Missing information

September 27, 2022
 by Paul McGowan

I am always fascinated to watch those slow-speed nature videos where you watch flowers open and plants grow. Or, the opposite. A high-speed camera that slows down the wing motions of a hummingbird so we might see what in real life we cannot.

Those cameras (both film and video) are essentially digital: frame by frame they capture a single moment in time. When played back they look continuous.

Each form of slow and fast digital capture offers a glimpse into that which our analog senses cannot perceive. In other words, we get more information than what we are able to consume in our analog world.

The same is true today in digital audio. Whether high-sample rate PCM or DSD, modern A/D D/A converters can capture and playback far more details and frequency extremes than we can perceive.

All of which to point out that somewhere along the technology timeline we moved from being able to seamlessly capture nearly all the information in a live performance (the end of the analog era) to today capturing more than our human auditory systems can use.

Now that it is no longer a question about missing data, the challenge becomes converting that data back into the analog world.

Subscribe to Paul's Posts

47 comments on “Missing information”

  1. The most important missing information for me is what you discover about a product after you’ve bought it, particularly if that information has a negative impact on the use and suitability of the purchase. Always DYOR (do your own research).

  2. I was listening to a podcast the other day, basically explaining that the human brain is a massive prediction device, rather than simply processing millions of bits of data. The data is as much stimuli for a preconditioned response. There seems to be a reality gap between the idea that we hear digital music digitally, that we hear it in 16/44 samples or whatever. The neuroscientists understand it that we may receive and process discrete data, but our brain effectively predicts what we hear, effectively what I think Paul refers to as analogue.

    So I don’t think Paul has to worry about converting back to analogue, we can do that ourselves.

    1. The data stream reaching the retina in our eye is already dramatically reduced when further transmitted via the visual nerve to our brain. However we believe that our eye are seeing the whole reality and that there is no loss of visual information! 🙂 On the other hand: the promise of HiFi is that the loudspeakers will produce the identical sound wave which would have reached our ear during the live event! Thus which degree of data loss (and additional distortions) would be acceptable? And do we know the parameters being most crucial for the prediction capabilities of our brain?

      1. I remember going on a wildlife drive in Wilpattu National Park in Sri Lanka and our guide pointed to a rather average bird of prey of some sort sitting in a tree. It’s head was rotated 180 degrees (no need for eyes in the back of its head) and he said it could spot its lunch from 3 miles away. Ever since I’ve always wondered at how compromised our senses have become since we stopped being hunter-gatherers.

    2. The brain has to be a prediction device as a survival tool. If you were in the savannah and you saw some rustling tall grasses you had to predict if it was an impala or a lion. This is one of the reasons that the brain tends to value negative information more than positive. If you assume it is an impala and you were wrong, you became food. But if you assume it was a lion and it was an impala, nothing serious happened.

      We still behave this way. Predictably irrational.

  3. I’m not sure if…. with the analogy of a visually (due to the speed of procedures) not followable course of details, transferred to the audible world (three completely different topics: human analog sensory perception/technical analog ascertain ability, visual/audible, capture speed/capture range) …you really want to imply current digital format’s and A/D conversion (in contrast to analog) have no limitations anymore in their ability to capture more than the completeness of what happens acoustically and improvement just lies in the D/A conversion and several folks just still prefer analog because the D/A conversion is not yet at the necessary level of performance?

    I agree this could be the case.

    But why is always the current technology exactly the point where this happens? I guess you claimed this or something similar already with every step digital made so far from 16/44 (and later revised due to better knowledge) and now again, although DSD1024 or something completely different might be the format which finally stores even more, as we might find out in 5 years from now.

    I admire your ability to cement your facts in the particular now. I’m just not sure if everyone in practice is always well and realistically advised with this.

  4. As I mentioned the other day I’m now streaming DSD 256 with optical Ethernet conversion to the I2S input. It is very, very good and by far the best digital I have heard in my system. BUT, it still does not sound good with massed strings. There’s still a glare compared to my analog rig. Is then, the problem in the Directstream DAC and does the Mark 2 fix this?

    1. My best files for comparison are pure DSD 256 downloads from Nativedsd.com. I have some Brahms and Schumann that I had high hopes for and so far I’m disappointed. Classical is the acid test.

      1. Classical orchestral music is certainly the acid test with digital. I’ve been fighting it for decades. I did find that the most improvement I got was from speaker placement, room treatment and room spatial tweaks. Sloppy attention to these factors smear sound awfully.

    2. That’s likely more to do with microphone technique and recording I would assume.

      However, you say it sounds fine in analog recordings?

      I’d be curious to compare two identically miked recordings, one with analog tape the other with DSD256. You might have something interesting there.

      Massed strings might easily be prone to IM distortion in which case it could sound screechy.

      1. I also think this case is rather a matter of different recordings.

        Paul, in all respect, you talk a lot about what sounds better and you certainly have quite some experience with each individually (how comparable the level I don’t know)…but you should probably also make those comparisons if you didn’t do yet.

        To my knowledge and the only ones I’m aware of, who in parallel record DSD (and partly PCM) and analog from the same microphone settings is Fone and Yarlung (they don’t use the DSD, but the analog recordings for their vinyl), not sure about Opus 3. Giulio from Fone for a certain time even made comparisons of LP‘s sourced from his digital recordings vs. the analog recordings and found out that the analog ones on LP sound better. For a certain time he even offered both LP versions commercially.

        All other comparisons would have to be done with analog recordings (instead of an analog live feed) played back all analog or transferred to DSD or PCM.

        For whatever reasons the latter comparisons are more negatively for the digital part than the first in my experience, but both usually are especially in case the recordings have a high potential of ambiance and 3D imaging

        IMO everyone who didn’t make those comparisons, must have decided about his approach by principle or by a quite selective view or test, but not beginning to end product.

        I wonder if Cookie made such tests…but I think no…she didn’t make records at her DSD era for Blue Coast as far as I know…but she certainly will have a lot of experience with analog and PCM based vinyl from her past I guess.

        Why switching live feed comparisons between digital and tape in a studio might give different results than comparing two end products made from digital or tape at the beginning, I don’t know, but it could be.

        Finally there are so many mastering gurus we read of who should have this experience…and to be honest, so far I didn’t read of any of those who know both sides well, who don’t or didn’t (in case of Sax, Ludwig, Ricker etc.) prefer analog.

        But at least DSD256 is quite new and could lead to a more balanced perception.

        1. Reference Recordings also released one album recorded and produced all analog for LP and produced in hires digital for a later LP release on heavier vinyl. It was Nojima plays Liszt.

          The analog version here sounds so much better, that I’d not dedicate this to the analog thing only, but mainly to different mastering processes.

          1. Jazznut

            You hit on a point that jumped out at me. Is more care being taken in the analog recording / mixing / mastering process, than what is taken in the digital process? Is it some form of recording bias? 😉

            It seems to me Paul is in the perfect position to demonstrate. Record in both PCM and DSD at the same time with the same mic’s positioning etc. don’t bother mixing don’t bother mastering.
            Don’t produce a whole album, just a simple 2 minute take. The larger and more complex the ensemble the better.

            Take the 2 versions and let us compare. Otherwise there’s too many variables.

            1. Yes, we could do, but PCM vs. DSD for me personally isn’t an open topic anymore. Too many folks without agenda, who listen, who don’t sell it, confirm the superiority of DSD. So even though I don’t have the environment and sources to properly compare myself, I believe DSD is clearly superior.

              The best chance normal consumers have to experience the DSD potential, is to listen to one of the few exceptional Mofi LP’s and their quality (especially of the DSD256 ones). Then every harming digital side-process and one’s own DAC quality is out of the equation.

      2. Yes that would be a good test but I don’t know of any such dual recordings. Failing that, I’m suspicious of the DA conversion in the DAC. Is PS Audio focusing on the output stage in the MK 2? Say, use of exotic capacitors, or no capacitors, eg chokes, and tubes? I’d love to see a marriage of a DAC with a Ypsilon VPS 100. Does that sound crazy?

    3. Happy I’m not the only one who hears similar anomalies with more than a few DSD releases. When I listen to less 4-5 instruments e.g. a jazz quartet, I hear cold, sterile sounding instruments (perhaps because of the extreme low noise floor). When more than a small group of instruments (like mass strings) come into play, the soundstage seems really congested, something I’ve never experienced before. This music does not feel natural to me, not in every release but in many of them. It may be the recording studio reverb units, I’m just not sure but it leaves me feeling cold. There are reasons I’m not jumping into this new and changing technology just as I did with the original CD players for 15 years. It’s why I always refer back to the live venue and what I hear in the presence of live music. There are many great recordings done in the studio that satisfy my soul but that’s not happening for me with what I perceive with the direction that things are headed towards. Many of you may like this feeling but I certainly do not.

      As much as I love the Bailey release, I compared it to his Telarc release and it may not sound as harmonically correct as the Octave release but I do feel the entire soundspace as if Zuill is performing alone in a large stone church with his Cello emanating from my speakers. It actually sounds like he is all alone in this acoustic environment with his beautiful instrument radiating like distant sunbeams radiating in every direction. This is the best way that I can explain what the differences are to my ears.

  5. There is some evidence that suggests that our ear / brain combination that we hear with does work on a sampling method. Thus, it maybe that we hear in a digital manner. However, if you think this understanding of the ear / brain combination has progressed to the point where we only need some limited amount of data to “hear” what we used to get from pure analog I suggest that you give a listen to some MP3 audio. 😮

  6. “Whether high-sample rate PCM or DSD, modern A/D D/A converters can capture and playback far more details and frequency extremes than we can perceive.”

    The above statement raises a question. We’ve been told that high rate pcm (352.8) capture doesn’t sound as good as DSD 256 capture. That leads to the conclusion that even though more data than we can perceive can be captured, it’s still missing something. So then if something IS still missing, and we take things at stated face value, they’re not on the capture side, All that’s left is the the mixing / mastering process or playback side.

    1. That’s a very astute observation Mike and one I have to think about. Clearly, we can show there is no missing information in the sense of steady state tones. And we know DSD256 and PCM 352 can capture frequency extremes greater than we can hear. So, maybe it’s something in the middle?

      1. Paul,
        I really like questions like you just posed in your reply. Those are the types of questions that move things forward and help provide deeper understanding when they are answered. (Fully or not) 😀

      2. Paul, You’re right that we do not hear fundamental tones much above 20 kHz, however, those frequencies about 20 kHz are essential to forming the sharp leading edges of drum strikes, cymbal strikes, guitar plucks, etc. We all hear those ultrasonic frequencies every day. We hear them not as fundamental tones, but as complex audio signals that we hear and record everyday.

        Snap your fingers. The acoustic signal is almost a pure impulse signal. It rises sharply and then in a micro second falls sharply. Bandwidth limit that signal to 20 kHz and see what you get. There will be no sharp rise and fall and there will be pre and post ringing. Pure sine waves are the building blocks of the complex acoustic signals that are part of everyday life. If a digital representation o0f those complex signals does not include any and frequency content above 20 kHz then there will be lots of missing information.

        1. Or you can read this nice “laboratory” paper. In this very well controlled experiment, people in their 20s had a very hard time hearing an unmasked high frequency sound. Indeed, even 20 kHz had to be at a rather high SPL to be heard unmasked. Music has extremely limited signals above 12 kHz and those high frequency signals would be masked by the lower frequencies. So I find your conclusion highly unlikely. Enjoy!

          https://www.degruyter.com/document/doi/10.1515/bmt-2018-0048/html?lang=en#:~:text=Abstract,lowest%20test%20frequency

          1. CtA, I am sorry, I must not be communicating very well. I have no problem with the article in the link you supplied, but that is not what I am talking about, and again, I’m sorry I did not make it clear what I meant. Here is a link to what I mean. The first figure in the article is an actual demonstration of what I mean.

            http://archimago.blogspot.com/2015/04/analysis-dsd-decoders-2015-windows-mac.html

            You can easily see the effect of bandwidth limiting ( to prevent aliasing ) on the impulse signal ( a finger snap ). The extreme example is what happens to the original signal when it is sampled using 24 / 48 kHz. Removing all of the frequency content above 22 kHz gives you a signal that is noticeable different than the original. My personal experience using my TASCAM hi-rez digital recorder allows me to hear the difference between the digitizing methods used in this diagram. Frequencies above 20 kHz are very important when we digitize sound signals.

            1. “No need to worry about the sonic output from any of these converters IMO. Conversion algorithms and software look mature with little difference between them.”
              “As I suggested last year, I believe that DSD –> PCM conversion is transparent.”

              I am not sure I understand your concern if I read this in the article. Anything below 20 kHz has not changed.

    2. Maybe some relevant acoustic elements are simply masked by more or less annoying digital artefacts/distortions added? 😉 As far as I understood digital reconstruction schemes they are based on most simplifying assumptions and on pure sine-waves as basic elements – quite contradictory to the characteristic transients in music. Interpolation between two sample points also uses more or less simple functions (linear to spline functions) also more or less bad approximations!

      1. Paulsquirrel,

        I’m not even going to speculate as to any of the whys. Your points are well taken and certainly well within the realm of possibilities, maybe even probabilities.

        By definition when you sample you have to reconstruct based on assumptions or predictions.

        1. A lot of good theories today Mike. Perhaps the industry can move forward. So far, I’m still singing the same song… “But Not For Me” At least for the present.

          How would you define extrapolation? I would “assume“ the same response.

          1. Stimpy2

            From where I sit it all has to do with the recording. The care and the manipulation of said recording.

            Format is secondary, although in theory it should be foremost.

            1. I know Mike, we’ve done this dance before. The best thing about today’s post is it’s starting to open up the door to other new thought processes that may answer some of the questions as to why I believe that DSD is not even close to approaching a satisfactory musical experience for everyone.

              If you read another of my comments today, I asked a question about the differences between Bailey’s Octave recording which is truly beautiful yet more romantic than Bach meant it to be (listen to Sigiswald Kuiken’s reading) but more importantly comparing it to Bailey’s Telarc recording that he made prior to Octave. IMO, the Octave release has its pluses and minuses. It’s certainly a beautiful sounding instrument with incredible harmonic integrity and a masterful musician but recorded a bit to close. When comparing it the Telarc version, I get a full sense of Bailey and his instrument in the center of a church radiating in all directions which is so much more realistic to a live event (I have been to two of the Partitas and Sonatas for Solo Violin in two different churches and it is so much closer to ‘being there’ then the Octave recording. On the other hand, I’ve been to some rotten live events where I left before the end and my listening at home makes me happier than listening live. It’s a double edge sword.

              And we both know a good recording produced in PCM or on Vinyl can be life-changing and touch your soul in a way that a poor recording can never do.

              So in the end, I fully agree with your comments today as usual but I wanted to throw my two cents in a second time. It’s possible that I may be the most senior person in our community at the age of 76 (going on 30 in my mind and 50 in my body) so I’ve been around the block more than several times regarding live and recorded music. Doesn’t make me the end-all and be-all, not even close, but it does give me history and I do have the ability to listen to sound the way I feel it was meant to be. My perception only, but I don’t think I’m the only person with these conclusions.

              1. And another thing Mike… Look up the definition of synthesize… Well all right I’ll give it to you

                Synthesize=Combining (a number of different things) into a coherent whole.

                This Definition immediately strikes a familiar note. When a continuous audio signal is chopped up, massaged and then recreated back to a supposed higher quality analog signal, we have possibly introduced many additional anomalies to the original. Perhaps a real Pandora’s Box.

  7. Dear Paul,

    I think you raise a very interesting theory, thank you!

    My question is: is more captured information necessarily consonant with greater emotional engagement to the reproduced musical event?

    1. You hit it on the head with what I spent a half hour commenting on because I couldn’t understand why I was feeling uneasy about the sound of this technology as I stated above.

    2. You hit it on the head with what I spent a half hour commenting on because I couldn’t understand why I was feeling uneasy about the sound of this technology as I stated above.

      The statement “Music hath charms to soothe the savage breast” is not happening for me when I listen to this technology at present.

    3. You hit it on the head with what I spent a half hour commenting on because I couldn’t understand why I was feeling uneasy about the sound of this technology as I stated above.

      The statement “Music has charms to soothe the savage breast” this is not happening for me when I listen to this technology at present.

  8. In Paul’s analogy the reason we are able to absorb more of the existing information is because it is slowed down, it’s down to timing.
    Does this infer that future improvements in digital audio will be with regard to the clock?
    Just a thought.

    1. Richtea: Good point. It’s remarkable that our ear/brain senses timing errors at such low levels. Maybe jitter has to be completely unmeasurable before we will not hear the negative impacts on our digital playback systems.

      1. I think you and Richtea may be onto something. This is treading in uncharted territory to the best of my knowledge but I don’t keep up with these topics in depth very often anymore. If there was an oscilloscope that has the bandwidth to see what the leading edge on the clock waveform is, unless it is a perfect pulse (no overshoot or ringing) it could very well be at least part of the reason.

Leave a Reply

© 2022 PS Audio, Inc.

linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram