LP after LP on my turntable and listening until 4.00 am in the morning. I can tell you what is far more commonplace in my household.

Ajcrock's picture

Michael is correct is saying that records don't ring. Bad phono preamp design does, though, produce harmonics just like any amplifier. This is caused by inferior amplification chips or a mismatch in the input impedance of the opamp and the selected load or impedance of the cartridge, which should be at least 10:1. I can back this up by stating I have performed tests on many opamps, and other forms of amplification. Using a spectrum analyzer, which is the correct way to measure this, one can easily see the harmonics. This is something a scope cannot do.
Besides, who listens to a square wave anyways.

pt11055's picture

...but have you actually looked at the output of a properly loaded cartridge into a properly designed preamp when playing a square wave test record? I wouldn't expect the cartridge or the preamp to ring...much...but test the entire chain once. And, if you include the analog recorder(s) used in the recording and mastering, you will have ringing, I guarantee it, lots of it, and worse than typical digital system.

The comment, "who listens to a square wave", clearly, nobody. I'm surprised that someone familiar with testing enough to have used a spectrum analyzer would even say that. It's a test stimulus designed to reveal transmission flaws. Do you listen to sine waves for entertainment? Pink noise? But how else would to make a good frequency response measurement?

christopherf's picture

Submitted by christopherf on Sat, 2015-01-31 13:37


This has to be the most unbalanced article; perhaps you will write a rebuttal in the form of a letter to the editor. Bob Clearmountain and Bob Ludwig are referenced in the piece; it seems such an old argument; I thought each format has its' own strengths but this article is very patronizing; it says what vinyl lovers are enjoying are "distortions"; the tone is patronizing; you can like vinyl but don't say it is better than CD; indeed!

Here is the link to the piece:

I am amazed that you have not grown weary of addressing this with the "experts" in this case the engineers; I regret that most people who read this article will take it as gospel.

Michael Fremer's picture
The writer gets the "Music From Big Pink" story completely wrong... too bad for these people because the vinyl train has left the station!
wolfman's picture

for days now my kids a listening to "Heidi" (a fairy tale). The records are played through a Nagaoka MP100 cartridge (100$). This cartridge is not able to pull all the highs and lows and get all the microdynamic right, it cannot put the music on a big stage etc., all the things we HiFi nuts are after, but still, more than once approaching the listening room fooled me that I thought real person are talking. We have the same on CD. Which in HiFi terms betters the sound of the Nagaoka MP100, but it never happend with the CD that I was fooled.
So you tell me what measurement(s) are responsible for this kind of deception. It is not a fairy tale what happened to me:-)

Ajcrock's picture

For those who have evidence of ringing please feel free to post pictures. I would love to see the photos. Please also include the opamps used in the amplification, loading, and spectrum analyzer results. Given a high quality phono preamp and correct loading, I have yet to see ringing. I can induce it using poor opamps or loading but not see it naturally occur in a high quality phono preamp. Also a square is not necessary. You should be able to use a 1khz test tone. But make sure your test tone was not recorded with harmonics.

cundare's picture

I have so much to say on this topic, but ya know what? I'll limit myself to just one comment. Michael, as a journalist, has outdone himself with this piece. Not because I agree 100% with his assertions or his references -- pencil me in at about 85% -- but because this piece spawned one of the most provocative discussions I've seen of such a nuanced and complex issue. When TechNet, Gizomodo, or even CNET (one of my ex-employers) covers this ground, you know that the inhouse author and the posting readers are likely to have no idea of what they're talking about. This page, though, despite the fact that I disagree with a respectable minority of the posters, was a breath of fresh air. Thank you again Michael. You remain one of my all-time favorite journalists.

Deli's picture

I hate to tell you all this, but vinyl is not the wave of the future. Can you admit to yourself you are investing hundreds, if not thousands, in a technology that will probably be dead in the next 20 years? It’s hard, I know, I look at my shelf of CDs and know those will soon be nothing more than fodder for my digitizing computer.
You don’t see Picard, busting out a vinyl record on the Enterprise. I’m not saying CDs or hi-rez digital downloads sound better than records; I wouldn’t touch that topic with a 10 foot pole on this website, but you must realize you are on borrowed time. But, really, come on, you know I am right.
Enjoy it as long as you can, and really do enjoy what ya got, but vinyl isn’t the long term solution. You can collect, and enjoy, and then even somewhat digitize your collection, but it isn’t going to last. Do CDs have flaws, yes—totally; do hi-rez downloads have flaws, yes, but less than CDs? Do I know what the future holds—no! But I can say, with decent certainty, it is not going to be a 12 inch black platter.
At least what I have can be cleanly uploaded into my computer, free of DRM, and I can continue to enjoy it without a subscription. Don’t get me wrong, I have no animosity. I remember hearing Lalo Schifrin on Dad’s Garrard turntable and being blown away, but I’m not going to invest money on a dying dog.
Eventually a standardized clean, clear, and hi-rez format is going to rear its head and you are all going to feel like goobers holding your unwieldy disks reminiscing about the past. So spending money on turntables and disks seems as long term logical as throwing it in a hole.
Something great sounding is eventually going to come to fruition with the ability to pause, rewind, resume, and skip—something sorely missing with vinyl. Those deficits make it instantly antiquated. Wait a moment and save your money for the next standardized format.
Instead, devote your money on better amps and speakers because those seem to have greater enduring value.

isaacrivera's picture

Us vinyl fans know very well that it is not "the wave of the future." It has been "dead in the next 20 years" for the last 30 years already.

Michael Fremer's picture
The billions of old records and new will forever be playable. It is hardly throwing money down a hole. Ironically it is the very fact that you can't pause, rewind, resume and skip that is producing another generation of vinyl fans for reasons I guess you can't comprehend. Your post is a useless attempt to turn people away from what is a trend gathering momentum. What are you afraid of?

The notion that everything that is new is better than everything that came before is ridiculous on its face.

However since that is what you are claiming, perhaps you can explain why CD enthusiasts are so dead set against higher resolution digital audio?

PureSilver's picture

First, Dr. Fremer, we met back in the '90s through the pages of Stereophile and I want to say thank you for many enjoyable and informative articles (“Don’t Be a Jerk” is my personal fav). More so, thanks for being a big part of me getting an LP in my system.

As for the topic at hand, I know I'm a little late to the show here, but... I'm an EE whose focus was in Digital Signal Processing (DSP), though I'm not active in the tech stuff anymore. I studied DSP largely because of my interest in Hi-Fi and CD players in the late '80s. I learned more about digital signals than I care to recall and I know and experimented digital limits- ringing, harmonics and artifacts that don't make much sense to ears but are easily explained with Fourier transforms and abstract math. I used to go to the lab and manufacture different kinds of digital signals because I wanted to see how they would sound- kind of like Jody Foster listening to electronic noise from washing machines in the movie "Contact".

Anyhow, fast forward a few years and I had a subscription to Audio Electronics in April of '96 (along with Stereophile and Glass Audio) and read the original printing of Mr. Bauman’s article. That article was a game changer for me- the math made sense, the electronics made sense, and the statistics made sense (though I don't believe the results have a high statistical confidence- but they weren't represented or intended to. I didn't 'buy off' on LPs because of an article though- rather I gave the medium a credible chance. It didn't take much time to confirm with my ears what Mr. Bauman’s article presented as a possibility.

So here I sit today with a CD transport (sorry Michael) and LP in my system. I enjoy my CD transport and external DAC to this day (in fact right now). Nothing is quite like a media that you can damn near play Frisbee with and then play through a Hi-Fi setup. The CD is for comfort and convenience and sounds good. My LP however is for when I want to really experience music, because it really does sound better (pops, noise, dynamic range and all). I don't know why it sounds better and frankly, after my almost Tron-like deep dive into the digital signal domain, I don't friggin care to know or explain it. My ears tell me vinyl sounds better and my ego about the DSP certificate resting on top of my EE degree is secure enough to not try to convince my head that a (at least my) CD Transport sounds better than an LP because I understand the theory at each and every step that the 1's and 0's take on their way to my amp...