Which best represents your opinion of 96kHz high res recordings vs. redbook?

[Jud]

If you compare the speed of sound to the speed of electrons moving through circuitry, how close are they?

 

Not close at all, of course. Why?

 

Up early to watch the Tour?

[AlainGr]

Hi Alain. Doesn't matter, it's how the filter is written/programmed to operate. So no matter whether inline or offline, a filter with both a passband and a stopband at a particular frequency is impossible. You could think of as slow an action as you'd like - a traffic cop, let's say, deciding whether a signal should pass - and if the cop must decide both to let the signal pass and​ to stop it, well, he has an impossible job.

It is probably the wording that I have trouble with, but I myself have some filtering to do to fully understand the notion In offline mode of course

[Jud]

It is probably the wording that I have trouble with, but I myself have some filtering to do to fully understand the notion In offline mode of course

 

It is impossible, so if you don't understand how it could happen, that is probably a good sign.

[wgscott]

Not close at all, of course. Why?

 

Up early to watch the Tour?

 

It would mean you don't have to worry about the time.

 

No. Kids have a swim meet.

[AlainGr]

[RealAudio]

Go take your shit in someone else's thread for a change, you senile twit.

 

 

Don't you think that's rather harsh.

[Paul R]

If you compare the speed of sound to the speed of electrons moving through circuitry, how close are they?

 

Have to be a little more specific eh? If you are talking drift speed, the speed of sound is much faster indeed - and it also varies greatly with the material and amperage involved. All those little electrons zipping along in random directions smashing about with the atoms of the conductor don't move very far...

[Jud]

It would mean you don't have to worry about the time.

 

Not so. Time is one way of looking at it, but it's the underlying math of the required filter performance being impossible that is the underlying cause. A very loose analogy would be the double-slit experiment: intuitively you think the resulting interference pattern is caused by two waves present at the same time and interacting, but the pattern occurs even when only one electron is allowed through a slit at a time.

 

Maybe a more intuitive way to think of the filter is to picture a scope trace: the signal is full amplitude from 1-22050Hz, then drops absolutely vertically to zero at 22050Hz and remains there for all higher frequencies. So you have two right angles and a vertical line precisely at 22050Hz. It's that vertical line, requiring the filter to simultaneously completely pass and completely stop 22050Hz, that is physically impossible but mathematically required if a digital signal is to be perfectly reconstructed into analog per the Nyquist theorem. Because such a true brickwall filter doesn't exist, we have to contend with filtering artifacts.

[bogi]

If you can't detect a difference with 96kHz, why bother with 192kHz or DSD?

 

I can hear difference between PCM and DSD on my DSD capable DAC and DSD option sounds me better, therefore DSD has sense for me.

 

The previous sentence is related to Redbook to DSD on the fly conversion. So it is not about source file format, but about audio data processing in computer and DAC.

 

We can speak about audible or inaudible differences of various data formats (inc. various sample rates). If we hear a difference, we cannot be sure if the difference is caused purely by different source data formats, or if it is caused by some difference in audio data processing. So the audible difference can be (partially) related (also) to software and hardware used in audio chain.

[beanbag]

If you spent more time reading and understanding and less time thinking of arguments, explanations would penetrate more easily.

 

The reason things are impenetrable is because you insert various bits of irrelevant physics and because some of the things are mostly correct, and some very wrong. It doesn't help that I don't say things precisely either.

 

An "ideal" brickwall filter is shaped like a step function. The transition is infinitely sharp, but the pass band and the stop band do not "overlap". It's still a single-valued function, which means that any given frequency has a particular attenuation, and not both.

 

You're right that any real brickwall filter has a finite narrow pass band. The reason is due to a concept you've mentioned many times - namely the inverse relation between frequency range and time range. A filter with infinitely narrow transition band takes an infinitely long time to implement because it has to count infinitely many cycles of the wavefrom before it can tell 22050Hz apart from 22050.000001 Hz. (delta_f *delta_t = 1 or 1/2 or pi, or something close to one) Any real filter only has a finite time to operate (e.g. the length of the song, or the number of samples you can hold in your filter algorithm) so that's why it has finite transition band.

 

To say that something acts "instantly" from one frequency to another is to get the time and freq domain effects mixed up. The brickwall doesn't act instantly, not because of "speed of light" but because of the opposite - it takes really really long.

 

This signal analysis stuff has to do with fourier transforms and etc, and not special relativity, speed of light, two slit experiment, etc etc.

[Jud]

I think the important thing here is that you have agreed an ideal brickwall filter is not possible to achieve. For the rest, I'll go through what you've said here and point out how we are actually expressing the same thing in different ways.

 

The reason things are impenetrable is because you insert various bits of irrelevant physics and because some of the things are mostly correct, and some very wrong. It doesn't help that I don't say things precisely either.

 

An "ideal" brickwall filter is shaped like a step function. The transition is infinitely sharp, but the pass band and the stop band do not "overlap". It's still a single-valued function, which means that any given frequency has a particular attenuation, and not both.

 

You're right that any real brickwall filter has a finite narrow pass band. The reason is due to a concept you've mentioned many times - namely the inverse relation between frequency range and time range. A filter with infinitely narrow transition band takes an infinitely long time to implement because it has to count infinitely many cycles of the wavefrom before it can tell 22050Hz apart from 22050.000001 Hz. (delta_f *delta_t = 1 or 1/2 or pi, or something close to one) Any real filter only has a finite time to operate (e.g. the length of the song, or the number of samples you can hold in your filter algorithm) so that's why it has finite transition band.

 

You've pointed out that any real brickwall filter has to have a finite transition band, and that the time required for such a filter to operate moves further toward the limit of infinity as the size of the transition band shrinks toward the limit of 0. Now picture in your mind the graph that accompanies an ideal brickwall filter where the limit of 0 has been achieved. Not "really, really small," but 0. The amplitude v. frequency graph for that ideal filter will show an absolutely vertical line. The line can't be other than vertical because that would mean the transition band had a greater than zero width, right? Now for the transition band to have zero width, you cannot name any two frequencies such that one is immediately on one side of the transition band and the other is the next higher frequency on the other side of the transition band, because I can always name a frequency in between them, right? So no, there is no such thing as a non-overlapping passband and stopband for an ideal brickwall filter. That completely vertical graph line tells you the passband and stopband must overlap at a single point along the frequency axis, the point where that vertical line occurs. That's the point at which the filter must as a mathematical absolute take infinite time to discriminate whether the frequency belongs to the passband or stopband because it belongs to both and thus the question is not decidable. If there was no overlap between passband and stopband, then no matter how small the division, the filter would eventually be able to discriminate - after the heat death of the universe, perhaps, but in principle it could eventually discriminate. Only where the frequency-vs.-amplitude graph line of the filter is truly vertical and there is absolutely no way for the filter to discriminate, no matter how long, can we say the time needed for the filter to operate is truly infinite.

 

To say that something acts "instantly" from one frequency to another is to get the time and freq domain effects mixed up. The brickwall doesn't act instantly, not because of "speed of light" but because of the opposite - it takes really really long.

 

These are two ways of saying exactly the same thing, and if you research you will find articles and web pages that use one or the other or sometimes both. You can say the filter requires infinite time to discriminate, or, exactly equivalently, you can say the filter's action must be instantaneous in order for it to operate, because if it takes any time at all it must take an infinite amount of time.

 

Another way to think of this is as the filter "telling" the signal which amplitude it must have along the amplitude (y) axis of the graph at the single point along the frequency (x) axis where the passband and stopband overlap. The filter has to tell the signal to be at full amplitude, at zero amplitude, at 3/4 full amplitude, etc. There are an infinite number of points in the vertical line segment between zero and full amplitude, and the filter has to tell the signal to go through each of them. So you can say that if telling the signal to go through one of these points takes any time at all, then telling the signal to go through all of them takes an infinite amount of time. Or you can turn that around, and with equal validity express it as the filter having to have instantaneous effect to be able to tell the signal to go through an infinite number of points in anything other than an infinite amount of time and thus even in principle be able to operate at all. A filter passing information instantaneously would be sending that information faster than the speed of light and thus violating special relativity.

 

This signal analysis stuff has to do with fourier transforms and etc, and not special relativity, speed of light, two slit experiment, etc etc.

 

As I noted when referring to the two-slit experiment, that was a very loose analogy by way of saying to Bill, when he referred to the relative speeds of sound and electricity, that it isn't the speed/timing of the waves/particles involved that you have to be concerned with, it's the underlying fundamental math and/or physics requirements. I was in no way indicating that quantum uncertainty and wave-particle duality are involved at all in the Shannon-Nyquist theorem and brickwall filtering.

 

Regarding special relativity and the speed of light, I've explained above why this is an equally valid way of looking at the same problem/issue you express as the filter taking infinite time to discriminate between two frequencies. If you don't happen to like that view, that's fine. As I mentioned at the beginning of this comment, the more important thing is that you agree an ideal brickwall filter is impossible in reality. This being the case, we have to deal with filtering artifacts, and that is the reason in principle for providing recordings at higher than 44.1Hz resolutions and/or oversampling in the DAC - to do a better job of dealing with those artifacts or not creating them in the first place.

[Paul R]

Or to boil that all down, in English, any PCM filter that can currently be implemented will screw things up either in the time or frequency domain, or in both, equally or not.

 

In general, the more complex the filter, the worse it will muck up one domain or the other.

 

And again in general, the simpler the filter, the less audible damage it will likely do.

 

it is therefore quite reasonable that a filter designed for 96k sample rates can do less audible damage than a filter designed for 44.1k sample rates, and an audible difference in the sample rates is possible for that reason alone.

[mayhem13]

Or to boil that all down, in English, any PCM filter that can currently be implemented will screw things up either in the time or frequency domain, or in both, equally or not.

 

In general, the more complex the filter, the worse it will muck up one domain or the other.

 

And again in general, the simpler the filter, the less audible damage it will likely do.

 

it is therefore quite reasonable that a filter designed for 96k sample rates can do less audible damage than a filter designed for 44.1k sample rates, and an audible difference in the sample rates is possible for that reason alone.

 

Quite right.....beyond reasonable but actually factual......

 

But the point is it simply doesn't matter to us mere mortal humans.....we can't hear it's effects and there's far more important audible factors out there that CAN be heard and demand far greater attention than Redbook vs HiRes.

 

I'm a speaker guy first and foremost....and therefor it would be quite unfair of me NOT to accept and expose just how poor current speaker technology is compared to the source/digital side of things. Equate it to watching a 4k movie through a dirty window....or better yet, ask some of the recording experts here on the limitations of microphones.

[Paul R]

Quite right.....beyond reasonable but actually factual......

 

But the point is it simply doesn't matter to us mere mortal humans.....we can't hear it's effects and there's far more important audible factors out there that CAN be heard and demand far greater attention than Redbook vs HiRes.

 

I'm a speaker guy first and foremost....and therefor it would be quite unfair of me NOT to accept and expose just how poor current speaker technology is compared to the source/digital side of things. Equate it to watching a 4k movie through a dirty window....or better yet, ask some of the recording experts here on the limitations of microphones.

 

Thing is, you can usually see outt of a dirty window, however imperfectly. Same is true for speakers, if the speakers are any good at all you can usually hear through them. Imperfectly.

 

Paul

[DigiPete]

. . .

I'm a speaker guy first and foremost....and therefor it would be quite unfair of me NOT to accept and expose just how poor current speaker technology is compared to the source/digital side of things.

Equate it to watching a 4k movie through a dirty window....or better yet, ask some of the recording experts here on the limitations of microphones.

 

 

 

Yes, speaker technology is not ideal!

Worse, most speaker setups repeat the same sub-optimal use of said technology again and again.

Most commonly:

 

- Unsuited cabinet shape: Square boxes with sharp edges and corners resulting in diffraction and resonances.

- Unsuited cabinet material: often low stiffness flat wooden with low resonance freq that are difficult to dampen.

- Use of passive x-overs: making life difficult for the poor amp and introducing distortion and phase change.

- Pot luck choice of amps: allowing users an extra chance to screw things up

- Separate high/mid drivers: confusing the brain where it is most sensitive

- Amp placement: far from drivers adding cost and loss of long speaker cables

- Room compensation: or rather lack of any.

 

How about we fully exploit available technology before we pine for something new to screw up?

[Jay-dub]

98% I'm for option 7 (there are practically perfect sounding brickwall filters, but over 90% of CD's are mastered with bad sounding ones). 2% I leave myself open to option 2, not that I can hear a 22kHz sine wave, but I think there is some direct time-domain processing in the human ear that can be affected by ultrasonic components.

[Jay-dub]

I might add, in response to the discussion of ideal brickwall filters being impossible, that in my opinion trying to approximate one is a sure way to make your filter bad-sounding. A good-sounding 20kHz brickwall should have stopband at least 2500 Hz higher than the passband, otherwise the impulse response will not be sufficiently localized.

[Jud]

I might add, in response to the discussion of ideal brickwall filters being impossible, that in my opinion trying to approximate one is a sure way to make your filter bad-sounding. A good-sounding 20kHz brickwall should have stopband at least 2500 Hz higher than the passband, otherwise the impulse response will not be sufficiently localized.

 

Then it is not a brickwall filter. "Brickwall" is so named for the nearly vertical line, or "brick wall," between pass band and stop band. You're describing a gentler slope, so not really a brick wall.

[4est]

Gosh how I hate it when someone else claims something doesn't matter to me. Speakers may be the single most important item in a stereo, but your insistence that speakers are the only area worthwhile pursuing is narrow minded at best. This thread is about sample rates not speakers.

Quite right.....beyond reasonable but actually factual......

 

But the point is it simply doesn't matter to us mere mortal humans.....we can't hear it's effects and there's far more important audible factors out there that CAN be heard and demand far greater attention than Redbook vs HiRes.

 

I'm a speaker guy first and foremost....and therefor it would be quite unfair of me NOT to accept and expose just how poor current speaker technology is compared to the source/digital side of things. Equate it to watching a 4k movie through a dirty window....or better yet, ask some of the recording experts here on the limitations of microphones.

[Jay-dub]

OK, Jud. Fine. "Lowpass." But: the filters I like are still much steeper than what can be achieved with low-order analogue lowpass filters, and I use the "sinc" effect in Sox. The reconstruction filters in typical CD players have around 5% transition bandwidth, and are frequently referred to as brickwalls; I like 12%.

[Jud]

OK, Jud. Fine. "Lowpass." But: the filters I like are still much steeper than what can be achieved with low-order analogue lowpass filters, and I use the "sinc" effect in Sox. The reconstruction filters in typical CD players have around 5% transition bandwidth, and are frequently referred to as brickwalls; I like 12%.

 

Have you tried/are you able to play around with oversampling prior to the reconstruction filter?

[mayhem13]

Gosh how I hate it when someone else claims something doesn't matter to me. Speakers may be the single most important item in a stereo, but your insistence that speakers are the only area worthwhile pursuing is narrow minded at best. This thread is about sample rates not speakers.

 

I'm not claiming it doesn't matter to you.........I'm simply stating that from a practical standpoint, it shouldn't.

 

That's a strong word, practical or practicality when you examine it's definition and explore its total disregard within the audiophile community. The demise of practicality is even happening here on a COMPUTER audio site where a computer's main function is to make our live's easier.....or simpler.....or...oooops.....here it comes.....practical?

 

Again,,dirty windows but instead of cleaning the glass, we're marketed a better view? You'd need to do the the work yourself in this case as to what's audible and what isn't. And as to change vs improvement?......that's a philosophical fundamental and I can't help you there but can only add, one needs to know the difference before deciding on anything. Successful marketers answer this question for you all the time and they know the route of the campaign from the onset........it's either to improve your life or some function within it OR to substantiate you. It's up to you to quantify both.

 

Enjoy!

[Paul R]

The demise of practicality is even happening here on a COMPUTER audio site where a computer's main function is to make our live's easier.....or simpler.....or...oooops.....here it comes.....practical?

 

I just want to point out that there are some misconceptions here - computers were invented to make impossible things possible - to make it practical to say, compute monthly billing statements from literally billions of call records. Or to fly very large and very complex aircraft both safely and accurately. Or to do mathematical computations in moments that would take a team of human mathematicians their entire lives to accomplish.

 

Practical in this sense means - able to be practiced. In the same sense as a Doctor or a Lawyer sets up a "practice".

 

They were never meant to be "simple" or "easy to use." It actually took a genius named Steve Jobs to make "easy" "natural" "simple" and a few other similar terms become not only practical when dealing with computers, but *expected* and *demanded* as well.

 

Computer Audio has two major things opposing it being something simple - first it is a young field, and there is still much argument/disagreement about what is and what is not true about it. Second, Audiophiles by nature tend to be picky detail oriented people- and nothing is simple when you demand to know the details and understand everything that is happening. (grin)

 

Now, back to your regularly scheduled program...

 

-Paul

[4est]

What if you are very pleased with your speakers as I am? My life is in upheaval and I haven't the space for my Soundlab ESLs that are in storage. While I await the opportunity in that area, I am pursuing others avenues. Maybe I am daft, but that seems a practical to me. Changing speakers is not always a solution, nor desired. To attempt to distil it into only that is short sighted. To push that view constantly while others are attempting to discuss something unrelated is inappropriate and rude. You may know an awful lot about speakers, but time and again you have displayed that you do not understand me or many other posters. Maybe it is time you give us that DIY monitor project you have mentioned. Give us something to work with. Better yet, how about a full on attempt at a full range one that will compete in a price class you feel is representative- say 5-10k. If what you say is true, we'll be floored, and not feel the desire to play with software and such.

 

Seriously, it is far easier to be a critic when you have nothing on the table. Show us what you can do. Need a cabinet or something built? I'd consider doing it for free for the cost of shipping for a prototype or even a production run if it is fiscally viable. Bi amp or tri amp if you care to. That will make it easier to model perhaps, but no DSP please.

I'm not claiming it doesn't matter to you.........I'm simply stating that from a practical standpoint, it shouldn't.

 

That's a strong word, practical or practicality when you examine it's definition and explore its total disregard within the audiophile community. The demise of practicality is even happening here on a COMPUTER audio site where a computer's main function is to make our live's easier.....or simpler.....or...oooops.....here it comes.....practical?

 

Again,,dirty windows but instead of cleaning the glass, we're marketed a better view? You'd need to do the the work yourself in this case as to what's audible and what isn't. And as to change vs improvement?......that's a philosophical fundamental and I can't help you there but can only add, one needs to know the difference before deciding on anything. Successful marketers answer this question for you all the time and they know the route of the campaign from the onset........it's either to improve your life or some function within it OR to substantiate you. It's up to you to quantify both.

 

Enjoy!

[mayhem13]

The demise of practicality is even happening here on a COMPUTER audio site where a computer's main function is to make our live's easier.....or simpler.....or...oooops.....here it comes.....practical?/QUOTE]

 

I just want to point out that there are some misconceptions here - computers were invented to make impossible things possible - to make it practical to say, compute monthly billing statements from literally billions of call records. Or to fly very large and very complex aircraft both safely and accurately. Or to do mathematical computations in moments that would take a team of human mathematicians their entire lives to accomplish.

 

Practical in this sense means - able to be practiced. In the same sense as a Doctor or a Lawyer sets up a "practice".

 

They were never meant to be "simple" or "easy to use." It actually took a genius named Steve Jobs to make "easy" "natural" "simple" and a few other similar terms become not only practical when dealing with computers, but *expected* and *demanded* as well.

 

Computer Audio has two major things opposing it being something simple - first it is a young field, and there is still much argument/disagreement about what is and what is not true about it. Second, Audiophiles by nature tend to be picky detail oriented people- and nothing is simple when you demand to know the details and understand everything that is happening. (grin)

 

Now, back to your regularly scheduled program...

 

-Paul

 

Aaaaagh........a good question or two raised!

 

Is computer Audio really a 'young' field? Think about that for a moment when you consider the huge volumes of data or 'bits?' processed that have come before this site and other like it. More marketing maybe?........at least consider the possibility.

 

.....and details........yes the details. How important the details because god knows we clearly don't have enough to think or worry about already. Should I be watching world news or the financial reports instead of debating this topic with you?.........probobly as both of those have a better shot of making my life better...........but more details?

 

Maybe the zombie masses with earbuds and twitter accounts aren't so mindless after all?

 

Practicality.........think on it.........and pay it forward.