The provenance issue that plagues hi-res audio (HRA)


Riddle me this: what is hi-res audio (aka HRA)? 24bit/96kHz digital audio? Nope. How about DSD? Wrong again. Vinyl? Good luck with that. Alright – give up?

Answer: high-res audio (HRA) is just a marketing term – nothing more, nothing less. That’s right, a marketing term. There is no widely accepted official standard that actually defines HRA. So technically speaking, HRA is simultaneously all of the aforementioned formats…or it’s none of them, depending on who you talk to.

Loosely speaking though, when audiophiles talk about HRA, they invariably mean digital audio with a higher sample rate than the CD’s 16-bit/44.1kHz. Even 24bit/48kHz qualifies. This presents a problem in that it presupposes higher sample rates automatically mean higher resolution and therefore higher sound quality, which a lot of the time just isn’t the case.

Now I know what you’re thinking: this is the part where I get on my high scientific horse and talk about the glory of Nyquist followed by some diatribe about the audiophile world’s arch enemy, the double blind test, and then perhaps I close with a few links to a Monty Montgomery article from Cue mic drop. Sound familiar?

Good news! That isn’t where this writer is headed. My biggest beef with the HRA movement isn’t whether or not I believe there are advantages to higher sampling rates, but rather their fixation with the sampling rate of the playback file and not the entire recording process. Because in the end, it is a master file’s source that ultimately dictates its fidelity.

I am however a true believer when it comes to high-res audio; not because I think we can magically hear more than 20Hz-20kHz, but because there are certain intrinsic advantages to both playback and recording where greater bit depths and higher sampling rates are beneficial. In order to reap those rewards, the entire production chain must be held at those higher resolutions. And if at any point in the recording process a lower sample rate or bit depth is introduced, the resulting studio master file could (should!) reflect that change.

For example, I think even the most staunch HRA evangelist will agree that just taking a 16-bit/44.1kHz file and running it through an up-sampler isn’t the path to salvation. If the source material in question was recorded at 16bit/44.1kHz but then mastered at 24bit/96kHz, is it still HRA?

Do you remember that new high-res remaster with which you were so enamored where the label dusted off the original analogue tapes and paid some engineer megabucks to do a high-res transfer? Do you really believe those old Ampex tapes taken out of the label’s vault are a true high resolution format? If not, how can the resulting digital audio master file created from those tapes also be considered HRA?

Finally, what if an artist records their next album at two different locations and track some songs in standard-res and others in hi-res? Now the mixing engineer has to upsample the standard-res ones to give the mastering engineer a “high-res” master. Is that still HRA?

The truth is none of the examples above really qualify as HRA, either because the source didn’t begin life as high-res or because a lower-res signal was introduced into the chain.

Yet these same scenarios play out time and again. It’s hard enough for an artist to make a hit record let alone micromanage their entire production chain. Besides, HRA is often an after thought, tacked onto the end by the label as a means to an extra buck. And because there is no widely accepted formal definition of HRA, they are allowed to call that 24-bit/192kHz master “high-res” even if the original recording was not. Awesome? Nope.

Ironically, the music industry realized that this was a major issue back in 2014 when a bunch of consumer electronic organisations and record labels tried to formalise the definition of HRA, which included tackling the issue of provenance head on. Those efforts begat the MQ system (no relation to MQA) that specified from what a given release’s master was sourced. Is anyone actually using the MQ system today? I don’t see much evidence of it.

Recently, the RIAA tried to address the issue of provenance by introducing a (ahem) “gorgeous” logo for use by artists and labels to certify their recordings as truly hi-res. HRA was defined as follows: “lossless audio capable of reproducing the full spectrum of sound from recordings which have been mastered from better than CD quality (48kHz/20-bit or higher) music sources which represent what the artists, producers and engineers originally intended.”

Perhaps you’ll agree with me that this is a very vague way of defining HRA and leaves open loop holes aplenty for labels to exploit in order to apply the HRA logo to their releases.

Maybe the industry doesn’t really want to standardize HRA because it would invalidate a lot of the releases presently being pawned off as hi-res.

Think about it for a moment: what if tomorrow, all the big record labels and hi-res providers all collectively agreed that for a release to be truly hi-res it had to be sourced, recorded, and produced at a sampling rate equal to or higher than the file being sold in download stores? Do you think Led Zeppelin’s back catalogue would qualify for HRA status? How about Neil Young’s?

Here’s what I always tell my fellow headbanging audiophiles when it comes to HRA: proceed with extreme caution. The conditions from which we might hear the benefits of HRA are rarely met, even with many a popular modern release and especially if it’s a remaster.

There are exceptions, and they do indeed sound magnificent for the most part, but unless the master’s provenance can be verified, I’d advise readers to empty your HDTracks shopping cart and back away. Because unless the industry truly standardizes the format, and high-res vendors strictly adhere to those standards, HRA is like a box of chocolates. You never know what you’re gonna get.

You can read more of Alex’s thoughts on audio over at his own Metal-Fi.

Written by Alex M-Fi

Alex M-Fi

Alex is co-founder and Chief Editor of, a website dedicated to the head-banging audiophile. His blood type is Type O Negative. Alex derives his income from writing software.


Leave a Reply
  1. Hi Alex, besides marketing reasons and creating endless opportunities to re-sell musical libraries by record companies, there is also technological progress. Untill the recent introduction of the MQA format, many audiophiles and specialized HD selling platforms have been focussed much towards sampling density and decreasing jitter with improved DAC solutions. In the race for better content performance some HD albums are proven to be just upsampled frauds..
    But now a new technology has arrived, which shows us that there has always been a much more audible negative effect presence on recordings AND mastertapes..! Time-smearing induced by old 16/44 analog to digital converters. Luckily enough the pre- and post- ringing of those ADC converters are known and can be compensated for by using a smart DSP algorithm, which is exactly what MQA is doing. I have personal listening experience with the strong improvement of the original recording this causes. It is much more significant than increasing the bitrate of a digital remaster of the original mastertape. The MQA dsp software is actually reconstructing the original sound before it was stored on the mastertape and this is a great technological achievement which, hopefully soon, all of us are able to enjoy. It will probably not be the end of HD audio. Like with video and digital photography, 3D etc. We have seen and heard it all before. But for now, I am confident that the RIAA committee made the right move (except for the logo..)

    • I don’t think MQA is going to make the major labels suddenly become honest about provenance. They don’t care any more and will most likely continue the deception. The way most business now operate is to make the most profit with least amount of effort and when people ask questions you lie and make excuses.

    • Let’s just say I have a wait and see approach when it comes to MQA. And Peter, think about this: Do you want your entire production and playback chain beholden to one company that owns the license for it?

    • Peter,
      I just got back from T.H.E. Show in Newport. I’m one of the few who got an A/B test of The Doors “Riders on the Storm” in Meridian’s room. The MQA version was different than the 24/96 version. But it was off. A Fender Rhodes doesn’t reproduce rain accurately and others who have heard the MQA track confirm the difference. I would like an explanation of how the rain changed.

  2. Speaking of ‘hi-res’, which pencils are considered best for winding tangled tape back into cassettes?
    I’ve always used a Faber-Castell but Nobby down the King’s Arms swears by a Staedler…an’ he should know; he’s one of them ‘ordiohfiles’…

  3. I have purchased 25 albums of music that I love via download from HDTracks and Acoustic Sounds in “HRA” from 24/48 to DSD, more as an experiment than anything else to see what value I perceived for this model of music purchase.To my ears (old) and with my gear (decidedly mid-fi), the best sounding one of the bunch as compared to flac cd rips of the same music was the one that was 24bit/48khz, which is Donald Fagen’s “The Nightfly.” I believe this is due to a combination of two factors; the care taken with the original recording, and the quality of the composing/performance (which is completely subjective of course). This album sounds better from an mp3 file @ 320 kbps than many albums sold as “HRA.” My takeaways: for my purposes used cds ripped to a lossless format are the enjoyment per dollar leader, and the subjective qualities of the music/recording are far more important than file format. I imagine the latter will always be the case, as if I don’t enjoy the music, I don’t give a rip about the “sound quality,” whatever that is.

    • The Nightfly was recorded on a 3M digital machine at 12 bits. 24 bits means you got taken. This album is proof the people at the front end are the most important thing. In this case Roger Nichols and Donald Fagan. Roger did an analog, digital, live shootout with one of the tracks, digital won so the album is digital. It was more realistic than analog.

      • The Nightfly was absolutely NOT recorded at 12-bits. It’s a full 16-bit recording, as that 3M deck was the first 16-bit multitrack digital recorder. It was so early in the history of digital recording that 16-bit analog-to-digital converters weren’t even available yet, so 3M used a 12-bit and an 8-bit converter, used 16-bits for the audio and I think the other 4-bits for parity or something. Anyhow, unlike some of the early units that were 14-bit (Sony’s first), it was a full 16-bit device running at about a 50kHz sampling rate – higher than CD or DAT.

        There’s an article about the multichannel remaster of The Nightfly done in the early 2000s that touches on how that 3M deck worked here:

  4. Alex,

    Thanks for bringing this topic of provenance forward. But, the issues extend well beyond HRA.

    It’s my opinion that there is genuinely a need for a standardized, unique identifier, that can be associated with a release master; both at a track level and at an album/composition level. Such information needs to be included with the info/labelling/tags associated with a release and maintained in a directory/database that is publically accessible. By virtue of a release mastering process, the output is something to be consumed. The consumer needs to be know what there are dealing with.

    This is not just an HRA issue. Not that I want to pick on TIDAL, or any of the other streamers, but I find albums being replaced in their catalog all the time. Particularly, when something like a “Complete Studio Albums” compilation/box set release is made available, somehow some of the individual “original” albums available in the catalog become the equivalent to a volume in the compilation set. How this gets noticed is the fact that in many cases the compilation is re-mastered in some form. When the substitution occurs on the original album and it’s not labeled as such in the catalog, it represents to me that their is more focus on the release name, artists, composers, etc attributes (which haven’t changed) and not the technical details of the release.

    Primarily, it’s my opinion that the release “labelling attributes” align more with the revenue and royalty streams than the technical details of the release. These alignments are in the opposite direction of the consumer.

    • I could not agree more. The box set issue drives me nuttin frazy.

      I will say this: The problem is more acute with HRA because the claim is higher sampling rates equals better sound and that just isn’t the case most of the time.

  5. This article reminded me of the parable “the King’s new clothes!”
    Is Hi Res audio the new snake oil in the audiophile pantheon of products?
    Or, are we just driven to have the latest, greatest piece of equipment or system by reviewers and magazines?
    I want to know when did we stop listening and enjoying music and start listening to equipment?

  6. Hi, Alex,

    I thought that these matters (circling around the HRA) were already sorted out?

    Of course, music recorded at lower resolutions and then up-sampled to meet the official HRA specs IS NOT a true high-resolution.

    About 18 months ago, I tried to do some testing and converted CD rips to DSD with the Korg Audio Gate application. I used 3 different CDs: one jazz record with exceptional mastering (from ECM label), one run-of-the-mill soul music sample (analog recording digitally remastered) and one re-issue of old (pre-WWI) solo piano recording.

    After comparing the original CD rip with the DSD converted file (using exactly the same 2 audio setups – one with speakers and one with headphones), I came to this conclusion: none of the recordings sounded better or worse after conversion. The only difference was the file size on a disk.

    Unless the entire process of recording/mixing/mastering is done in high resolution (at least 96/24), the HRA moniker is a bogus term meant to deceive consumers. That means, no old 2-track or 4-track tapes should be considered as a “high-resolution source”, even if they were recorded at 15ips.

    I have stopped obsessing with “high resolution” a good while ago, because today’s music (truly hi-res) does not appeal to my taste in general, with very few exceptions. And I can still find great music recorded in analog technology and mastered in an artistic way.

    • Wise man. I do think well mastered material in HRA can sound absolutely top notch and frankly, I see zero reason to dither down in this day and age.

  7. Vinyl. And CD’s on sale. Done. I don’t mess with 24/96, downloads, dsd, MQA, or whatever the voodoo of the week is this week.

  8. The last sentence resonates with me. I’ve bought “high res” music from HDTracks and have been dissapointed. An exception would be Aerosmiths Rocks album which was bloody great!

  9. I would agree with Alex that any acronym is to be skeptical about. Take MQA. No, you take it. But we have to be skeptical and fair. HD Tracks had some issues with tunes labeled HD that were not, meaning some were upsampled, but many, arguably, most are HD. For one, some DACs cannot be fooled and they will display HD content as such. Or they won’t. Second, I have heard the differences between 16/44 and HiRez, that latter were “better.” The percentage better is easier to hear on superior systems. Take Jon Strong’s “Bad News on the Mountain”, CD on 44.1 and from Linn, 192K…with the latter you hear “more.” Deeper bass, more explosive transients, cleaner highs, better separation of instruments and layering, more personality…

  10. Hi Alex,

    I really enjoyed your commentary on HRA provenance. However, I have a question on your statement “Do you really believe those old Ampex tapes taken out of the label’s vault are a true high resolution format?”. In particular, is it even relevant to discuss resolution for an analog recording? An analog recording represents the original analog wave form from the recorded event; it’s all there, nothing is sampled. It would seem that the issue of resolution becomes relevant only when the analog recording is digitized by sampling at a particular frequency (i.e., 44.1kHz, 96kHz, etc.). For the resulting digital master, higher resolution would therefore imply more data points and a closer approximation to the original analog signal. Just like a connect-the-dots drawing, more points, closer together, produce a smoother curve. That’s why I always thought of true analog recordings as having “infinite resolution”. Am I looking at this the wrong way? Thanks — HIK

    • Yes. Analog most certainly has limits and I hate to break my fellow vinyl junkies heart BUT modern digital recordings can replicate far more dynamic range, real frequency content (don’t talk to me about vinyl rumble noise as high res content), and just offer more to the recording engineer than those old Ampex tapes on their best day. Don’t believe me ask most engineers.

      But even behind that, all I’m really saying is why is there this pervasive notion in the industry that taking a recording made in the early 50s is somehow going to be transcoded to true HRA? It’s preposterous in my book. The best you can do is get a pristine copy in digital format, i.e. you know the digital side is more resolute so you can ensure you have a high fidelity copy when the engineer performs the transfer – a testament to digital not analog if you ask me.

    • I’d like to break your heart too. Nothing in the phono chain equals the resolution of a CD. Check the signal to noise ratios. One bit equals 6 dB. The issue of resolution is always relevant. I just got back from T.H.E. Show in Newport. In the seminar on specs the most important one is the signal to noise ratio at one watt of output. That one rarely reaches CD resolution, 16 bits in any amplifier. As for your connect the analogy well The Nightfly sounds good and it’s 12 bit digital fools people all the time. And don’t forget most Cd’s were mixed to sound good in car. I had a client in the 90’s that had a soundtrack of his for a movie, he remixed it for a car and my living room. They all sounded great in the proper locations, not so good otherwise. Good people and skill outweigh everything else.

  11. I don’t disagree with you for the most part. Much of this is hype and marketing to make $$. But MOST of the time I do find the hi-res file superior to the CD or Tidal stream. Case and point…while I was reading your excellent article, I did a multiple “blind” A-B-C with Paul Simon’s new album, Stranger to Stranger. I compared the CD to the Tidal stream to the 24/96 download from HDTracks. The hi-res download easily won every time. Its a great album too (in any format)! I certainly buy alot less thanks to Tidal, but sometimes that splurge on the download is worth it. And not even the stereo downloads can beat most of my classical multichannel SACDS (though u can download alot of those now too). My computer dac cant play those multichannel downloads back so I either just burn them to a DVD-A or buy the SACD since my Oppo plays anything you throw at it!

    • All I’ll say is that you can’t compare multichannel to stereo as you have two very different masters there. I do wish bands experimented more with multichannel; it’s still an undiscovered country in my book.

  12. Nice piece loaded with common sense. Do you think MQA might be an extension of the record companies’ efforts to market HRA? Record companies made a bundle when people replaced their lp’s with cd’s. Then they made another bundle when people replaced their first press cd’s with remastered versions. Then file sharing happened and the music business has been in shambles since.

    I always saw HRA as the record companies’ efforts to resell their back catalogue once again, but it doesn’t seem to be working. Here’s where MQA comes in. MQA allows large HRA files to be streamed with less bandwidth. Do you think the record companies are hoping that MQA will entice people who might not be interested in buying HRA downloads to purchase something like Tidal (assuming they make a deal with MQA)?

    • I just don’t know enough about MQA and up until this point I have not seen Mr. Stuart answer some of the real tough questions – mainly revolving around licensing.

  13. John (and everyone else):

    I hear ya. However, I think the “provenance ” and quality of the entire recording/production chain is generally (there are always exceptions) genre specific. Those of us who spend more time with classical and jazz know this. It is in the “modern” genres of pop, rock, dance, and electronic that this is THE major issue as far as something above 16/44 (or even 320 lossy) being worthwhile. I do have my favorite EDM, but the simple fact is there is no reason to purchase it in greater than 16/44 (if it is even available, which it usually is not) because anything more will simply reveal more of the recordings limitations and not bring me any more listening pleasure. For most modern classical and jazz, hi-res (> 16/44) makes a difference because the recording/mixing/master quality is (usually) there.

    I know John and many posters here like to point out the dearth of exciting modern music at shows, but if I was a high end manufacturer I would be hesitant to play too much of this music also because so much of it is so poorly made. I know there are exceptions (I own some of it) but they are the exceptions that prove the rule…

    • I don’t believe in exceptions proving a rule, especially when it comes to demo music. There is music aplenty to be found that’s modern AND sounds good – one must dig for it or go visit Zu Audio. If audio show demonstrators want to play new stuff they must first decide if a broader audience is what they’re after. Some are quite content with the way things are.

      PS I didn’t write this article. Alex did.

      • C’mon John, we all know in Munich you couldn’t get enough of that Norah Jones tune every exhibitor played over and over and over and….or was it Diana Krall? 😉

          • You know I have this weird fantasy where I win the lottery and then immediately go high-end speaker shopping whereby I torture the sales rep by playing one extreme metal track after the next (while I’m sitting there critically listening and enjoying myself).

            OH the fun I would have…

            That list would be so meticulous crafted for maximum obnoxiousness. I mean who doesn’t love Pig Destroyer at a high-end demo?

            Maybe I would stick in a Enya tune in there just to screw with the rep too!

            Sail away, sail away, sail away…

          • My friend did that at Dynamic Audio in Tokyo a year or so ago with some enormous JBLs! Track after track of quasi-EMO-metal (my take) played really loud. Was both amazing and terrifying, especially as my friend’s source was an iPhone!

            Enya? Dude. That’s a four letter word ’round here.

      • Oops, sorry for mixing you and Alex up. I would love a “audiophile SQ & modern” article on your favorites (with lots of suggestions in the comments section 😉 if you have not already done something similar…

  14. A lot I agree with here, but I don’t agree that tape converted to high res digital isn’t hi-res. Because it’s the highest res version I can get.

    I don’t really care about the technical argument.

    I simply know that I have lots of high res material sourced from tape (some of it converted to DSD as opposed to PCM) and much of it is the best sounding version I have of that album – better than the LP, better than the CD.

    All I really care about in the end.

    • All agreed Danny.

      However, I want you to consider that maybe it wasn’t because the recordings were all sourced from tape, but rather the engineers involved took the time and effort to make a really great sounding digital copy of it1 That’s my point.

      Here’s an example:

      Album was originally tracked to analog tape. It gets released on CD and vinyl first with the CD transfer being pushed to be very loud and as a result, loses a lot of its original analog warmth in the process. Vinyl was sourced from the CD master because heck, the label thinks it can kill two birds with one stone. Pressing engineer backs off on the limiting though to get it through the lathe.

      Alright, so now there is this huge online debate of audiophiles who claim the vinyl sounds better (well it does) and that’s due to the intrinsic benefits of the format (it isn’t). And so the story goes…

      Then, HRA comes a long and the label takes those dusty tapes out of the vault once again, does a high-res transfer, and THIS TIME the engineer they’ve hired doesn’t push the master so hard as the CD and in fact masters it for fidelity not sound.

      Did the fact that it was recorded on tape originally have anything to do with why it sounds better via HRA? And moreover, what would have it sounded like if it was all recorded in HRA from the get go?

      This are the questions I just want you to *think* about Danny! 🙂

      • Alex, when vinyl junkies talk about vinyl sounding better than HRA, they’re referring to vinyl cut directly from the analog master tape, not from a CD master. Even with less limiting, vinyl cut from a CD master is really nothing more than a CD copied to vinyl; who needs that?

        To make fair comparisons between vinyl and HRA, the comparison should be based only on vinyl cut directly from the master tape. That will significantly close the performance gap between HRA and vinyl, with vinyl often sounding superior.

        • Ah, a pure analogist!

          To tell you the truth, I would much rather have a pure digital transfer of an old recording for posterity than rely on old tapes that deteriorate after a while.

          I also believe that here, HRA is an invaluable format for archival.

  15. Alex, good points raised. After dipping my toe into a few hi-rez downloads I’m not convinced and now produce my own from a decent vinyl rig and ADC at 24/192. Still ironing out the last remaining pops and crackles but the results are proving far more worthwhile. Although not for everyone excellent affordable ADCs like the Sugarcube bring it all closer to plug and play.

  16. I can definitely say that after purchasing The Van Halen Hi-Res The Collection 192/24 on HDTracks about 1.5 Years ago which I was extremely excited about the studio REre-RE release and boy did this material sound like shit compared to how it was advertised by the studio. Not even sure if this was better then redbook. I emailed HDTracks many times to ask them to explain exactly what was going on and response was mute and just said this is what the studio gave us! I am not saying HDTracks is dishonest since they have no bias except to sell but the studio should be ashamed with this garbage. This MQA (at least you don’t need to buy new material) is suspicious not because of the tech but the studio is not trustworthy!

  17. Great article Alex (simply LOVE your approach by the way to everything hi-fi)

    What’s funny is, (being I live in Australia) I have bought MANY CD’s in my time (vinyl too) and depends on the pressing. It can be chalk and cheese to other pressings (even some 128k MP3’s have sounded better than the CD)

    Some of the CD’s I have bought have sounded flat, like they were recorded from old tape. This is the problem with ‘digital’ you can record a bad quality song off the radio at 192k but won’t sound as good.

    So more the studio that records it and releases where it should be 🙂

    • Thanks Steve!

      I’ll bet though that some of those flat sounding CDs were a product of bad mastering more than anything else. Happens to me all the time – especially if you listen to popular forms of music.

  18. Are there any studios/labels that publish at least what resolution the master was recorded?

  19. I am late to see and comment on this article, but I believe I have found a major flaw in Monty’s reasoning in his xiph article as to why sampling rates about 44.1 khz do not bring any benefit.

    That’s because the Nyquist-Shannon Sampling Theorem requires two samples per frequency to capture the frequency. However greater than (not equal) two samples are required to record the frequency amplitude and timing accurately.

    Here is a visual demonstration by an electrical engineer at Qualcomm on Quora.

    If you read his explanation about the number two regarding the theorem, you can see that when you use just two samples for a particular frequency, while it’s possible to reproduce a sine wave, you can see that they differ in amplitude (volume) and timing.

    If we take 4 samples per frequency as being necessary to more accurately reproduce amplitude and timing information, and divide 44.1/.4, this is 11.1khz and falls well within the audible spectrum.

    I’ll try and get into correspondence with him and mathematicians who are more knowledgeable about Fourier transforms and the theorem than I am (I did maths undergraduate).

    I believe I have finally cracked it, I never believed my ears were deceiving me, and spectral analysis of my Hi-Res files (mostly 24/96) show most of them, especially the ones I most enjoy are recorded at that bit-depth and frequency.

    I need to investigate further, for example Onkyo claims bass is tighter, I hear the same too, but low frequencies are already massively over-sampled at 44.1khz already. I am not researching other explanations such as wave propagation properties (sound is a pressure wave, it’s only a sine wave when electronic). In the meantime I believe I have taken down Monty’s core argument as to why sampling beyond 44.1 does not bring audible benefits and you do not need “Golden Ears.”

    About ABX tests, I believe there is a simple psychological reason why everyone fails them, even in 8bit vs 16bit and mp3 vs lossless. When you perform the test, your core beliefs are being challenged like a religious person fighting to claim God is relevant, and your brain gets into a fight or slight mode sort of when our ancestors were in the African Savannah under threat, and when that is the case, you focus on what you need to do to survive, openness to experience, being in a relaxed state of mind and experiencing the emotions that music generates will get you killed. As for those who don’t already have an opinion, if they are told there is a difference, because of our optimism bias, we will try to focus hard to find a difference or form a certain opinion to satisfy our egos. These theories require further investigations.

    I am currently writing all this for publication, I am double checking the facts first.

    • Correction
      *I am now researching other explanations such as wave propagation
      instead of not

      Can the mod edit my above comment please?

    • You are not even close to making the argument on sampling. Two samples will do job just fine. The theory of digital audio would not work if you are correct and some good people like Roger Nichols a nuclear engineer would have noticed as early as 1970 when they started laying the groundwork for it. He went to engineer some very fine sounding records, see my earlier comments on The Nightfly.

      And everybody doesn’t fail ABX tests. I very rarely failed any or any blind test in my youth for a simple reason, I got paid.

      • I want to post an academic article from an electrical engineer about the flaws in the application of Nyquist Sampling.

        “It is a common misconception that the Nyquist-Shannon sampling theorem could be used to provide a simple, straight forward way to determine the correct minimum sample rate for a system”

        If you look at figure 2 on the paper, at a frequency that is an exact multiple of the sampling rate, say like there’s a frequency of 6 khz for a sample rate of 48khz, you get aliasing.

        By sampling more frequently and using a higher bit-depth of 24 which results in fewer quantization errors , you reduce the chance of aliasing which is what I think results in lower frequencies sound tighter. I need to further investigate.

        What Monty says “All signals with content entirely below the Nyquist frequency (half the sampling rate) are captured perfectly and completely by sampling; an infinite sampling rate is not required. Sampling doesn’t affect frequency response or phase. The analog signal can be reconstructed losslessly, smoothly, and with the exact timing of the original analog signal.” is wrong.

        It’s a mis-use of the Nyquist Shannon Sampling theorem. You need more than double the sample rate to reproduce a frequency, and that reproduces a sinusoidal wave for that frequency, but it does not do so with the correct timing or amplitude. There are other examples in the paper demonstrating this.

        This is his conclusion from the paper

        “Measured purely from a sample rate perspective, increasing the signal sample rate
        will always increase the signal fidelity. It will often decrease the cost of any analog anti-aliasing and reconstruction filters, but it will always increase the cost of the system digital hardware, which will not only have to do it’s computations faster, but which will need to operate on more data.”

        Pcm will never perfectly capture and reproduce the original analogue wave.

        Can someone show me a Double blind audio test with a reasonable comparison between digital audio formats (e.g not flac vs 64kbps mp3) where significantly more than half of participants correctly guessed the correct answer? With Vinyl vs CD, there are obvious cues.

        Since you’re listening to both audio streams at a similar time, your brain is able to intuitively reconstruct or average the two.

        What I think should be done is a test where participants under go an mri scan when they are listening to 16/44.1 audio and 24/192 with both formats in one session mixed, and return a week later to do the same scan, but this time, with the same songs in reverse format, i.e the 16/44.1 not switched to 24/192, with both tracks sourced from the same master.

        • Let me just provide another scenario which disproves that the Nyquist-Shannon Theorem perfectly reproduces all frequencies below the Nyquist rate.

          In my above comment, I provided a scenario where you can get aliasing when a frequency is at an exact multiple of the sampling frequency, in the case of the academic paper, ten times, in my example, 4 times.

          What happens when a signal is sampled 3 times, e.g 48000 samples is an industry standard sample rate for DVD and Blu-ray. However, for a frequency 1/3rd 48000, so that would be 16,000 which is a frequency within the audible spectrum, a perfect reproduction cannot occur.

          In a sinusoidal oscillation, with three samples for the frequency, there are two scenarios, you either get a flat tone, or the amplitude of the wave that depends on the phase.

          Therefore the Nyquist-Shannon Theorem fails to perfectly reproduce all frequencies below half the Nyquist rate.

          What Nyquist actually said was that it can only do so perfectly in Theory.

          • I need to correct myself, I wrote the above comment in a haste from the excitement of the discovery. It’s too easy to write a comment, and you can’t edit or delete a past comment.

            Three samples or any multiple of the sample rate for a frequency causes issues with the Nyquist-Shannon sampling theorem. That’s because any possible multiple of that frequency is possible and could be produced. A higher sample rate would a more accurate reproduction between the samples, I need to produce graphs to illustrate this.

            In the above example, you do not get a flat tone.

      • Your comment regarding the nightfly is regarding 12 bit audio, I didn’t say the bit-depth of CD was in-adequate but rather the sampling rate.

        I agree that CD bit-depth of 16 which provides 120db when 24 bits is dithered down provides all the dynamic range of the audible spectrum.

        I am still on the fence about 16 vs 24 bit, I believe 24 bit is helpful to the DAC as it doesn’t have to over-sample as much to average out the samples. This is a problem that can be rectified however, and yes 16 bit should be perfect. The saving in data rate from 16 vs 24 bits should be used to provide a higher sample rate i.e 384khz.

        • Leon,

          You cited a marketing paper not an academic publication. As for The Nightfly it was recorded at 12 bits / 50kHz sampling rate. I think you are making assumptions based on one sine wave. You need to examine how well the Nyquist Theorem works with actual music. More sine waves and a 44.1kHz sampling rate works just fine. It is has enough cushion to allow roll off in the filters just outside the audible range.

  20. Great article, as an AV engineer I am quite critical in my listening; I am astonished by the poor recording/mastering on modern pop music in particular (and I’m talking about big artists) I heard an album recently that had SO much clipping and the drums sounded bloody awful!

    It would be pretty simple to include dynamic range in a HD standard to ensure the levels/compression is sensible.