'HD Download debacle' investigation published on HFN&RR June 2011

How many of the folks who make arguments about "Nyquist's theorem" believe Harry Nyquist heard digital audio?

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

I know I was just clarifying that because it seems to come up as a misunderstanding all the time haha.

Isn't that similar to asking whether or not Claude Shannon knew that digital electronics could be implemented for audio applications?

 

How come complete amateurs have no problem showing the complete lineage of a recording but these people can't? If you are getting a download you better demand it or don't buy it. Verification. Ever seen one with even a checksum file?

 

I've only ever seen a checksum from folks who rip and distribute online. Not saying that as a promotion for gray area downloads BUT! it demonstrates how serious the fans take the music. Companies are trying to sell us stuff and they can't be bothered to provide evidence that we are getting what we are paying for.

Most people don't care. Or better put, most people don't know what a checksum is nor why it is important to have one. This of course also contributes to most folks not caring what the lineage is etc. Yet, for those that do (and such pricing is certainly targeted towards us that do) why can't the companies take the small amount of time to simply let us know with evidence, if the stuff is what it is supposed to be, we'll pay

 

It's been a long time since I studied this, but I do not that statement is completely correct, and that is one of the big misunderstandings in audio.

My gut rememberance is that theorem applies in some perfect world which simply does not exist in real recording chains.

http://en.wikipedia.org/wiki/Nyquist–Shannon_sampling_theorem
says in part
"...The theorem assumes an idealization of any real-world situation, as it only applies to signals that are sampled for infinite time; any time-limited x(t) cannot be perfectly bandlimited. Perfect reconstruction is mathematically possible for the idealized model but only an approximation for real-world signals and sampling techniques..."
(oh yeah, THAT's what I meant to say...)

 

I'm going to propose a crazy idea. Absolutely nuts. You'll probably tell me I should head straight to the funny farm.

But I think that before HDTracks or any other download vendor offered these albums for sale, they should have listened to them to determine if they were actually any advantage over the redbook versions. If you're selling an allegedly premium product like this, for a premium price, then for the sake of your reputation, you do not want to sell the same old CD-quality files that are upsampled. Nor do you want to be selling albums that really offer no sonic advantage over the physical CD. The way to determine this is by careful listening. Also, they should obviously check the spectrum of frequencies for signs that they are "cheat" hi-rez files.

It is possible they innocently offered what they thought were high resolution sourced files, but if they did not take the minor effort involved in listening and checking for signs of a frequency cutoff (i.e. 22.1 kHz, for example), that seems a bit sloppy to me. Or naive. As a famous president once said, "Trust, but verify."

 

And I'd also like to know what ADC's and DAC's someone is using if they can't hear a difference between a 44.1khz signal an the analog original.

 

It's not even necessary to listen to the files. You just need to open them with software that does Spectrum analysis. It's as simple as that!...

 

Hi Jody,

I am an Electrical Engineer and have been working with DSP systems for many years. I often get that debate and it's quite easy to explain. You see, an ADC and a DAC do their job properly if the reconstructed analog signal is indistinguishable from the original. Because anlog signals are subjected to noise at a "microscopic" level, that means that the reconstructed one will be subjected to noise as well. For example, a pure sine wave with 0% harmonic distortion(only in theory), fed into an ADC, and into a digital signal processor acting as a talk through interface, and out through a DAC will output a sinewave with the exact same characteristics(provided that the anti-aliasing and anti-imaging filter cutoff frequencies are > twice the bandwidth of the audible frequencies, along with the sample rate of the circuitry).

If no DSP operations are performed on the signal(EQing, filter, etc.), then my above statement holds true. However, you are probably thinking of characterstics of the signal comming in and out of the ADC and DAC. Please remember that those two system blocks are different than the DSP(which does the processing and EQing).

Regards

 

The components used to implement the whole ADC/DAC chain would have to be ideal to be able to implement the Nyquist theorem.
The brickwall filters typically used for 44.1 are not able to contain all the energy from an impulse signal, and output pre-echo and post-echo signals which smear the original analog signal.
This white paper explains it a lot better then I:
http://www.ayre.com/pdf/Ayre_MP_White_Paper.pdf

 

Thanks to both of you

I guess I can still learn while on summer vacation from school haha. One question is that if you took a mastered track and encoded in a lossless file of 44.1kHz, and another one with 96 or 192kHz, how many people can actually hear the difference? Are people comparing 16/44 files from a CD to needledrops at 24/96? Because that would make more sense.

 

Hi,
So I'm still curious - which ADC's and DAC's do you think "do their job properly" and reconstruct a 44.1khz digital signal that is indistinguishable from the analog original? For arguments sake, lets say the original is a 1/2" two track tape of rock 'n roll music recorded in the 70s.
Cheers,
Jody

 

Their quality control seems highly suspect, having multiple instances of upsampled audio being passed off as Hi-Rez and dropping tracks on certain releases. It is not like they have a gigantic catalog, where a few mistakes could be explained away. HDTracks apparently puts up for sale on their site any audio the original label gives them without much investigation.

 

That would be astonishing news to the many many people who have implemented Shannon-Nyquist successfully over the years.

If you meant that ideal theoretical performance cannot be achieved, that's a different thing. It's also true of many theorems used in applied science (not just digital audio), yet systems based on them still 'work' for their purposes. And that's true of digital audio. If the imperfections are kept below audible thresholds, the system is effectively working 'perfectly'

That's a white paper touting a company's technology. It's written almost as if oversampling had never been invented (even though their own filters use 16X oversampling). But the crux is statements like this,presented without proof:

"This time-smear is interpreted by the ear-brain as both a lack of image precision in the soundstage, and also a subtle smearing of the musical sounds together."

and

"But redistributing this same total energy leads to significant gains in musical realism."

And this *is* the nub...whether tweaks to technology have audible benefits.

If Ayre has really done several man-months of listening tests in support of their product, they should publish those experiments.


I suggest reading this white paper instead:

Sampling, Oversampling, Imaging, Aliasing

which concludes:

 

Nyquist was significantly involved in the first practical implementation of PCM for audio transmission at Bell Labs during WW2 and would have been continuously exposed to digital audio in a professional context until his retirement in the mid 1950s, if not thereafter.

 

That probably depends on the way you wish to determine "indistinguishable."

 

No surprise

I haven't heard a downloaded hi res file that I thought sounded great.
I've done Linn and HD Tracks to sample the product. Didn't return to either site.
Perhaps Wall Street bankers have been brought in to "enhance" the business model. HD music derivatives. Oh boy.

 

Indeed.

Not that it matters. The argument Barry was trying to advance here was specious from the start. Nyquist could be deaf, blind, dumb, or or an autistic monkey and that would have no bearing on the validity of his theorems (of which the sampling theorem is only one).

 

I concede that my assumption that most readers over the age of ten would have immediately identified the fallacious reasoning was probably optimistic.

 

Hi Cam,

There is no "fallacious reasoning", other than that which leads some Internet cowboys to draw conclusions based on a simple question.

What was the state of "PCM for audio transmission" during WW2 (or in the mid 1950s)?
How much music recording do you think those systems were used for?
And if any, what was the state of the gear used to evaluate said recordings?

My point was - and remains - that which Yogi Berra said so eloquently:
"In theory, there is no difference between theory and practice but in practice, there is."

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

That's probably the smartest thing Yogi Berra ever said!

 

I'm sorry. Keith is an idiot with some axe to grind for lack of a better expression. He clearly doesn't understand the fact that analog master's recorded to CD sound far worse than those same masters recorded hi-rez A/D. CD's simply cannot handle the full bandwith of the analog master w/o compresssion/normalization.

I'm not sure I buy this conspiracy theorist. I have that frampton DL and it blows away any red book copy I have even if upsampled to 24/96.

 

Uhhuh.