24/192 Downloads ... and why they make no sense?

[crisnee]

I think there may be a slight misunderstanding as to the attack part of the musical instrument sound. Attack doesn't mean hard attack, it includes even the softest beginning of the note (when played on an instrument).

 

I think the kind of test Miska was referring to involved artificially removing the attack part of the notes and then trying to discern their origin. One could not create this sort of sound test by playing an acoustic or electrified acoustic instrument. So he was not in anyway besmirching music, just possibly unnatural sounds.

 

So I say Prairie Oysters to you and your ilk.

 

-Chris

 

 

[Teresa]

In the beginning of SACD 13 years ago, some people were claiming CD sounded better and that SACD sounded too soft. The consensus was that these people got used to the unnaturally over-etched hard sound of 16/44.1kHz PCM and that when exposed to realistic sound they were underwhelmed. My advise was to listen to live acoustic music and to completely quit listening to 16 bit music for at least a couple of months. Once one is deprogrammed from 16 bit audio they will have a hard time being able to listen to it again IMHO.

 

Quote from Harry Pearson in my interview with him http://analog-lovers.blogspot.com/2011/02/harry-pearson-interview-from-analog.html :

 

Harry Pearson"I believe the sources of PCM distortions are beginning to be well understood, if not scotched. But what many engineers are doing is "sweetening" the PCM sound by adding colorations that are designed to make the experience less exhausting, but, which, at the same time, remove the sound further from any known reality."

 

[Julf]

"the test itself is flawed by having "easy-peezy" a capella male vocal material, rather than something which challenges 16 bit"

 

That is of course a very valid comment, but I have also received comments stating that "natural" human voices are by far the best test material. My main concern was that a capella material would not have enough high frequency content to show a difference between 44 and 96 kHz, but I don't see why it doesn't challenge 16 bits.

 

"As I've suggested throughout the thread, people can easily satisfy themselves about the differences they may hear between various sample rates and bit depths by simply downloading the many free or low-cost samples and tracks available at many sites"

 

One problem is that often the hi-res material has been mastered differently from the red book material, and we know that even small changes in overall level can make a huge difference in how we perceive a track - not to mention changes in EQ etc.

 

The other problem is that if you yourself download the tracks fro listening, you know which one is the hi-res one, and unfortunately our brains do tend to emphasize our beliefs - if you believe hi-res is better, you will hear the hi-res sounding better, and if you don't believe, you won't hear a difference. This has been stated by several people on this board - "Maybe you don't hear a difference because you believe there is no difference?".

 

[Julf]

"Micheo has provided the results of lots of tests, and he is a professional in

the audio business"

 

Ah, but is he famous in Audiophile circles? Does he write for TAS?

 

[Julf]

"Maybe you should have used the fourth movement of Beethoven's Ninth."

 

Would have loved to, but there is that pesky copyright thing...

 

[Julf]

"From BIS, a recording company that sells SACDs from 24/44.1 source recording?"

 

I have no other connection with BIS apart from Robert kindly helping me by arranging access to the material, but your claim sounds like a rather grave one - do you have any proof?

 

"If you aren't against hi-rez, I don't know who could be."

 

Then you don't know much.

 

I came to CA because of my interest in Hi-res, and having gotten burned by HDtracks, I found the "Audiophile Downloads" and "Music Analysis - Objective & Subjective" forums extremely useful. I have used a fair bit of money to support HDtracks, eclassical, B&W Society of Sound, 2L, iTrax, Linn and others.

 

I can definitely hear a difference between some of the hi-res tracks and red book. Or at least I think I can. But I can't tell if that is from different mastering. Or it could even be all in my head.

 

So I came to Computer Audiophile as a believer in Hi-res. I am quickly loosing my faith. One reason is seeing how few of the people who claim they can hear the difference are actually willing or interested to do a blind test. But more importantly, what is turning me away from hi-res is actually Computer Audiophile.

 

The way many of the supporters of hi-res seem to defend their faith rather fanatically and irrationally makes all my "faith-based belief system" warning radars go off. Enough to make me at least question the whole hi-res concept. I guess that if you are part of a fanatic religious sect, questioning is the same as being against?

 

 

[Miska]

Upsampling the "Digital Crap"

You can read about over here:

 

That story has word "crap" couple of times too many over it. But the guy doesn't seem to understand what he's seeing... Luckily I have handling just for that case.

 

 

 

 

What I failed to see is how he thinks the "Petrov" one would be better, looked bad to me.

 

 

[PeterSt]

But more importantly, what is turning me away from hi-res is actually Computer Audiophile.

 

[after finishing this post, I added the above later as a bom shell ...]

 

No need to trash CD. Modern, quality DACs make CD sound almost perfect. -So pure and resolved. Stepping up to 24/96 just gives a tad more fluidity, slightly better micro dynamics and a little bit more low level information. This is my experience from owning and trying out some of the best DACs in my own system. Currently I am using the Phasure NOS1 DAC, which is by far the best I have heard. Together with the accompanying software player XXHE, they lift 16/44 playback up to another level. In fact, the special “Arc Prediction” feature gives you the impulse response of a high-rez recording. (The creator of this marvel is no other than PeterSt, so he knows a thing or ten about the subject).

 

PS: I had a great time with my old 3.6s some years ago. My present loudspeaker system is different, but I still use a full size ribbon tweeter similar to the Magnepan.

 

So, this is from one of my customers. It is especially to be noted that he is the only one who tells that Hires is better than CD.

For a kind of reference, also look here please : An interesting article about HiRes recordings, which is a thread in my own forum about the same article. Thousands of customers, and NONE is even interested in the subject. Pedal is, but this is because he advocates Hires. And, as far as I can tell from my own forum ... the only one. As in here, several threads ran about it, and I never found anyone to be pro-hires, except for Pedal again.

 

This should tell us something. I mean, without any self-advertising in order (look what I've all written in the CA thread here), it MUST be some sort of conclusion that my indirect "debate" about transients does something which is supposed to happen in Hires. Without exception all these XXHighEnd users apply the Arc Prediction filtering, when their DAC allows it. So, it is not even about a Phasure NOS1 DAC ...

 

The interesting part of it all, or merely intrigueing for myself, is that I can well expect Pedal to be more experienced on this all than myself. You don't know the man, but I do, so trust me on this please.

Whether he really is correct in Hires being better is something else, but it is easy to give him the advantage of the doubt, because it will be about WHICH Hires is judged. And as you may see in the link I referred to, he sticks to maybe 30 well done Hires albums for his judgement.

 

When someone listens to my system, and which almost each time is about auditioning the NOS1, exactly nobody even comes to the idea of listening to some Hires or compare. Not even Pedal did (but he already was a customer at the time). It is no subject. The whole subject is not needed.

 

If I were not right on a couple of things, how come that the whole subject of Hires actually does not exist on my forum ? How come that precisely nobody but me has something to say about it, when that only-one-Pedal stirrs the pot ? And really, my forum is not about the NOS1 you know.

 

To put it all down to the merits : Yes, Hires must be better. But whatever I randomly pick out of the over 400 I have myself, it is not. And so IT IS NOT. Not for the ones I sometimes like to try. That 30 - or 300 or whatever good ones do exist, does still not tell that Hires "works". Point is, when only a few % "work" then I can't find them.

And *when* I found one, the only thing really to tell is something like "at last, phew ... this doesn't sound worse than the CD version".

 

Peter

 

 

 

[PeterSt]

 

 

 

 

PeterSt :

So yes, to open another Pandora's box. Hires sounds way too soft to me. There's no life in it.

 

Pedal:

Now, this is an interesting statement. You do hear a difference, but you don’t like it.

 

Well, I like to emphasize this, also because it seems to be implied that "all who claim to hear a difference do not participate that (Julf's) Test".

So, for me this indeed works the other way around; I do hear the difference, but I never like the Hires.

 

When this would be throughout -so that assumed for now- it would come down to me applying to that test, to next pick the 16/44.1 decimated as the best sounding one ...

 

Fun.

 

[crisnee]

"So I came to Computer Audiophile as a believer in Hi-res. I am quickly loosing my faith."

 

Me too, I just thought, yeah it must be better. But I too am losing my faith. Not that I had a lot, I just thought at first glance, it made sense.

 

But folks here and the fact that I hear very little difference, and not necessarily a better difference, are making me wonder.

 

The folks here who so often restate that they love it so blah blah, but rarely if ever detail exactly what's so different. Just a vague sense that the music is just more musical, flows better, is finally real. And some go so far as to say that at long last it's finally really enjoyable.

 

So I guess the love I've had for recorded music was all a dream, where somehow I did what? to fool myself. For surely I must have been fooling myself all along, perhaps had an imaginary hi-res friend who filled in the extra bits and hz, OTOH it's not possible that the hi-res apologists might be fooling themselves. And what about those poor duffers that listened to the first Victrolas and exclaimed, is it real or is it Memorex, I mean shellac?

 

And then sometimes I wonder if some of us just have very different hearing than others. I don't mean better, or worse, just different. I don't know how that would work, somewhere the brain's involved I would guess, in the translation. It all doesn't make sense, unless it really is simply that people are fooling themselves.

 

If that's all it is, it's really discouraging, especially when it comes to sites like this. It's discouraging because there's really nothing to be done about it. You can't unfool yourself, the best you can do is consider that you might be foolin. Which ever side you might be on, as if there are sides!

 

And on the very other hand, does it matter if we fool ourselves, either way? No, except to our bank accounts, one way or the other. Either way we enjoy our way better, one way just costs a lot more.

 

Still, there's that lingering doubt, am I fooling myself? Am I missing something? Am I paying for naught? Am I a tin ear? I am a golden ear, right?

 

Back to Fred Neil. Listen to the first few lines of "Please Send me Somebody to Love" on even just an ok system and res of any kind won't make no. (unfortunately the recording has a lot of dropouts but the sound between them is good)

 

-Chris

 

 

 

 

[PeterSt]

OK, what you are talking about is the instantaneous signal value going up very quickly. if you measure at the output of the DAC or amp, the signal is a voltage, if after the speaker, it is sound pressure. But "volume" describes the RMS or peak-to-peak aggregate value of the repeating signal.

 

If you keep holding the signal high for several seconds, you still have a signal value (but classified as a DC value), but it has no volume (because no air is moved and no sound pressure is produced).

 

OK, what you are talking about is the instantaneous signal value going up very quickly. if you measure at the output of the DAC or amp, the signal is a voltage, if after the speaker, it is sound pressure. But "volume" describes the RMS or peak-to-peak aggregate value of the repeating signal.

 

If you keep holding the signal high for several seconds, you still have a signal value (but classified as a DC value), but it has no volume (because no air is moved and no sound pressure is produced).

 

 

Earlier in the thread I talked about my view on 1:1 and how it could be a good idea to follow that view.

But now you overdid it ... :-)

 

I wanted to put forward DC myself yesterday, but was sure that it would imply the wrong thing. A too long hold ...

 

So, it sure implies a temporal DC shift all right, but this happens throughout music. If you only look at the micro detail of it. Like in 500 subsequent samples. So, all the time the voltage is on the plus side now (or minus), and nothing changes for the making of music. Easiest to think about this is a sudden very steep transient in the bass, and next "DC" is up and stays there as long as the increase in general amplitude level to one side (oh yes) stays.

 

Julf, you are correct (especially look at the last quoted sentence).

Merely : I may not be able to explain what really happens electrically. I mean, when looking at the samples, them representing voltage, and what the effect of a transient is. An attempt :

 

(please remember, the transient as per my definition is a one-way up (can theoretically be down as well), not going down again (at least not as instantly))

 

Since the transient is to be seen as an instant "pin" going that one direction only, its effect will be similar to connecting that 1.5V battery I talked about. Connect + to + and - to - and the diaphragm will come forward - and stays there. The sound you hear seems to be like some "electrical connection", but all you heard was the instant coming forward of the diaphragm. A small, but very powerful push. An SPL change. And one time only. DC is shifted by +1.5V.

 

Now music;

There the same happens, but this time subject to the attack of the instrument etc. itself PLUS the decay which will be slower for the instruments with the fastests attacks (remember, synths are excluded). So, now the, say, 1.5V is applied again by the attack, the transient it as instant as with the battery, the "volume" increase (but please think sample's amplitudes) really is there, and while the increase was instantly, the decay is relatively slow and the amplitude drops again.

This is, of course, not the same as DC, but the temporal effect is the same; supposed the decay lasts 1ms, I am pretty sure that during that time the DC Offset stays for that time, but decays.

 

Side note : NOS1 users have some advantage in understanding here (as do I myself) because there's real time DC Measurement in it, with 0.1mV resolution;

What happens during music is easy to see and plainly confirms what I just said. So, DC Offset actually rides on the lowest frequencies, because they determine the "fundamental" (ouch) on which all further frequencies ride.

Maybe I shouldn't have said this because now more questioning (this) may arise, but this is how it works. And of course, this is right in the sample's amplitudes, with btw DSD as the best example to observe it (why ? because of that easy smallest "volume" step possible I talked about earlier).

 

So, true; DC has no volume as such, except for that one moment it is applied (or deapplied). This is the transient I talk about (all the time).

These transients happen throughout, although I would not state that this is each ms. But think about e.g. Mark Knopfler and his voice; Is this a frequency of say 50ms, or is it just a happening transient each 50ms (or whatever ms of course). I say it is the latter, and I always refer to it as on/off sound. It's actually squares for its rise time, but it is not for its slower decay. It is on/off, and may carry a frequency by *that*, although it is not allowed to call it a frequency as we tend to refer to it (like in sines). It still is an on/off frequency though.

 

Now hopping over to synths afterall ... it is exactly this on/off which does not show at all with any normal filtering means. Of course not, because the "On" got burried in the remainder of the wave, and will show the "lump" only as that same (DC) Offset I talked about. Amplitude level may be reached, but spread over numerous samples, averaging all.

 

And yes, with the filtering Miska proposes that seems to be solved because the attack is OK, but sadly the (indeed extra long) postringing now merges (and conflicts) with the next attack (but think in far more detail and in the technical transients). If that only comes (repeats) fast enough, with the notice that a violin will.

 

This view of mine happens when you just look into the waves (but take a couple of months to get experienced on it) which from there creates that 1:1 view. Not only because that's exactly what you would be doing, but because it should be done electrically. Thus, no fuzzying filters.

That this by itself creates some problems in the frequency domain is another matter ... which sadly is not so much "another" at all, because all is related and no real good way exists. Yea, sample high enough.

 

Peter

 

 

 

[Julf]

"Am I a tin ear? I am a golden ear, right?"

 

Well, at least I am a better than average car driver.

 

[Julf]

"Merely : I may not be able to explain what really happens electrically."

 

Are you familiar with differential equations?

 

 

[PeterSt]

Still, there's that lingering doubt, am I fooling myself? Am I missing something? Am I paying for naught? Am I a tin ear? I am a golden ear, right?

 

No man. Your system is just not resolving enough !

Wasn't that clear to you right from the start ?

 

But somehow this is key;

Those who claim to clearly hear the difference (Hires being the better one) have a common denominator : their system misbehaves or the presented Hires album is technically no good at all (was upsampled, was downmixed - anything).

 

By now this is close to statistics;

The number of "good hires" I was presented (hey, really, listen to this then !!" is numerous, and not one single time it was right. Yes, amongst those CA readers, who may or may not be ashamed to admit it by now (and some participating in this very thread).

 

Remember : NOT ONE TIME.

Still it was brought to me as THE example. It was their pick.

Their pick from something which clearly was better to them over any Redbook. Also of course their best example - not to fail.

 

HOW ?

 

[crisnee]

"Well, at least I am a better than average car driver. :)"

 

Can you prove it? I suggest a doubly blind test; and I'm perfectly aware that some people think the test will apply extra stress and weaken your abilities.

 

-Chris

 

 

 

[Julf]

"Can you prove it?"

 

Are you saying I don't know perfectly well how well I drive? Are you doubting my word? Are you claiming I am crazy?

 

Why don't you go for a drive yourself? You will see, with your own eyes, how bad most drivers are!

 

[Mark Powell]

There are three groups. those that have 'accidents' that are their fault, those who have 'accidents' that aren't their fault, and those who don't have 'accidents' at all.

 

The first two groups would not survive five minutes in a light aircraft. Not even the third group is totally 'accident' free. One idiot here recently crashed into a small village surrounded by open countryside. "The engine has never stopped before, so I don't see the need to regularly practice my forced landings." And other idiots call it an 'accident'.

 

[Julf]

"The first two groups would not survive five minutes in a light aircraft."

 

So I guess the fact that I did for all those hours in my logbook makes me belong to your third group? I think surviving 30 years of motorcycle riding has been more of a challenge.

 

I gave up flying when I didn't have the possibility to do enough regular flying to keep up my proficiency. You don't want a rusty pilot in a rather busy airspace.

 

[Miska]

This is, of course, not the same as DC, but the temporal effect is the same; supposed the decay lasts 1ms, I am pretty sure that during that time the DC Offset stays for that time, but decays.

 

You can see how the sound pressure decays from speaker step responses, like Figure 7:

http://www.stereophile.com/content/aerial-acoustics-model-7t-loudspeaker-measurements

 

And yes, with the filtering Miska proposes that seems to be solved because the attack is OK, but sadly the (indeed extra long) postringing now merges (and conflicts) with the next attack (but think in far more detail and in the technical transients). If that only comes (repeats) fast enough, with the notice that a violin will.

 

Which is still just roughly half of length of what many common linear-phase filters have at one side the impulse...

 

If you put overlay those with the step responses from speakers, you can see that it becomes partially relevant for tweeters.

 

This is of course solved by hires, at 88.2 the ringing is half the length, at 176.4 it is fourth and at 352.8 it is eight'. Then it pretty much fits inside the tweeter step/decay. But it requires that the original source rate is at least that high. Upsampling won't solve it.

 

 

[Mark Powell]

I have ridden big motorcycles for many years. Idiot car drivers are always pulling out. There is a UK 'Think Bike' safety campaign.

 

Monday a grandmother and two children were hit by two motorcycles while crossing a busy road at nearby Weymouth. Obviously, like so many car drivers, she thought they were 'only motorcycles' so could be ignored. Taught her a sharp lesson that did. The mororcyclists were unhurt.

 

[Pedal]

Both camps do the same mistake: -Hi-rez skeptics say its wasted money because all you get is upsampled red book. The hi-rez fans says CD sucks because of hard, brittle, uninvolving, etc sound.

 

But each format should be judged for its potential, not from a (randomly) bad implementation of it. To my ears hi-rez is superior when done correctly. Never mind the early hi-rez titles sounding more or less bad (most of them originated from old scrappy analog master tapes).

 

The same goes for Red Book 16/44. CD replay of today is completely free of any harsh or brittle sound. CD reply has become almost perfect in my ears. I hear no “additive” distortion from CD any longer. This can be experienced even from inexpensive DACs costing less than $1000 together with hard disk reply, substituting expensive CD transports.

 

Stepping up to dedicated high end DACs like my previous DIY Buffalo II, I think the quality of CD was at least on pair with good analog. The final frontier, however, was reached when I bought the NOS1 DAC. –You want believe how much low level detail and micro dynamics this DAC can dig out of your old CD collection! You don’t know the boundaries of Red Book unless you have listened to something like the NOS1 DAC.

 

 

[Pedal]

This hi-rez debate is not very important to me. And I didn’t spend that much ink on it, neither here on CA or at the www.phasure.com forum.

 

But hi-rez is here to stay. The early trials of SACD and DVD-A, started about 10 years ago, as physical media. Since then we have seen a rapid growth in computer audio and downloads. What we call “hi-rez” is already the standard format of recording. Practically all new music today is mastered at 24/96 or 24/192. More and more people have enough internet band width and an entry level 24/96 DAC costs pocket money and can be fitted in your pocket too. So, in the future “hi-rez” downloads will be the norm even among non-audiophiles. The selection of great sounding, modern recordings will increase and everybody will be happy.

 

 

[sandyk]

Peter

You are selling a US$3,500 DAC. For what, to listen to MP3 or CD only? (grin)

The fact is that many people are able to appreciate the differences between RBCD,and SACD and the other high resolution formats, and very few will ever get to own your DAC. This means that for most people the high res formats are worthwhile.

Kind Regards

Alex

 

[Julf]

"So, in the future “hi-rez” downloads will be the norm even among non-audiophiles. The selection of great sounding, modern recordings will increase and everybody will be happy."

 

I dearly hope you will be proven right. But watching the record industry and Apple, I think you can color me somewhat skeptic/concerned.

 

[PeterSt]

Are you familiar with differential equations?

 

Already before the age of Lotus123 I simulated what would happen to my annuity (?? -> annuiteiten) morgage throughout time.

This, at lacking the proper formulas at the time.

 

The other day (somewhere in CA) I posed the impossibility of calculating the chances with applying the Dutch Lotto and the number of forms to join in. It is not quite related to differential equations, but it shows how results can be obtained anyway by applying brute force "computer math". So here you go :

 

A.

1 1

2 2

3 3

4 4

5 5

6 7

 

B.

1 7

2 8

3 9

4 10

5 11

6 12

 

C.

1 1

2 7

3 8

4 9

5 10

6 11

 

Each are two "forms" to join that Lotto. Which pair gives the best chance for the most money ?

Of course you'd have to know the rules, and I don't even know them myself anymore.

 

Anyway, from computer simulation I got that joining in with 2700 forms would by guarantee give back 2/3 of the money. This optimized the combinations, and without optimization it could easily be 5% return money only.

 

At the time I worked at "Rijkswaterstaat" (what about differential equations ?) and the majority of work on the mainframes was about waterspeeds, througput etc. etc.

Not there, and not at any time later I could find someone able to do this math with formulas.

The base of it, of course exists (about "including laying back" - I hope you know what I mean), but in this Lotto "combinations" ... not that I know of.

 

An even better example could be me hunting the formula for primes. I got into the subject because of ever recurring prime number contests. Who (in computers) has not done that.

Of course I won the contest, after ONE YEAR everybody being allowed to improve over the other, which was about the one being a few cpu cycles over the other "and where can we gain some". But, because this ended up in the most sneaky own-found rules (applied in the Fortran program), I was close to a formula in the end. Only once in the few thousand numbers the formula didn't work. It looked like a matter of time to get there, but the time span already was a year.

And I liked fishing better.

 

Peter

 

PS: To answer your question ... No - sort of :-)