MQA is Vaporware

[John Dyson]

Be careful -- that horrible bug might start being resistant to treatment, then we might all be stuck with it.  It is a good idea to 'be careful' about what one is doing nowadays.  The world used to be much simpler, but not anymore -- just be safe.

 

[John Dyson]

45 minutes ago, Ishmael Slapowitz said:

Funny, because I often hear folks state they don't want their music files or gear "infected" with MQA..😍

If one doesn't end up with really bad problems -- the system will still have 'cooties' (oh, I mean MQA recordings :-)).

 

John

[John Dyson]

7 hours ago, Shadders said:

Hi,

When you examine class D specifications, they restrict the measurement bandwidth to 22kHz for THD, as the noise above this is excessive, and would present THD figures woefully worse than class A/B.

 

I do not see any reviewer dismissing class D for its higher frequency (22kHz+) noise when compared to class A/B amplifiers. In fact, there is never a complaint from anyone regarding this - reviewers and the consumer.

 

What this tells you is, that the reality is that high resolution files are not required, no one can hear the difference of the extra content, and any difference is due to mastering etc.

 

MQA is irrelevant, since no one can hear the differences or extra content. The fact that class D amplifiers are so bad in the 20kHz+ region proves this - no one can hear it, they only believe that the can.

 

Regards,

Shadders.

Agreeing, and furthering you statement about higher frequencies.  Actually, excess HF is troublesome in certain kinds of analog HW -- and can certainly be problems in the digital realm (aliasing for example.)

 

If one  looks at op-amp specs (even really good ones), the distortion tends to increase fairly rapidly above a  nominal frequency in the range of 5kHz through maybe 20k or more at times (I have some very interesting and detailed references in that regard -- if interested.)

Also, the shape of the distortion curves depend on signal level and amplifier (even op-amp) loading.  Such curves are *very* interesting, esp on so-called 'premium' type op-amps.  Some are actually pretty good (some in the TI series.) 

 

Negative feedback (of course) doesn't solve all problems, and lack of negative feedback doesn't solve all problems also.  It is a matter of competent design, and it is simply not competent to supply excessive unusable HF material to any piece of HW.   It is generally best to remove analog signals above a reasonable maximum -- say think about starting to roll off at 22kHz, do something at least at 30kHz, and it is nice to be down quite a bit by 50kHz.

 

Actual, real pro equipment typically forces a rolloff somewhere above 20kHz (DolbyA, for example is pretty much extinguished at 40kHz)  simply for EMI and other such issues.  Oh so often, when I look at 'high res' material with sample rates above 48kHz, most of the so-called audio is digital interference and noise reduction splats.

 

I sure hope that people who see 'interesting' stuff above 20kHz are very often either seeing audio that they cannot hear, or even  more often seeing various forms of distortion or various kinds of coherent (e.g. digital or RF generated) noise.

 

Now, if there is 120dB (equivalent) audio at 22kHz, there might be some people who can 'hear'  the sound -- most likely the ACTUAL audible effects will be artifacts from distortion.

 

In fact, most of the 'improvement' by dealing with significant signal above 20kHz is an *increase* in distortion -- which sometimes creates a mirage of detail.

 

John

[John Dyson]

15 hours ago, sandyk said:

 

Hi John

 I wouldn't exactly call the rise in distortion from around 5kHz with the LME49720 (formely LM4562)  a problem.

 (still well under .0001%)

 Incidentally, Mark ,who was on the original design team reported that the HA version (metal can) sounded better that the DIP version despite measuring the same with their Audio Precision gear. I found the same too. (Perhaps improved heat dissipation was the reason?)

 

Kind Regards

Alex

  

LME49720.pdf 1.04 MB · 2 downloads

You are pointing to a manufacturer's spec sheet.  They always always give ideal distortion characteristics.  What was the source/load impedance while measuring the distortion?   If they specify the soruce/load impedances, they are usually not in the more challenging ranges.

Here is a very illuminating document about the actual performance of op-amps...  You might REALLY be disappointed...  This kind of information (and experience) that distinguishes the experienced analog engineer vs. someone who dabbles.  (It is okay to be a dabbler, it is just that data-sheets are not the be-all/end-all...)     Datasheets are 'good', but are usually not 'enough' to do a competent design (either experience, or knowledge of underlying device physics/behavior like with BJTs can really be important.)  Application notes won't solve all design problems beyond a data-sheet either :-).  (BTW -- I have a wonderful location that has many of the old National/RCA/Signetics/NEC/etc *complete* databooks online -- *legally*).

 

Opamp measurements:

https://www.dropbox.com/s/avnprbaoxw1ng8m/opamp_distortion.pdf?dl=0

 

Databooks/etc:

http://bitsavers.informatik.uni-stuttgart.de/components/

 

 

[John Dyson]

2 hours ago, Paul R said:

So guys - you can argue this back and forth until the cows come home. Logic should be:

 

1. Do hi-res recordings sound different?

    No. ->. Stop

 

2. (Yes) Do they sound different because they are different masters?

    Yes -> choose the one you like best. Stop. 

 

3. (No) Does the high res recording sound better? 

     No -> Stop

 

4. (Yes) Why does the high res recording sound different and better? 

    Go for it with the 14 mainstream theories of why, or invent your own. 

 

Personally, I usually think that 24/96 or above sounds a bit better than red book. The why I usually wind up at is the filters make a difference.  I do not care what the filters are particularly, I just want the sound that pleases me the most. After all, I am the one listening to it. 

What 'sounds' better can be very different from something that IS objectively better.  Simply because frequencies above about 20kHz cannot be heard directly doesn't mean that the IMD or other effects along with circuitry/software don't make a difference.

 

Sometimes, there is the subjective sense that 96k/24 sounds better than 48k/24, and I cannot (will not) argue either way about that.

 

John

[John Dyson]

26 minutes ago, Paul R said:

Well, does it really matter? Resource limitations have essentially become non-issues. Go with the higher sample rate to be on the safe side.  😎

I agree that it is okay to use higher sample rates, but wider bandwidth and higher sample rates aren't needed or actually beneficial for listening (linear applications.)  Also, higher bandwidth than needed/usable is a burden, not an asset.  (What I mean by 'burden' is that a lot of electronics, and even some software, can create more artifacts when presented with unneeded signal.)

It is my suspicion (and opinion  of others) that the 'difference' often claimed for wider bandwidth source material is actually additional or change in distortion more than anything else.

On the other hand, if I have to up/down convert over and over again -- I'd' rather keep the higher rate.  That doesn't mean that a nice rolloff well above the audible range is a bad thing (e.g. starting at a dB or so at 25kHz.)

When I rolloff for audible band, my stuff is essentially 0dB at 20kHz and a few dB at 21.5kHz.  It is nailed entirely at 23.5kHz.  When needing to support wide bandwidth (to make places like HDtracks happy -- they like to see lots of noise above 25kHz), my software doesn't always keep the audio and above-audio bands together, but separates them for processing -- it mitigates IMD to do the split before processing, then recombine.  (In fact, doing the split in the audible bands is also helpful.)

 

 

[John Dyson]

1 hour ago, sandyk said:

 John

 The 33 page Data sheet that I attached is possibly the most detailed opamp data sheet that I have seen. It does show numerous graphs of distortion into various loads etc., but you are correct in that it doesn't show the input resistance value used.

 However, in practise they are prone to instability/oscillation problems when directly driving cable loads of 100pF or more without a series output resistor.

 

Years ago, I also had several of the National Data books that you referred to, but they are obsolete now due to much newer devices. 

 

Regards

Alex

LME49720.pdf 1.04 MB · 0 downloads

Pllleassse -- look at the document that I referred to -- note the various input/output conditions/impedances, etc.  Spec-sheets like the one that you attached do not handle several very important  characteristics (source impedance vs.  input impedance effects are a relatively significant effect on distortion, yet is almost always poorly specified.)   The moderately low source impedance case is NOT always real world.

Even JFETs can have significant loading on the source because of capacitance effects and unbalanced  +,- inputs.  Lower voltage noise are usually worse (because of the lower dynamic input impedance, and it varies.)  Higher voltage noise devices tend to be smaller geometry, and therefore lower capacitance.  Also, a high impedance input often runs at lower currents which will cause a little less apparent varying DC input resistance.  By far, the AC loading effects are the most poorly specified.  (Devices running at lower bias currents tend to have more voltage noise and less current noise, also smaller geometry devices tend to have higher intrinsic resitances which increase the voltage noise -- making them more useful at lower currents.  Super low noise BJTs tend to be large geometry -- because the specsmanship is usually about input voltage noise, even though current noise is more important for higher source impedances.)

 

The problem with the spec sheet generally is that they don't always give real-world curves.  For example, what is the distortion at 10kHz with a 10k source resistance and a 600ohm load, at 0.1V, 1V and 10V output AC, and gains of 1, 10, 100.  There are lots of conditions, where the spec sheets give an idea as to the performance, but it isn't always complete enough.

 

I did a quick scan of the specs -- think about an op-amp in non-inverting mode (therefore dependent on the matching between input transistors/network), and running gain of X, source impedance of 40kohms, (common mode input impedance is very high -- so 40kohms is possilble), and at frequencies between 1k through 100kHz (assuming that some miscreant audio source supplies lots of high frequency signal -- wanna make sure that the IMD doesn't reach back down to audio.)   Read the specs CAREFULLY...

 

Here is a problem -- what is the input capacitance?  Why do you want to know the input capacitance?  Because it varies with the signal and internal state of the circuit and feedback, etc.   When the signal varies, and you have a non-zero source resistance, that varying input capacitance causes distortion.  In fact, this varying input capacitance is worse because usually a non-inverting amplifier circuit isn't AC balanced on the input -- so there are all kinds of weird distortions based on the unbalance that leak into the design.

 

So, you have a circuit, with a relatively high impedance source, and no useful spec that works with a general case of input impedance.

 

The input impedance of the op-amp circuit varies vs signal/frequency/etc -- where is that specified?  Do you know that is a significant source of distortion in op-amp circuits?  The distortion in an actual circuit will be significantly worse than the specs because of these facts (and others.)

 

Again -- read that very interesting document, and note the missing information from the 'common' spec-sheet -- even the many page attempt-to-look-complete.  Working engineers know about the missing specs, and know about the need to do tests or require selected or special order parts...  Real world production design IS sometimes a cr*p shoot on complex devices.  (If they keep on making them for longer than a few years.)

Doing your own high quality BJT/JFET design is the best -- if you can...  Of course, in production, that isn't quite as practical as using a chip that you pray that you can keep on purchasing.

 

John

 

 

[John Dyson]

On 6/6/2019 at 8:46 PM, esldude said:

Would be interesting to filter out everything below 20 khz.  Slow the speed to 1/4th, and amplify what is left to listen to it.  Is it only tape noise up there or is anything whatsoever related to music.  I think it will be noise only. 

Yes -- it will be noise (imd/hd splats from the lower 20kHz & sometimes 'tones').

[John Dyson]

2 hours ago, John_Atkinson said:

 

Analyses I performed 20 years ago showed that many recordings, even from analog sources, had music-related content  above 20kHz.

See https://www.stereophile.com/features/282/index.html

 

Note that one of the letters published in response to this article was written by Bob Stuart.

 

John Atkinson

Technical Editor, Stereophile

 

There are certainly some cases where there is 'music' up there, but most often, it is just distortion, tones, etc.   The distortion splats look just like music -- -I can create them at will if you want -- just break my DolbyA decoder to let it cause them again. BTW -- whatever music  up there is of such a VERY VERY low level that is meaningless when there is 'music.'   As soon as you add in the various forms of distortion, the 'music' is meaningless, and better just to remove it (it is a burden for the electronics, and cannot be heard -- PERIOD.)

 

Also, don't talk about 'vinyl' being able to reproduce above 20kHz -- it is so very contorted/distorted that the very small amount of meaningful music -- is damaged anyway.

 

Best to put on a competent engineering hat -- not so much a marketeer trying to benefit a 'cause'  $$$ of some kind. If the technically unneeded 'information'  (and actually burdensome) above 20kHz 'makes you feel good' -- I cannot discuss that. 

 

John

[John Dyson]

On 6/8/2019 at 3:53 PM, Paul R said:

 

Hi John - 

There is plenty of music content above 20khz, and nope, it ain't distortion. Vinyl can reproduce it just fine.  Has been that way for ages. Not saying anyone can actually hear it, at least not directly. However, if you stick your paw in an ultrasonic record cleaner bath you are sure able to feel something. 😉

 

Can you be a little more specific about what you are talking about? 

Methinks that you cannot hear the  'content',  and like noise/music above 20kHz from vinyl -- it is messed up anyway.   If you think that any signfiicant pop material made before 1990 has 'clean' highs -- well I have a bridge to sell you...  Most recording situations used DolbyA/DolbySR (and its littler brothers/sisters.)  The splats from the old HW are about as strong as any 'music' -- perhaps stronger when you have those nice 'cymbals' from 'high quality' recordings.  (reference:  when I decode professional material and compare with a true DolbyA -- on a specific recording, using DolbyA HW cymbals are decoded at a suppressed level -- guess where that suppressed energy goes?  think about 20kHz?  It isn't that the DolbyA cannot respond quickly -- it is that the response is SO FAST that intermod is so severe that much of the energy is lost outside of the audio band!!!)

 

Get a pair of Earthworks or good DPA microphones, and record directly onto digital -- then you have music material above 20kHz or so -- *that cannot be heard or detected by humans*.  Please do not count the resulting distortion products which do 'make a difference' when errant signals much above 20khz mess things up.  (I am not one of those 'must brickwall at 20kHz' people -- but every extra kHz above certain signal level will more and more likely be troublesome.)

 

One nice thing about my decoder -- it doesn't produce those splats...  (That is what I was talking about breaking my decoder.)  DolbyA encoded and decoded material is full of splts, similar to the vinyl being all contorted WRT signal quality above about -- say, 0Hz :-).

 

On the other hand, pros are 'big boys' and know how to deal with the HF problems, and they oftne like to keep high sample rates/raw bandwidths for purposes other than direct listening. (I am not counting myself as one of those 'big boys' -- just talking about the pros)

 

John

[John Dyson]

On 6/8/2019 at 6:38 PM, John_Atkinson said:

 

Yes, the >20kHz content is correlated with the sounds of recorded musical instruments.  When the music stops the overtones disappear.

 

John Atkinson

Technical Editor, Stereophile

 

The > 20kHz content is correlated because the distortion splats are correlated.  It can really fool even an experienced observer -- which are distortion splats, which are music or which are both....   Since normal humans cannot hear anything at reasonable SPL above 20kHz, then the only real engineering answer is to remove the (mostly) splats.  I never say that one MUST remove the useless information much above 20kHz, it is just that it doesn't help in any audible way, and increases the burden on the subsequent electronics.  Nothing wrong with keeping some material above 20kHz, sometimes there is benefit to avoiding sharp rolloffs (but has nothing to do with pre-ringing -- a bogus concept.)

 

John

[John Dyson]

15 hours ago, sandyk said:

 

 Sorry John, but highly experienced Recording and Mastering Engineers such as Barry Diament who these days only records in 24/192 .aiff with GENUINE musical content to >55kHz do NOT agree with you.

He finds that recording at  24/192 gets him to virtually identical to what his mic feed sounds like.

Recording at 16/44.1 doesn't even get close for Barry.

 

Regards

Alex

But -- when there is music at that frequency (when there is) -- you cannot hear it unless you start getting into the high SPL levels -- then other problems ensue.  It is a ZERO win for distributing material above 20kHz or so.  (I am NOT saying relgiously at 20kHz -- just that the cost/benefit drops rapidly.)

BTW -- it is okay for 'experts'  to be wrong all of the time...  That is okay -- but I am correct about the 20kHz thing.  Also, the 'splat' thing is certainly dependent on the situation...  You are speaking wtih Barry that he is using a pristine internal/professional recording source.  For distribution -- lots of nonsense gets mixed in.  I look at all kinds of distributions that get to the consumers -- The same material/different distributions  have different distortions/ control tones/etc.

 

Geesh -- I can create a test tone just as high in frequency as anyone else, but above about 18kHz (21-22kHz in some very special people), it cannot be heard.  Don't even bring in the idea of transients -- that has nothing to do with it -- unless the SPL is powdering you.

Note that I am a 40yr EE/Computer (EE Analog, all kinds of computer, degree in EE -- not technologist) person from Bell Labs, on the other hand -- a recording technologist knows his world -- that is okay, but what I am telling you is a physical true engineering fact (not only as a technologist -- user of equipment.)

 

You are comparing Barry's (or whoever my daddy is bigger than your daddy expernt) LAB experience with real world material that is sold to consumers -- different things.   The splats are almost IMPOSSIBLE to distinguish from pure material -- and in fact the splats might LOOK nicer, but they are wrong.   (Actually, I do believe that I can seperate some of the splat energy from the signal -- the DHNRDS decoder does that, but it all depends on the circumstance.)

Barry can generate a 100k test tone, then I can generate a 101k test tone -- got the idea? 

 

John

[John Dyson]

OMG -- sorry about replying to my own post -- but regarding 'splats' which happen *to some degree* on almost any automatic gain control device.   The splat (IMD) characteristics depend on both the gain control mechnaism and the signal.

 

I got some very encouraging results with the DHNRDS decoder WRT 'splats' (haven't checked in a few weeks) - doing ANY DolbyA processing on the old HW, encoding or decoding, will produce IMD splats across the audio spectrum -- even without extreme transients in the audio.  (The effect is paradoxically a softening/fuzzing of the sound.)  

 

My DHNRDS measurement show ZERO measurable splats in the ever so important higher audio frequencies, even where they could happen.  This very nice observation does NOT utilize the absolute IMD avoidance of band splitting... 

Band splitting: the DA decoder splits the audio band up so that the audible below 20kHz doesn't create splats into the super-audio range, and the super-audio range cannot create IMD that drops into the audio range.

 

Additionally on the DHNRDS DA -- recently made some trajectory shaping improvements -- even below 20kHz, on pretty intense material, I can not measure any IMD splats above 14kHz when simply bandlimiting the input audio below 14kHz on the input.  This shows that the IMD control is really good.*

* This means that even without the band splitting, the IMD is very substantially (seems like completely) suppressed.   I know that it isn't really 100% suppressed, but it is created and put in a different -- less audiblle -- place.

 

What is a 'SPLAT'? (my nomenclature, not standardized)  It is a burst of intermodulation energy that often comes from gain control changes being mixed with th audio.   A 'SPLAT' can actually be any kind of intermodulation energy -- sometimes looking just like a cymbal crash encroaching into and beyond 20kHz frequency range.

 

Normally, with the fast/relatively ragged DolbyA attack/release, you'll get splats as strong as any normal audio above 20kHz -- with a slow RMS style compressor, the splats still happen -- but are of lower magnitude.

So, generally, a fast fet-style limiter will tend to produce splats more easily (unless very carefully designed) than a smooth RMS compressor...  The DolbyA technology is almost  a worst case of a fast FET-style compressor that doesn't explicitly clip the audio.  (DolbyA could be worse than it is/was, but R Dolby was a brilliant engineer/researcher, and did an excellent job with the technology that he had.)

[John Dyson]

13 minutes ago, Rt66indierock said:

 

I've said more than a few times I can detect ultrasonics from a crash cymbal when I'm hitting it (bone conduction probably). But I can't detect them listening 10 feet away. But then Richard Vandersteen considers my hearing abilities "lucky." 

Deep in my past memories -- I do seem to remember that bone conduction does have wider HF bandwidth than 'hearing'.  I know nothing about the actual mechanism (one can always guess) -- but your possible implication about bone conduction did wake up that stale/old memory.  It might be proven wrong, but I'd guess that you are actually manifesting the bone conduction that the old papers spoke of (I mean decades-old papers.)

 

John

[John Dyson]

I see Blackmer as more as a really innovative semiconductor audio applications EE, who understood the details of transistor (and diode) behavior.  When I think of Blackmer, I think of his innovative  semiconductor audio applications work, and some understanding of the *very good* concept of the 'RMS' scheme that DBX style compressors used.  (The DBX style of RMS is NOT the same as what a non-audio EE normally thinks of as RMS.)

On the other hand, he might have some insight into other areas, but I am not convinced of usefulness of any high frequency material beyond 20-22kHz other than at very high powers to destroy/damage things.   There are also some nonlinear effects whereby above 20kHz sonic energy that can be used to produce audio from intermod...

So, yes, above 20kHz can be detected, but no -- they are not useful for music (unless 'music' include heating/physical damage/ or using intermod to produce pin-point audible sonic effects.)

 

John

[John Dyson]

On 7/4/2019 at 3:35 AM, Paul R said:

 

You don’t agree that ultrasonics are handy for noise shaping? I think they are a great place to push noise into and subsequently, easily filter out. It is why DSD can and often does sound so superior to redbook. 

 

Going on about people being being able to hear ultrasonics or not is just a great big whale sized blue herring. 

 

That is a red-herring -- removing the potential distortion  without dithering has little to do with the audiblility of material which is at 20+kHz.  .  Material isn't ulrasonic until it is ready to be applied to a transducer.  Lots of good stuff can happen digitally, including oversampling/decimation (with processing in between, like filtering.)

At 44.1kHz, we aren't talking places where a lot of audio procesisng is going on.  Most of the audio processing going on in a consumer's hands is the on/off control and slow volume changes.

 

John

[John Dyson]

7 hours ago, Ishmael Slapowitz said:

saying 16/48 is "superior" to 16/44.1 is pie in the sky.  The difference is imperceptible. But of course, you can hear a mosquito fart at 20 paces. 😍

I always say that I prefer 48k over 44.1k sample rate for various esoteric reasons -- but for direct listening, 44.1k and 48k are the same.  That is, unless one of the fortunate VERY few who can hear at 21k or above.   Even then, most of the actual material on at least older distributed recordings are NR distortion splats.

 

John

 

[John Dyson]

1 hour ago, John Dyson said:

I always say that I prefer 48k over 44.1k sample rate for various esoteric reasons -- but for direct listening, 44.1k and 48k are the same.  That is, unless one of the fortunate VERY few who can hear at 21k or above.   Even then, most of the actual material on at least older distributed recordings are NR distortion splats.

 

John

 

I mean to say -- most of the signall above 20kHz on older, consumer material is NR distortion splats.  It is so easy to prove with the equipment used in the day.  It is not a matter of opinion, it is just what happens.  In the 'lab' or 'recording studio', there is a much greater likelihood of actual (unhearable) audio being represented by frequencies above 20kHz --  but then there is the responsibility of maintaining the quality of the analog electronics (avoiding the distortion mechanisms.)  Please refer to the op-amp distortion document that I pointed to a few weeks ago -- there be lots more dragons above 20kHz -- even starts happening sometimes around 10kHz --  than most people realize.

Much of the time (most of the time) the differences heard when changing filters at the 20kHz boundry are either timing issues with filter behaviors or differences due to distortion products reflecting into the audible range.

 

John

[John Dyson]

4 hours ago, The Computer Audiophile said:

But you’re somehow one of only two people in the world who hear the difference between two identical files 😁

I *thought* i heard a difference on 'a certain test.'  I do lots of testing ALL OF THE TIME, and depend on my hearing...  However, just using hearing is not reliable when done casually.  At least, for reliable results, I really do need a better controlled environment, and at least use some statistics for comparisons.  For me, 'Sounds good' isn't even reliable -- esp at the end of the day.

 

Most difficult -- listening to/comparing  the entire recording.  It is so easy for the same recording to sound differently from play to play.

 

*in my case* the only way that my hearing is reliable at all -- listening for very specific instances of distortion, and whether or not that distortion is reduced or increased.  Even then, the relative measurement requires great care.

 

The general case of determining if a recording *is the same* -- super difficult, and for me requires statistics.  Almost every casual attempt at comparision results in embarassment.

 

John

[John Dyson]

13 hours ago, sandyk said:

 John

 I value your online friendship, and have enjoyed participating in your quest to rectify the problems in many earlier Dolby encoded releases.

 If you are willing, I am now quite able to (hopefully)  demonstrate to you that you weren't mistaken originally, just as I did recently with several high profile, and suitably qualified members with well over 5,000 posts EACH  in this forum.

Regards

Alex

My hearing (like most other people) is NOT reliable in that kind of test scenario.  I just did a casual evaluation, and like in a lot of cases, I heard a difference because I was listening differently.  A difference in mindset, time of day, phase of the moon, can let someone hear material differently.  Being reliable in doing a comparison of complex material as a whole requires a lot more concentration than I could dedicate or even have available for the effort.

 

The ONLY way that I am even marginally reliable is when I am listening for a specific sound/distortion, where not needing to listen to the entire set of complex iteractions.  In these cases, I have found that I made an improvement in one form of distortion, but audibly blind to the travesty that I had created elsewhere 'in the instantaneous sound-scape.'  When I need to do the small changes for distortion mitigation, it is best to take very little baby-steps in the changes.  It is amazing about how narrow the information-bandwidth is for hearing.

 

For comparing/evaluating complex material as a whole, I wouldn't trust my own results (other for casual conversation) without careful ABX type evaluations.   This observation of my own limitations comes from truly vast experience making mistakes and ending up red-faced.  The one test casually done where I heard a differece -- I did HEAR a difference, but it was due to mental/information processing state differences, not the audio itself.

 

sound -> conversion to audio -> conversion to objects -> conversion to experience.

 

I can hear only so many 'objects' at any one time, even though I can experience a recording as a whole.  Also, I can distinguish lots of simple tones, but not very many objects at the same time.  There is a 'bandwidth limit' in the aural 'object processing', at least in my own hearing mechanism.

 

John

 

[John Dyson]

2 hours ago, Paul R said:

 

Well, I actually meant the web, but I was on Usenet, Bitnet, and Arpanet. I get what you are saying. Heck, I remember having use bang paths for email.

I had a root account on ihnp4 :-).

(explanation -- ihnp4 was one of the more common servers on the bang path -- routing was mostly not automatic -- but some servers had a little more inteligence.) ihnp4 was very common bang path constituent for UUCP networks also.  Nowadays, everyone can route to anyone.

 

John

[John Dyson]

2 hours ago, crenca said:

 

That is the position of most consumers (of digital anything) - they have to rely on an authority (i.e. the specialist) for knowledge about the reliability/truthfulness of fill_in_the_blank product/software/service.

 

In the audiophile niche, besides a handful of (mostly consumer driven) forums such as this one, where are the reliable authorities and specialists?  Certainly not writing for the traditional "Audio press" such as Stereophile and TAS.

 

Chrislayeruk, Barrowboy (who posts here as Tintinabulum), and the like are astroturfers fur sur.  Roon has a real astroturfer problem, but then they have a larger MQA problem in that a combination of circumstances (past working associations, being in the UK, etc. ) led them down the road of compromising with the MQA fraud in the first place.  They of course are not the only ones...

I truly don't believe that there are fully reliable authorities in almost any field.  All too often, there are financial interests, personal interests or opinon/feeling that overrides rationality (lock-in to erroneous technical opinion.)  Sometimes, even if a person knows the actual facts, they end up demuring because of overly strong dissenting opinions -- often because of the error sources that I mentioned above.  Even when I truly, 100% know a technical answer, I will sometimes demure (not in the right mood to deal with controversy.)

 

Even in areas where I am truly an expert, I make errors, from time to time have erroneous opinions (because not having a current interest/forgetting details,etc.)  Lock-in is a problem that all technical people tend to have.

 

Even technical people can be led astray by other technical people.

 

I think that the biggest problems are financial interest or personal bias coming from misguided technical reasoning.  Here is an example:  I worked with a guy who was a pioneer in a certain field, he was far senior to me -- but he kept advocating using a really error prone source for an electrical delay (synthesizing the RAS/CAS delay for first generation dynamic ram) -- he actually advocated using a series resistor, depending on input capacitance and threshold for CMOS gate -- this RAS/CAS delay had a rather precise timing requirement.  I tried to come up with a reliable digital timing method (TTL wasn't fast enough), finally I 'gave up' and advocated using an analog  delay line (we couldn't clock our circuits fast enough for proper digital resolution) -- but he advocated the R/C & threshold delay scheme, using a pot for a production tweak -- implementing the RAS/CAS delay that needed to work over the entire industrial temperature range.   That expert (truly, he was a technical expert) lost his contract job because of that conceptual lock-in error.  Using a tweak for the RAS/CAS timing over industrial temperature range (CMOS thresholds/characteristics/etc) - it would have been a production/maintenance/support disaster.  (Early 1970s')

 

I know that my little anecdote wasn't all that 'short', but I am trying to explain that it is very tricky to find someone who really does give an accurate technical opinion all of the time.  Best that one can do -- listen to more than just a few experts with differing agendas, and then use common sense.  No-one is immune to both sides of the problem.

 

I do believe that 'technical experts' should try to strive for more integrity* (myself included), and also the user base shouldn't 'buy-in' so very quickly to snake-oil...  How does one detect 'snake-oil?'  I have no good answer for that.  I wish I did.

* When I speak of integrity, I don't just mean 'honesty', but I mean the entire package that includes knowing-ones-own-limits.

 

John

 

[John Dyson]

2 hours ago, Paul R said:

 

There real problems are with those that think they know something, but really have no understanding of what they are talking about. Nor the will to study hard enough to learn. 

Those are the people that poison our hobby, especially those who "get on a mission to save the poor audiophools..."

Even trying to correct a easily technically provable error is sometimes not easy or helpful unless it is possible to relate to common sense and day-to-day usage. *I believe that pushing the facts about every little misunderstanding isn't a good thing...  When there IS an ongoing discussion, an actual bit of technical accuracy CAN be helpful -- I hope 🙂.  I wish more people with actual knowledge were more easily accessible.   Better access to experts in specific fields could have been incredibly helpful for some of my recent misconceptions, but information is sometimes difficult to find.

 

Some issues, like MQA, bother/worry me personally because I want to maintain my own access to good quality, unmolested, non-DRM music.  I also worry about the rest of the world -- working effectively full time for years to learn enough about a specific field that will help 'OLD' (1960s through 1990s) music quality/availability in the future.  There IS altruism in the world, and I know of maybe 2 or 3 people working very hard, and others contributing from time-to-time in the somewhat altruistic effort. (Common participants to this forum and some fairly well known professional names have helped the effort.)  My own project is NOT limited to (for example) DolbyA at all, and requires resurrecting a lot of difficult to find long-lost common knowledge (and also some hidden know-how.)  I have been a big-time victim of misconception -- much of it due to my own resolvable limitations.  It has taken a LONG time to get some kinds of accurate information.

 

Can you 'save' people from totally misguided ideas?  My answer is -- maybe, if the idea is important enough, but otherwise it is best just to let-go of any kind of 'crusade' (no religious intent for the usage) to correct everything.  This is a hobby for most people -- a little bit of misunderstanding is perfectly fine :-). 

 

Frankly, I wish MORE people who can contribute actual technical facts could find a way (and the time) to discuss things without being perceived as being know-it-alls or stir controversy.  Gaining access to real information, beyond what is currently available, has been troublesome in my current project.  Some actual experts DO demure.   Maybe the most difficult problem for those of us who don' t know everything -- trying to find people who understand their own limitations, and who TRY to avoid passing on their own possible misconceptions.

 

Audio/recording/etc can be very technical -- and there sometimes might be a tension between the artistic temperament and the kind of knowledge needed to truly understand what is going on.  I guess - most important -- remember the goal.  Participating in the hobby can required a very different mindset than the technical knowledge needed to implement the tools of the hobby.  It is easy to wrongly assume that 'understanding the use' is the same as 'understanding the supporting technology'.  They are NOT the same things.

 

John

 

[John Dyson]

2 hours ago, botrytis said:

 

I do spend a lot of time, at work, reading patents so I can determine was is real and what is not, for future research. Patents do not have to be actually usable. All it has to do a point out a unique tech or process. This way they CYA.

 

Your patent comment really hits home with me.  The Sony DolbyA patent is a near perfect example of a CYA patent.  I believe that it is possible that it is maintained simply to benefit from someone else's completion of the project.  They patented a key part of an obvious DolbyA implementation (making the feedback straightforward parametric instead of the impossible-in-DSP-land audio feedback scheme as in DolbyA HW.)  Even though their patent is worse than inadequate to implement a DolbyA (the most important characteristics of DolbyA are ignored), they thought that they could cause almost any good implementation to infringe, therefore benefit from someone else's work.  Some of the 'facts' as stated in the patent are simply 100% wrong, yet enough is accurate that it is likely to convince a judge about the relevence of the patent.  Patents like that suck -- I dont even think that it is possible to make a DolbyA decoder sound good even using the architecture in the patent -- but it sure *LOOKS* good. (US 5,907,623).  The current expiration is 'fee related', but I'd suspect that they could resurrect it pretty quickly.  Also, there are patents like US6807278, which are similar, but my project doesn't infringe that at all either!!!

If starting with a technique similar to the way that the patents describe, a developer would be very susceptable to lock-in.  Luckily, for my own sanity, I purposefully forgot the Sony patents, and implemented using a VERY foreign technque that supports greater flexibility.  Patents aren't evil, but they can cause problems (and benefit holders of essentially garbage patents.)

John

[John Dyson]

8 minutes ago, firedog said:

Red herring argument. Either the measurements are correct or not. People can differ in the interpretation of those measurements or the understanding of their importance. 

Where is there “blind belief”? If you can show their measurements are incorrect, show it. 

Lots of people on this site  have said they don’t think various measurements are definitive. That’s their opinion and their right. 
Daverich’s vague accusations are just that. People aren’t called “shills and trolls” because they present “facts and measurements” that differ. They are called shills ad trolls when their positions are contradicted by the measurements and they have no reasonable alternative explanation, nor alternative “facts and measurements”. Alternative opinions without showing a basis for them aren’t the same as alternative “facts and measurements”.

I don't think that most people argue that when (AS AN EXAMPLE):  there is a recording that has signal at up to maybe 40kHz -- that the signal isn't there.  It is more that there can be interpretations if the signal is useful...  Sometimes a signal can be coherent from some other signal device, sometimes a signal can be an IMD splat from an NR system, sometimes the signal can actually be audio from a microphone.  Secondarily, there are issues about the actual quality or useful signal level of the 40kHz signal, and most importantly -- can anyone detect that signal as directly being a part of 'music'.

 

The discussions are more fine than a simple disagreement about specific facts.  One problem with these discussions happens because it can be a little tricky to separate out the aspects of the discussion and not confuse them.

 

Depending on agenda, the 'facts' can be somewhat convincingly interpreted in different ways.

 

Organizing the facts in a discussion can be tricky, and there seem to be a lot of traps and pitfalls when there are so many different weightings of the importance of the facts, let alone judging whether or not that the facts are accurate.

 

This problem associated with organizing the facts, understanding the ramifications, weighting the importance  ALONG with the emotional aspects conceptually reminds me of 'herding cats'. 

 

John