Music Player Network Home Guitar Player Magazine Keyboard Magazine Bass Player Magazine EQ Magazine
Page 1 of 42 1 2 3 ... 41 42 >
Topic Options
#441389 - 04/05/01 01:38 AM George, Watch this!!!....(96k)
Nika Offline
MP Hall of Fame Member

Registered: 03/22/01
Posts: 2938
Loc: Ft Wayne,IN,UNITED STATES
You're about to observe me implanting my large foot into my gaping mouth!! Everyone gather 'round!!!


George,

I noticed on another thread that you lamented that the PCMCIA bus wouldn't handle the throughput for 96kS/s. This is just begging for me to ask your opinion on 96kS/s recording/mixing/effecting/mastering.

Before you dive in I think it fair to give you the premise that I come from. I'd recommend reading my comments about this on Roger's forum so that we don't end up revisiting paths that have already been explored. The good reading starts way down in the following topic starting with the posts from 3/5/01 at 10:33AM. In there somewhere are two long posts of mine that give you a synopsis of where I come from on this.

Thank you, George. You're the first Ultra Professional recording engineer I've talked to who has expressed much of a desire to have 96k ability. I'm just curious as to why that is.

Sincerely,
Nika.


Topic thread:
http://www.musicplayer.com/ubb/Forum2/HTML/001473.html

This message has been edited by Nika on 04-04-2001 at 11:27 PM
_________________________
For more...

Top
#441390 - 04/05/01 03:42 AM Re: George, Watch this!!!....(96k)
Anonymous
Unregistered


I know I'm not gm, but...

A pal of mine is testing out the RADAR 24. We ran tones through it, recording them at 24bits. We swept the tones up to 24k. At 44.1k, we could still hear the tones when the Radar just quit recording anything. In other words, thanks to Nyquist we lost high end. At 96k this obviously didn't happen, and the tones were still recorded up past the point where I could hear anything even at what I'm sure was dangerous volume.

No I can't tell you what the monitors were - but I'll find out.

Oh yeah, and otherwise I thought the RADAR sounded pretty good (not better than tape, just different). Anyone else using this unit, or want me to try something let me know.

Julian M

Top
#441391 - 04/05/01 09:12 AM Re: George, Watch this!!!....(96k)
lrbreez Offline
Gold Member

Registered: 04/08/00
Posts: 675
Loc: Walnut, CA, USA
At the NAMM show MOTU gave a demonstration on the merits of 96K. They showed that there is a definite difference (for the better) with 96K. It was very impressing!
Buddy

Top
#441392 - 04/05/01 12:41 PM Re: George, Watch this!!!....(96k)
Nika Offline
MP Hall of Fame Member

Registered: 03/22/01
Posts: 2938
Loc: Ft Wayne,IN,UNITED STATES
I know of the demonstration that MOTU does. I have seen it and critiqued it a few times. While I need to be cautious what I say, I will say that that demonstration does in no way refute anything that I have said above.


*watching my words carefully*:

I think it would be unfortunate if people were to buy into the merits of 96k because of the presentation that MOTU does. Next time you see this, ask if ANYBODY at MOTU can hear the difference that they are trying to visually paint. I did.


There. I think that that wording is subtle and safe. I'd be happy to take further discussions of that offline.
_________________________
For more...

Top
#441393 - 04/06/01 10:33 PM Re: George, Watch this!!!....(96k)
alphajerk Offline
MP Hall of Fame Member

Registered: 03/06/00
Posts: 7950
Loc: asheville nc usa
[what] IS 96khz going to give me? especially for twice the HD space and in a lot of instances half the tracks. im curious, i can hear a difference between 48 and 44.1 [not on a bass per se but on stuff like cymbals and such] i havent the chance to play with 96 khz yet but im not sure im going to hear much of a difference [wont know until i try though]. is it better because the LP filters are set higher up or is it better with stuff like EQ so the phase shift can "float on out" into the higher frequencies?

dave derr remarked about his Fatso JR not being able to be digital due to the low sampling rates but from what i've read about it, it seems awfully close to what david hill is doing DIGITALLY with his HEDD [albiet at 192khz, although Jules loves his @ 44.1]

that coupled with most mics just dont have the frequency response above 22khz anyways and neither do most pres, earthworks claims they do but not many others. not to mention our ears arent capable of detecting it either.

so are we gaining in the recording or with the processing at 96khz?

im not sure if this is what nika was after with his question or not.

bit depth is VERY EASY to tell differences. in fact you can easily illustrate it visually with you monitor operating at 16/24/32 bit depths and it is VERY apparent... however i cant think of an analogy for sampling rate.
_________________________
alphajerk
FATcompilation
"if god is truly just, i tremble for the fate of my country" -thomas jefferson

Top
#441394 - 04/06/01 11:55 PM Re: George, Watch this!!!....(96k)
Anonymous
Unregistered


A 15kHz sine wave being sampled at 44.1 is "having its picture taken" only 3 times per cycle. We listen to much more complicated (and interesting) waveforms in that frequency range. Things as complex as say distorted guitar will "look" mighty similar to that sine wave to the converter won't they?

julian m

Top
#441395 - 04/07/01 12:44 AM Re: George, Watch this!!!....(96k)
Nika Offline
MP Hall of Fame Member

Registered: 03/22/01
Posts: 2938
Loc: Ft Wayne,IN,UNITED STATES
Quote:
Originally posted by spoomuz@hotmail.com:
A 15kHz sine wave being sampled at 44.1 is "having its picture taken" only 3 times per cycle. We listen to much more complicated (and interesting) waveforms in that frequency range. Things as complex as say distorted guitar will "look" mighty similar to that sine wave to the converter won't they?

julian m


Julian,

You make a great point, and this becomes a fantastic argument. In fact, this is the point that Mark of the Unicorn makes in their presentation that was mentioned above.

They show that the number of samples done per waveform increases as we increase the sampling rate. This should become more necessary and audible as you approach the Nyquist frequency, such that at 10k, instead of sampling a waveform 4 times, we now sample it 8 times. This will most assuredly translate into better resolution by the time it hits our ears. Even better, at 1k (clearly in the human hearing range and even in many instrument's ranges including the human voice) we sample it 96 times instead of only 48 times. This means that we sample it 48 times per half/waveform - a doubling of the resolution. For this matter, 192k and even 384k will only improve this fact farther.

To restate your point, as we approach the Nyquist frequency, the resolution of high frequency information increases proportionally.

Unfortunately:

This argument is one of the great fallacies of digital audio, and is fast becoming one of the best marketing hoaxes in our industry.

The Nyquist frequency is dependant upon a lot more than just the obvious notion that we need to sample both at a high and low point in the waveform in order to capture the frequency of the waveform accurately. It is actually dependant upon a mathematical principle that states that a sine wave can only be drawn through two given points one way. This, in combination with the anti-aliasing filters used dictates that even with only two given points, so long as they don't both fall on the infinitely small zero cross point, the waveform that was originally sampled will be accurately reproduced by the digital to analog converters.

Thus, even if we only sample the 10k wavefrom twice in each half cycle, the D/A converters actually compensate for that by COMPLETELY filling in the rest of the waveform in the only mathematically possible way. Further, even if the points sampled do not hit the peak, the peak will still be represented by the D/A converters. Even at 20k, where we are only sampling once in each half of the waveform, and even if that once is infinitely close to the zero crossing point, the complete waveform will be reproduced by the D/A converters.

My immediate inclination was to doubt this, but upon vast research and testing, this phenomena that looks to be awkward when you look at the computer screen is actually 100% correct. This is the basis of MOTU's demonstration, that the signal LOOKS better on the screen. Unfortunately, we're not listening to the screen, and what they're showing us is only a slice out of the middle of the entire conversion process.

I hope this clears this up for you. Please feel free to ask any questions if part of this is still bothering you. I've never tried to type this out before, and am not quite sure how it will read.

Thanx again for asking!
Nika.



This message has been edited by Nika on 04-06-2001 at 09:53 PM
_________________________
For more...

Top
#441396 - 04/07/01 01:03 AM Re: George, Watch this!!!....(96k)
spookmuzik Offline
Senior Member

Registered: 09/13/00
Posts: 190
Loc: Seattle,WA,UNITED STATES
Ok - maybe there is something else I need to know in order to understand. If the converters work according to the mathematical certainty of a sine wave, does this mean that a sampled moment of music is broken down into sine waves - given that any complex waveform is a sum of component waves (is that part of Huygens principle?).

In other words - how does the number (1011101001000110 for example - too lazy to go past 16bit)represent a musical moment -

as A: part of a non-sine complex waveform which this number in combination with those which precede and follow it provide a "picture" of the sound we want to hear

or B: a large number of sine waves - at different frequencies, which added together represent the complex waveform we want to hear

???
Or something else altogether?

Julian M
_________________________
Keepin it Reel to Reel

http://www.dusty45s.com/

Top
#441397 - 04/07/01 01:20 AM Re: George, Watch this!!!....(96k)
Nika Offline
MP Hall of Fame Member

Registered: 03/22/01
Posts: 2938
Loc: Ft Wayne,IN,UNITED STATES
Julian,

A is correct.

This gets very difficult to describe without the aide of visual diagrams, but let me try.

If we take a complex waveform whose fundamental frequency is 20k, and then add a filter to it to eliminate all frequencies higher than 20k, you will be left with a perfect sine wave at 20k. If you then sample that sine wave at any two evenly spaced points you will have two digital samples, neither of which will be at the peak amplitude of the wave. Based on the fact that D/A converters act on the mathematical principal mentioned above, when you reconvert it back to analog you will end up with a perfect sine wave again.

Because the complex waveform was really just a 20k sine wave with a bunch of harmonic overtones over the top of it that we couldn't hear anyway, our ears only heard a perfect sine wave at 20k both before the conversion process and after the conversion process. For this reason, it was a 100% accurate conversion of all material at or below the Nyquist frequency. For the sake of human ears there was absolutely no loss.

Since this analogy is actually reality in the world of digital audio, we can thus prove that A/D D/A conversion processes will perfectly sample and reproduce 20k waveforms. Since this is true, does it not also make sense that any frequencies LESS than 20k will ALSO be perfectly sampled and reproduced?

That is the essence of how the process I mentioned above allows for 48k to be perfectly adequate to reproduce all frequencies in the human hearing range, unequivocally.

Make sense? Shall I attack from another angle?

Nika.

This message has been edited by Nika on 04-06-2001 at 10:24 PM
_________________________
For more...

Top
#441398 - 04/07/01 01:29 AM Re: George, Watch this!!!....(96k)
spookmuzik Offline
Senior Member

Registered: 09/13/00
Posts: 190
Loc: Seattle,WA,UNITED STATES
Nika - thanks for the knowledge!

Okay - I understand how a waveform butted up against the edge of our hearing *might* not require its "unhearable" overtones. However, if we are listening to an A from a piano, and then an A from a guitar in the same octave, the only reason they sound different is the different ratios of overtones the two instruments produce. These overtones are within hearing range for sure. If the A is sampled only as a sin wave we lose timbre no?

So the original argument returns - a sine wave is not my sound of choice!

Julian M
_________________________
Keepin it Reel to Reel

http://www.dusty45s.com/

Top
#441399 - 04/07/01 01:40 AM Re: George, Watch this!!!....(96k)
Nika Offline
MP Hall of Fame Member

Registered: 03/22/01
Posts: 2938
Loc: Ft Wayne,IN,UNITED STATES
Julian,

Yes, this is true, but each of the peaks and valleys of a complex waveform are sinusoidal, and thus the same concept applies.

If you sample an A on a piano at 440Hz, you will be actually sampling a 440Hz fundamental with an 880Hz overtone, 1760Hz overtone and several others combined with it to create a complex waveform. Some of these frequencies will be over 20k. Your ear will not hear these, and the filter will eliminate them from the sampling. The end result is a series of sampling points that in total represent that complex waveform. Any time the direction of the wave changes direction, the way in which it does that must be sinusoidal, and must be under 20k. No waveform under 20k is going to "change direction" in a way that results in a sinewave that has properties of being higher than 20k. The D/A will "redraw" the waveform through these points exactly as it was sampled in.

It's a tough pill to swallow. I know. I spat it out several times myself. Are we getting there?


Nika.

This message has been edited by Nika on 04-06-2001 at 10:43 PM
_________________________
For more...

Top
#441400 - 04/07/01 01:49 AM Re: George, Watch this!!!....(96k)
spookmuzik Offline
Senior Member

Registered: 09/13/00
Posts: 190
Loc: Seattle,WA,UNITED STATES
Nika,

A tough pill indeed! The part I'm not buying is that any change of direction *must* be sinusoidal. "Square" waves are the basis of digital signals in the first place, and the sawtooth pattern is something we have used in various ways - Synth sounds for example. No?

Julian M
_________________________
Keepin it Reel to Reel

http://www.dusty45s.com/

Top
#441401 - 04/07/01 02:18 AM Re: George, Watch this!!!....(96k)
Nika Offline
MP Hall of Fame Member

Registered: 03/22/01
Posts: 2938
Loc: Ft Wayne,IN,UNITED STATES
Yes, this is true, we do use square waves and triangle waves in many things.

The reality, though, is that these are only complex waveforms that have harmonic structures such that the addition of all of the sine waves creates a wave that is so complex as to be a square. If you filter everything above 20k from that square wave, however, you will end up with less than a true square. The closer your fundamental is to the 20k, the less your ear actually hears a square and it actually hears a sine wave. This is why the ear can't hear timbres above 10k - because it filters out all of the overtones and all it leaves is a sine wave.

So even though a synth may be able to create a square wave at 60hz, it is only a square wave to the point that the harmonics start to get above 20k for our ears' sake. For this reason, if we limit the frequency to 20k, the square wave can't be perfectly square and the triangle wave can't have a perfect peak. The sides of the square will actually slope up to the top in a perfect sinusoidal fashion of a 20k waveform. Just the same, the peak on the triangle wave will actually be perfectly sinusoidal of a 20k waveform. For this reason, the higher the fundamental, the less of a triangle and the more of a sine wave it is. At 20k, you can't have a triangle waveform. You can only have a sine wave. The same is true with the square waves. The same is true with any complex waveform.

For this reason it remains true that any waveform that has been filtered over 20k can only change directions in a sinusoidal manner.



Try it. Draw on paper a 10k complex waveform - one cycle. Don't try a square wave, it is too complex to do this on. This means that you'll cross the zero plane in equal increments downward vs. upward. Now go through and redraw the wave filtering out any part of that wave that is going to excede 20k. What you will end up with is a much less complex waveform, and everything you have there will be sinusoidal. Then draw your sample points - two while it is above the zero crossing, and two while it is below. Now take a new sheet of paper and trace those four points. Try to draw a sinusoidal waveform through those four points in any way except what you originally sampled? If you think you can do it, you didn't apply a good enough "filter" to your redrawing. You must "filter" it tighter than you did.

I'm not sure if that whole part made sense or not. The truth is that that does work, and that is indeed how it's done. Here, have another tablet......*grin*


Nika.
_________________________
For more...

Top
#441402 - 04/07/01 02:25 AM Re: George, Watch this!!!....(96k)
Nika Offline
MP Hall of Fame Member

Registered: 03/22/01
Posts: 2938
Loc: Ft Wayne,IN,UNITED STATES
Let me actually add to our little test.

First, draw a 20k waveform on a transulucent piece of paper that is proportional to the 10k complex waveform drawing that you're about to do.

Then, when you do your "filtering" stage, make sure that no slope on your paper is steeper than your 20k drawing above, and also that no peak angle is sharper or more triangular than your drawing above. If so, what you have is not truly 20k filtered yet. In order to be more triangular than 20k you must have additional frequencies in there. You need to make sure that no peak is sharper than your 20k drawing in order to insure that your not disguising higher harmonic overtones in your waveform drawing.

Then do your sampling and such.

Then, when you are reconstructing your waveform at the end from the sampled points, make sure that you're not trying to put any peaks in that are steeper than your 20k peak. Once again, if you are you're disguising harmonic structures that are higher than 20k.

I think you'll see what I mean.


I'm going to go watch a movie now. I'll try to post one more time before I hit the sack.


Thanks!
Nika.

This message has been edited by Nika on 04-06-2001 at 11:30 PM
_________________________
For more...

Top
#441403 - 04/07/01 02:28 AM Re: George, Watch this!!!....(96k)
spookmuzik Offline
Senior Member

Registered: 09/13/00
Posts: 190
Loc: Seattle,WA,UNITED STATES
"Draw on paper a 10k complex waveform - one cycle. Don't try a square wave, it is too complex to do this on. This means that you'll cross the zero plane in equal increments downward vs. upward. Now go through and redraw the wave filtering out any part of that wave that is going to excede 20k."

OK Nika, you lost me here. If I am drawing a waveform on a piece of paper, the x-axis is time, and the y-axis is amplitude right? The frequency is 10kHz *or* it is not. There is no 20kHz part to a 10k wave, no matter how complex it is. Which part of the wave am I "filtering"? There is no representation in my waveform diagram which represents frequency.

Am i using the wrong picture?

Thanks,
Julian M
_________________________
Keepin it Reel to Reel

http://www.dusty45s.com/

Top
#441404 - 04/07/01 02:31 AM Re: George, Watch this!!!....(96k)
spookmuzik Offline
Senior Member

Registered: 09/13/00
Posts: 190
Loc: Seattle,WA,UNITED STATES
Ahhhh! The slope of the waveform represents frequency? No hurry to answer, I'll be back another time too.
Julian M
_________________________
Keepin it Reel to Reel

http://www.dusty45s.com/

Top
#441405 - 04/07/01 02:35 AM Re: George, Watch this!!!....(96k)
Nika Offline
MP Hall of Fame Member

Registered: 03/22/01
Posts: 2938
Loc: Ft Wayne,IN,UNITED STATES
Frequency of the fundamental (10k) will be established by how often you cross the zero plane.

And there IS 20k material in there as part of your complex waveform. Even though it will only cross the zero plane twice in this picture, there will be lots of little rises and falls in various amounts while it is above and below the zero plane. Just because these rises and falls don't cross the zero plane doesn't mean they don't represent frequencies. Those are exactly those frequencies that you're arguing you'll be able to capture!!

Make sense?

In other words, if you have a 10k complex waveform, but while the wave is above the zero plane it happens to rise and fall five times before the wave dives below the zero plane, what you have is a 10k waveform with a 50k harmonic overtone. That information will obviously be filtered out in the next part of your test.

If you have a 10k waveform which rises and falls one additional time before it dives below the zero plane, but each of these little rises and falls is a triangle, then you have a 10k waveform with a 20k harmonic and a whole bunch of additional odd order harmonics over that. (A triangle waveform is all odd order harmonics. A square waveform is all even order harmonics). Thus you would have 10k, with 20k, 60k, 100k, 140k, etc. Since you would filter everything out except for the 10 and the 20, you'd be left with:

A 10k waveform that rose and fell one additional time before crossing the zero plane, but each rise and fall was a perfect sine wave.


And to answer your question of "does the slope determine frequency?" the answer is "sort of", but only when combined with the shape of the peak. This is a very bad scientific or mathematical answer, but for the sake of this explanation it is fine. Frequency is really defined by how many rises and falls you have. Overtones CAN be established by the slope and the shape of the wave.

Nika.

P.S. I have not started my movie yet.

This message has been edited by Nika on 04-06-2001 at 11:50 PM
_________________________
For more...

Top
#441406 - 04/07/01 03:20 AM Re: George, Watch this!!!....(96k)
spookmuzik Offline
Senior Member

Registered: 09/13/00
Posts: 190
Loc: Seattle,WA,UNITED STATES
Nika,

Thanks again, that makes sense now. The overtones are represented in the drawing by the "wiggles" Any wigglin goin on which was faster than the sample rate would be filtered out in keeping with Nyquist. At the same time as the A/D is prevented from seeing something which isn't there (ie the 34k showing up as 16k stuff) it also ... is prevented from missing any frequency information that would be heard anyway.

So ... If I believe that i can't hear anything above 20kHz (save that for later) then 44.1 should be enough. Except for that slope on the filter which unfortunately starts rolling off highs quite a bit lower than 20kHz.

Hence 48k sample rate which I assume is intended to allow the filter to begin its slope further up - preferably outside the range of hearing?

Perhaps I have been persuaded! Leaving aside the argument that human hearing may be being underrated, I could accept that 48k samples per second is enough to get what would seem to be all the information.

Next Problem Though....

Doesn't the information we "can't" hear affect the information that we can hear? For example - the way we EQ stuff is we use phase shift and addition to boost or cancel a target frequency. Well aren't the 50k waves affecting phase relationships between lower frequency waves until we SLICE!? Leading to a different sound coming out than the one that went in?


Julian M
_________________________
Keepin it Reel to Reel

http://www.dusty45s.com/

Top
#441407 - 04/07/01 03:41 AM Re: George, Watch this!!!....(96k)
Nika Offline
MP Hall of Fame Member

Registered: 03/22/01
Posts: 2938
Loc: Ft Wayne,IN,UNITED STATES
Julian,

My wife decided to postpone the movie until tomorrow morning.

The bottom line with the point above is that, depending on the quality and design of the filter, 44.1 or 48k are perfectly adequate to COMPLETELY represent any signals under 20k. Thus, for the sake of what is commonly accepted as our ears' hearing ability (from 20Hz to 20kHz) 96kHz recording is totally unnecessary. For further information on the theories of the potential validity of 96kHz, see the topic that I referred to above in my original post. There are indeed some theories that are worth exploring, only one of which is the "psychoacoustic" theory that we can percieve information that our ears are not attributed to being able to hear. I must tell you that this is, I believe, the weakest of all of the theories.

As for the notion that things that we can't hear can affect the things that we can hear, the answer is "no". Defiantly "no". Your ear acts as the same type of filter that we discussed above. If we take a 1kHz sine wave and then add all kinds of processing to it of very high frequencies (50k and such) so that in the end it doesn't really look like a 1kHz sine wave at all, and then put a filter on it that filters out everything over 2kHz, all you'll be left with is your 1kHz sine wave. I don't care how much junk you added. Once you add that 2kHz filter, it's right back to a 1kHz sine wave.

The thing is that your ears work this way also. If you take a clarinet note and add all kinds of garbage at ultra high frequencies to it for the sake of who-knows-what and whatnot, by the time you listen to it, all you'll hear is the clarinet.

If, however, you take the same clarinet and add some eq to it at 2.5k which also induces some wacky 50kHz stuff to happen, it would be incorrect to say that the 50kHz signal is CREATING differences that you can hear. What would be more correct is to say that you are processing the signal at 2.5k and some side effects at 50k, but that your ear won't hear the results of what happened at 50k. All that you'll end up hearing is the clarinet and the change of it at 2.5k. The 50k didn't CAUSE the change. It is a BIPRODUCT of the change, and an inaudible one that will be filtered away.

Fair enough? Have a glass of water with that pill you swallowed......*grin*
_________________________
For more...

Top
#441408 - 04/07/01 04:15 AM Re: George, Watch this!!!....(96k)
spookmuzik Offline
Senior Member

Registered: 09/13/00
Posts: 190
Loc: Seattle,WA,UNITED STATES
Quote:
Originally posted by Nika:
If, however, you take the same clarinet and add some eq to it at 2.5k which also induces some wacky 50kHz stuff to happen, it would be incorrect to say that the 50kHz signal is CREATING differences that you can hear. What would be more correct is to say that you are processing the signal at 2.5k and some side effects at 50k, but that your ear won't hear the results of what happened at 50k. All that you'll end up hearing is the clarinet and the change of it at 2.5k. The 50k didn't CAUSE the change. It is a BIPRODUCT of the change, and an inaudible one that will be filtered away.

Fair enough? Have a glass of water with that pill you swallowed......*grin*[/B]



Arrrrgh - horse pills!
What I am suggesting is that if (as you say above) a change at 2.5k can cause a change (albeit inaudible) at 50k, why cannot the 50k change what we hear at 2.5k. The idea is not adding a whole bunch of processing and then applying the filter - rather it is that the filter itself, in removing high freq info (before conversion)cannot then in D/A return the effect of any high frequency stuff on lower stuff.

Julian M
_________________________
Keepin it Reel to Reel

http://www.dusty45s.com/

Top
#441409 - 04/07/01 04:16 AM Re: George, Watch this!!!....(96k)
alphajerk Offline
MP Hall of Fame Member

Registered: 03/06/00
Posts: 7950
Loc: asheville nc usa
HA! the sampling rate [X axis] creates a finer resolution for the bit depth [Y axis] to capture not necessarily adding HF content but giving us a finer resolution picture over the ENTIRE frequency spectrum. it is only limiting the importance of this to HF content that is now able to be captured!

THE LIGHT GOES ON IN MY HEAD.

now i see why 96khz could be worth the investment even though the speakers we listen to anti-alias the audio anyways at 48khz. i can see how processing would be better as well.

like trying to blow up a 24 bit 72dpi picture wont look as good as a 24 bit 1200dpi picture. the 72 dpi will look fine from a distance but the closer you get the more pixilation you will see. the 1200 dpi picture will be very detailed.
_________________________
alphajerk
FATcompilation
"if god is truly just, i tremble for the fate of my country" -thomas jefferson

Top
#441410 - 04/07/01 04:17 AM Re: George, Watch this!!!....(96k)
Emile Offline
Gold Member

Registered: 08/25/00
Posts: 589
Loc: Montreal,,CANADA
Wow Nika what great info you've just presented. Very concice and greatly explain I must add. Although I'll make have opinion about the worthyness of 96KHz when I ear it, I do understand the theorytical point of view you're giving. Thanks a lot!

You might want to take a look at my D/A calibration post, I'm sure you could bring some light on the subject...

Emile

Top
#441411 - 04/07/01 04:29 AM Re: George, Watch this!!!....(96k)
Nika Offline
MP Hall of Fame Member

Registered: 03/22/01
Posts: 2938
Loc: Ft Wayne,IN,UNITED STATES
Quote:
Originally posted by spookmuzik:

Arrrrgh - horse pills!
What I am suggesting is that if (as you say above) a change at 2.5k can cause a change (albeit inaudible) at 50k, why cannot the 50k change what we hear at 2.5k. The idea is not adding a whole bunch of processing and then applying the filter - rather it is that the filter itself, in removing high freq info (before conversion)cannot then in D/A return the effect of any high frequency stuff on lower stuff.

Julian M


Julian,

Perhaps it's too late and I'm tired, but I just can't think of another way to answer this than "no". Any change that you make at 50k will not affect anything in our hearing range.

Except.......

For a phenomena having to do with subtractive overtones: an issue I do discuss in the other topic I mentioned that relates to 96kHz sampling. The concept of subtractive overtones playing a part in our listening CAN have something to do with a benefit of ultra high sampling. Read the blurb in the other forum and then let me know. I've yet to hear this theory actually demonstrated to be true, but mathematically it has potential - a lot more than the idea that we increase our resolution at 15k if we increase our sampling rate.

The bottom line is that (and remember here that I've done a LOT of research into 96k, and you'll see evidence of the listening tests I've done) I still can't hear it.

Talk to you soon,
Nika.
_________________________
For more...

Top
#441412 - 04/07/01 04:33 AM Re: George, Watch this!!!....(96k)
Nika Offline
MP Hall of Fame Member

Registered: 03/22/01
Posts: 2938
Loc: Ft Wayne,IN,UNITED STATES
Quote:
Originally posted by alphajerk:
HA! the sampling rate [X axis] creates a finer resolution for the bit depth [Y axis] to capture not necessarily adding HF content but giving us a finer resolution picture over the ENTIRE frequency spectrum. it is only limiting the importance of this to HF content that is now able to be captured!

THE LIGHT GOES ON IN MY HEAD.

now i see why 96khz could be worth the investment even though the speakers we listen to anti-alias the audio anyways at 48khz. i can see how processing would be better as well.

like trying to blow up a 24 bit 72dpi picture wont look as good as a 24 bit 1200dpi picture. the 72 dpi will look fine from a distance but the closer you get the more pixilation you will see. the 1200 dpi picture will be very detailed.


Alpha,

Umm. Uhh. Err. Actually, no. This is entirely NOT the point. The point is that, due to some wacky mathematical principles and some research done by Nyquist, if you blew the picture up it WOULD look as good if it were audio.

Again, 96k has absolutely NO advantages in better reproduction of frequencies that are in our hearing spectrum. If you came away thinking the opposite, then try reading this whole topic in a mirror or something, because you took the point absolutely 180 degrees out of phase.

Kindly,
Nika.
_________________________
For more...

Top
#441413 - 04/07/01 04:37 AM Re: George, Watch this!!!....(96k)
Nika Offline
MP Hall of Fame Member

Registered: 03/22/01
Posts: 2938
Loc: Ft Wayne,IN,UNITED STATES
Quote:
Originally posted by Emile:
Although I'll make have opinion about the worthyness of 96KHz when I ear it...........



Emile,

There's a problem. You won't hear it!! *smiling* But I would highly recommend the reading I've recommended in the first post of this topic. I go much more in depth about what you MIGHT be able to hear rather than what you CAN'T hear. Let me know your thoughts on it.


Thanx!
Nika.
_________________________
For more...

Top
#441414 - 04/07/01 04:53 AM Re: George, Watch this!!!....(96k)
Emile Offline
Gold Member

Registered: 08/25/00
Posts: 589
Loc: Montreal,,CANADA
Maybe I can supplement Nika's view with the screen analogy.

Say you had a straight line of a certain size, with only two dots you can reproduce it because you can fill (join) between the two dots without having to sample each dot constituting the line. Because of the pure symetry of the sine wave and its inherent caracteristic two dots at any given point of cycle of as sine wave can only be in that distance from both the x and y axis at that given frequency. So by knowing where those two dots occur and knowing they are part of a sine wave, you can automatically figure out its frequency and shape thus reproduce it through algorythms.

I have no scientific base for what I've just stated, so don't take my word for it. It's just what I came up with from reading the post. Back me up Nika if I've made sense 'cause this sure is a fascinating subject.

Good night everyone!
(it's 1:57 am here in Montreal)

Emile

Top
#441415 - 04/07/01 05:03 AM Re: George, Watch this!!!....(96k)
Nika Offline
MP Hall of Fame Member

Registered: 03/22/01
Posts: 2938
Loc: Ft Wayne,IN,UNITED STATES
Emile,

This is absolutely correct. Assuming that an anti aliasing filter is in place, any two sampled points can accurately represent all of the space between them, just as:

----------------------

can be represented by just

...+............+....

if we know that we are dealing with only a straight line, a sine wave can also be represented by

...+.................
................+....

if we just know some basic things about the sine wave including what the anti aliasing filter is set at and what the Nyquist frequency is.

And that silly little mathematical principle, my friends, is the foundation for part of the great marketing hoax of 96kHz.

Nika.

This message has been edited by Nika on 04-07-2001 at 02:09 AM
_________________________
For more...

Top
#441416 - 04/07/01 05:11 AM Re: George, Watch this!!!....(96k)
spookmuzik Offline
Senior Member

Registered: 09/13/00
Posts: 190
Loc: Seattle,WA,UNITED STATES
Alrighty!

I just spent a while rereading the other thread, and I have to say I'm with you. I have never had the opportunity to do the kind of listening test you have done. The only way I could tell the difference between 96k and 44.1 (both at 24 bit)was running tone (generator went up to 24k)through a Radar 24 and hearing the drop off at 16k up to somewhere near 20k. This obviously is about the filter, and not the sample rate. At a frequency within our hearing range the level dropped to nothing (and the waveform stops being drawn on screen, while the VU meters on the old tape deck just keep on showing signal as you sweep up.

I'm on the west coast - so its just now 11pm for me. So I'll keep yapping away. Is there a discussion of bit depth with this kind of clarity?
How many bits is enough? Can they sell us 48bit stuff next year?

Julian M
_________________________
Keepin it Reel to Reel

http://www.dusty45s.com/

Top
#441417 - 04/07/01 05:22 AM Re: George, Watch this!!!....(96k)
Nika Offline
MP Hall of Fame Member

Registered: 03/22/01
Posts: 2938
Loc: Ft Wayne,IN,UNITED STATES
Is there a discussion of bit depth with this kind of clarity?
How many bits is enough? Can they sell us 48bit stuff next year?

Julian M


Nope. All that bit-depth gains us is a wider dynamic range. When you add an extra bit you add 6db of potential dynamic range, but no additional resolution within that range. So the only practical means for adding much to the bit depth of your recordings is if you want to record some crickets out on a runway and have planes fly over your head and not clip your inputs.

24 bit gives you the potential to record the most dynamic music made. You can record music that is as quiet as 7db (there isn't a microphone that has a noise floor lower than 6db) in an anechoic chamber all the way up to the 1812 overture with canons firing. We can now fully capture the entire dynamic range that the human ear can hear.

Nika.



This message has been edited by Nika on 04-07-2001 at 02:24 AM
_________________________
For more...

Top
#441418 - 04/07/01 05:30 AM Re: George, Watch this!!!....(96k)
spookmuzik Offline
Senior Member

Registered: 09/13/00
Posts: 190
Loc: Seattle,WA,UNITED STATES
Nika,

You rock man! I mean thats just some serious 'splainin! And the best part is you've helped me reaffirm my belief that I can focus my budget on front end stuff, improving my monitoring situation, and working on the room itself.

.... and maybe someday I'll have a car (buses just quit going to my house after 12:30AM)

Julian M
_________________________
Keepin it Reel to Reel

http://www.dusty45s.com/

Top
Page 1 of 42 1 2 3 ... 41 42 >