Music Player Network Home Guitar Player Magazine Keyboard Magazine Bass Player Magazine EQ Magazine
Page 3 of 42 < 1 2 3 4 5 ... 41 42 >
Topic Options
#441449 - 04/08/01 03:24 AM Re: George, Watch this!!!....(96k)
Curve Dominant Offline
MP Hall of Fame Member

Registered: 10/29/00
Posts: 4223
Loc: Philadelphia USA
originally posted by 1176:
>>Probably cause he fell asleep from all the dribble being spouted out by Nika. Even HTML starts looking interesting. Maybe we could get Nika his own forum for his birthday or something. I bail pretty quick when his posts start filling up. I wonder why a guy who knows so much is schelping gear at Sweetwater...Nothing personal, just my .0005 cents <<

originally posted by Roger Nichols:
>>Good work Nika...
On the Steely Dan "Two Against Nature" album, we mixed to 44.1k and 96k at the same time. Nobody could tell on playback which was which... NOBODY.
Roger <<


I'll take Roger's assessment of Nika's credibility over some anonymous quack any day. Just MY .0005 cents. I'm preparing to help produce soundtracks for a non-profit film production house that educates inner-city at-risk youth, and I appreciate being provided with knowledge that will prevent these folks from investing big $$$ in mundane technology. REALITY CHECK: This forum is not a chat room.

curvedominant

This message has been edited by Curve Dominant on 04-08-2001 at 12:29 AM
_________________________
Eric Vincent (ASCAP)
http://www.curvedominant.com

Top
#441450 - 04/08/01 03:57 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
For the sake of 1176 and his comments I'll attempt to be concise.

I have researched 96k and done listening tests. I have yet to hear a difference. I was, therefore, amused and intrigued when George made his comment about recording 96k on a laptop. I was sincerely curious why, so I asked.

Then a couple of people posted about a misconception of the benefits of 96k. While answering those questions I needed to go much more in depth than this type of forum typical provides. I did want to straighten that out for the sake of Julian who was sincerely curious about this.

As for Alpha - read the material. I know exactly what your point is. It gets old saying this. I know how big the box is. Please just read the material and then call me and I can explain on Monday. It should go very quickly.

As for George's lack of commentary - this may sound strange but I had FOUR people call me AT HOME today to comment on this thread and each one of them said the exact same thing - "I wonder why George didn't commment". "It's unfortunate that George didn't have anything to add". Etc. I have to tell you that I thought that it was pretty bizarre to garner that much attention for this topic or any topic in this forum. I assume George has his reasons. At this point I'm a little more curious knowing what those reasons are than why he wants to record 96k on a laptop. Oh well, in due time. I'm really not trying to play devil's advocate with him, I sincerely am curious why he wants this capability. Maybe I'm missing something.

As for Orgasmatron, why do I suspect that that may not actually be his first post? After all, if I had something like that to say, I probably wouldn't use my regular handle. Then again, maybe not.

1176, thank you for your comments, both positive and negative. I'm just trying to help where I can and learn where I can. I apologize if my means of doing that is tedious and boring for you.
_________________________
For more...

Top
#441451 - 04/08/01 04:04 AM Re: George, Watch this!!!....(96k)
Curve Dominant Offline
MP Hall of Fame Member

Registered: 10/29/00
Posts: 4223
Loc: Philadelphia USA
Alphajerk:

I'm relieved to know that I'm not the only dawg who's been kept at home on a Saturday night for the cause of research and development. I have spent this entire evening combing over both this thread, and the "Digital Wanking..." thread on Roger's board, and I'm starting to feel a little guilty that I don't have to pay for this quality of education. That having been said...

Anti-aliasing in computer graphics is a counter-intuitive use of blurring. Originally called Fuzzy Fonts, this technique, which makes text more readable on computer screens, was pioneered in 1972 at the Architectural Machine Group (a predecessor of the Media Lab at MIT). It replaces the hard edges of 2-color pixelization with gradated-degrees of the opposing tones, thereby creating the illusion of smoothness. Does it have a parallell with anti-aliasing filters in audio convertors? Nika and Roger could probably give you a better answer to that question than I could, but my best guess is: no. I have the impression that we are talking apples and oranges. But since George has been conspicuously silent on this topic, it seems logical to pass this off to Roger and get his input.

Enjoy the chronic. I'm temporarily abstaining, but the bar is stocked at the CurveDominant facilities, so all is not lost.

curvedominant
_________________________
Eric Vincent (ASCAP)
http://www.curvedominant.com

Top
#441452 - 04/08/01 04:15 AM Re: George, Watch this!!!....(96k)
Orgasmatron the mighty Offline
Member

Registered: 04/07/01
Posts: 7
Loc: like i would tell you,WV,UNITE...
Tweaks Alpha jerk behind the ear like a 5th grade teacher to the really annoying student*
YO! Not thinking outside the box? I personally kow how much time Nika has devoted to this subject and I know one thing : My boi Nikalito is trying to tell you the sky is blue and you are saying "hmm.. i don't think you are looking at the right angle.. maybe it's Puce"
Look up
The sky is blue.
And close your mouth, or you'll drown like a chicken. It makes you look dumb too.
Orgasmatron the Mighty

Top
#441453 - 04/08/01 04:15 AM Re: George, Watch this!!!....(96k)
Orgasmatron the mighty Offline
Member

Registered: 04/07/01
Posts: 7
Loc: like i would tell you,WV,UNITE...
Tweaks Alpha jerk behind the ear like a 5th grade teacher to the really annoying student*
YO! Not thinking outside the box? I personally kow how much time Nika has devoted to this subject and I know one thing : My boi Nikalito is trying to tell you the sky is blue and you are saying "hmm.. i don't think you are looking at the right angle.. maybe it's Puce"
Look up
The sky is blue.
And close your mouth, or you'll drown like a chicken. It makes you look dumb too.
Orgasmatron the Mighty

Top
#441454 - 04/08/01 04:19 AM Re: George, Watch this!!!....(96k)
alphajerk Offline
MP Hall of Fame Member

Registered: 03/06/00
Posts: 7950
Loc: asheville nc usa
i thought i recognized orgasmatron, just didnt want to spill the beans. i got caught hassling Alice Kooper a while back

how about i not read it over and just leave it at skimming it over. im quite familiar with anti-aliasing and plan on simply waiting until i hear 96khzto decide [which i would be willing to bet i stick at 48khz] and know enough about nyquist to not worry too much about it. despite the marketing push for 96khz, im in NO rush to use it.

anyone think PCM will be replaced with DSD? im kinda in the wait and see concerning digital stuff seeing how i get more than adequete results with my minimal digital investment... all my money goes towards mics and pres right now [unless i am having to replace whats blowing up ! ] its why i dont waste my money on PT and why i dont waste my money on high$$$ convertors right now. im even still mouse mixing until these damn idiot manfacturers put out a REAL hands on controller. still thinkin about a motor mix [the cheapest with motorized faders].

i've been estatic with my recordings of late anyways and the weaknesses that im working on correcting ive got a bead on and it has nothing to do with 96khz or digital or convertors. the source, mics and rooms are really my passion over anything else. not to mention the bands...
_________________________
alphajerk
FATcompilation
"if god is truly just, i tremble for the fate of my country" -thomas jefferson

Top
#441455 - 04/08/01 04:28 AM Re: George, Watch this!!!....(96k)
Curve Dominant Offline
MP Hall of Fame Member

Registered: 10/29/00
Posts: 4223
Loc: Philadelphia USA
Nika posts:
>>I'm really not trying to play devil's advocate<<

I'll play devil's advocate - in George's defense! Perhaps he's recording to a laptop in 96K because he can! Like, why one would climb Mount Everest: because it's there. I'm just trying to dispell the impression that this whole exercise is to pile on GM. C'mon, George, we're not the enemy, we're just talkin' here. Wahzzup?

Nika,

How does digital audio anti-aliasing "fill in the blanks?" Just curious.
_________________________
Eric Vincent (ASCAP)
http://www.curvedominant.com

Top
#441456 - 04/08/01 04:28 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
Alpha,

Are you still awake? Is there a number I can call you at? Or will I wake the babe (or the baby also)?

Let me know. Let's just step outside and settle this one like real men.


Nika.
_________________________
For more...

Top
#441457 - 04/08/01 04:31 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
BTW, I'm now thoroughly confused at who Orgasmatron might be. I'm not sure he even knows me.
_________________________
For more...

Top
#441458 - 04/08/01 04:36 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 Curve Dominant:
Nika posts:
[b]>>
Nika,

How does digital audio anti-aliasing "fill in the blanks?" Just curious.


Curve,

Great question. Unfortunately we're getting to a level that I'm not sure how well I can type the answer. Just email me your contact info and I'll call you next week. It's a pretty simple concept, but I thought that this last topic was a pretty simple concept and it's apparently taken 65 posts to still not answer it adequately. I'd rather just take it offline for now. Fair?

Thanx!
Nika.
_________________________
For more...

Top
#441459 - 04/08/01 05:12 AM Re: George, Watch this!!!....(96k)
Curve Dominant Offline
MP Hall of Fame Member

Registered: 10/29/00
Posts: 4223
Loc: Philadelphia USA
Nika,

More than fair, rather, I appreciate your offer to educate me. My apologies - I didn't concisely word that question - no worries. I suppose I could do the homework on the mathematical issues. I don't have a precise knowledge of digital audio conversion anti-aliasing, but I feel as if it can be understood on an intuitive level. I'm a composer, so it's not as if my ass is on the line or anything, but I find myself irrationally fascinated with this subject.

By the way, I recieved the SW print collateral - tante grazie! I luv the 2001 ProGear Directory, and I've been carrying the SweetNotes issues around as reading material while on the subway, etc. I'm going to call you folks before I buy anything new from now on.

curvedominant
_________________________
Eric Vincent (ASCAP)
http://www.curvedominant.com

Top
#441460 - 04/08/01 05:40 AM Re: George, Watch this!!!....(96k)
Lee Flier Offline
10k Club

Registered: 09/13/00
Posts: 15398
Loc: Atlanta,GA,UNITED STATES
Alpha, Curve, here's the thing that I think you guys are missing: yes, anti-aliasing is totally different in audio than it is in graphics, and the two are totally different in the way they are represented digitally. This should be obvious when you consider the astronomical differences between the size and frequency of sound waves vs. that of visible light waves. We have nothing even remotely approaching a vehicle that can travel at the speed of light, but there are already planes and spaceships that are faster than the speed of sound.

So, representing visual data (light) digitally HAS to rely heavily on "interpretation" rather than actual real-time representation. Whereas sound waves are big and slow enough to represent the whole thing, given a reasonably powerful computer like the ones we have on our desks these days. In other words, given the audible frequency spectrum of 20Hz-20KHz there are only so many places that a pair of sample points can fall on the graph, no matter how complex the waveform is, and that number is large but not at all unreasonably so.

So... if we filter out everything above 20K, and even if we sample a complex wave only once for each time it crosses zero (which is covered by 44.1K), the angle that results from bisecting the two points can ONLY, mathematically, equate to one waveform. There is no need to "interpret" what that waveform "might possibly" be as there is with graphics; it's a mathematical certainty what it will be. If it were a different waveform, even a little bit, the angle between the two points would be different. So all the computer has to do is reproduce the waveform which it knows it is. With graphics, the computer does indeed have to "fudge" a lot and guess where to fill in the blanks, because there is no way to ever have enough data points to analagously represent light. Comprende?

Of course, filters are not perfect and that's one reason why 48K can sound better than 44.1K. And there are other possible benefits to 96K that I won't get into right now because I'm about to fall asleep. But, for now let's just say that 44.1 is enough to TOTALLY represent any audible waveform no matter how complex, assuming you are filtering out any harmonics above 20K (and who cares if you do, even if we can "sense" shit above 20K it's pretty much a guarantee that something either in the recording or playback chain is going to be unable to record or reproduce those frequencies anyhow).

--Lee
_________________________
What The...?
http://www.what-the.com
http://www.myspace.com/whattherock

Radio KIRA
http://www.radiokira.com


NEW CD out April 7th! >> Get it here , or here ! <<

Top
#441461 - 04/08/01 05:44 AM Re: George, Watch this!!!....(96k)
alphajerk Offline
MP Hall of Fame Member

Registered: 03/06/00
Posts: 7950
Loc: asheville nc usa
sorry, i was spacing the past hour. i sent you an email but i realized what time it is/was.

i dont feel the need to settle it either since i dont much care about the 96khz attainment like curve said, i know im never going to bother climbing mt everest but i climb all sorts of mountains in my backyard, some with fantastic views [i can see 4 states at once] im sitting right now over half a mile high [higher than that really ] but there are certainly some who will... and some will die trying to climb mt everest. 96khz wont kill me.

hey george, ill show you how to post pics on this forum if you show me how to use an Oxford
_________________________
alphajerk
FATcompilation
"if god is truly just, i tremble for the fate of my country" -thomas jefferson

Top
#441462 - 04/08/01 05:52 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
Lee,

Thank you for chiming in. I did cover many of the other theories regarding benefits of 96k elsewhere, but I'd love to have your opinion.

Thank you also for clearing up the graphics issues. I know absolutely nothing about graphics as you are about to see!

In the meantime, I think I have put together a visual aide. See below:



You'll notice in the top picture a very complex waveform.
The fundamental to this waveform is at 11.2k.
The first harmonic is right at about 23k.
There are several other harmonics. I ran it through a scope.

The bottom picture is what you actually hear. What I did was apply a filter to it at 20k.
Notice that this is no longer a "complex" waveform?
This is also exactly what the anti-aliasing filters do on the A/D converter.
Thus, by the time it actually hits the converter chip, it looks exactly like this.
Moving on to the converter chip, now:



You'll notice that in the bottom picture now I have drawn in the samples that the converter will actually take.
This is for sampling that is done at 44.1k

Now, what we are left with is this:



The mathematical principal states that there is only one way for the D/A converter to redraw a wave given these sampling points.
Any guesses as to what it will look like?
Any guesses as to what it will sound like?
Exactly, it will look and sound exactly like the bottom frame in the first picture - which is, as we established, exactly what you heard going in.

I hope this makes sense. If I can't explain this concept now with the visual aides I need to switch jobs!


Thanks again, Lee. It's nice to have your help.
Nika.


This message has been edited by Nika on 04-08-2001 at 03:01 AM
_________________________
For more...

Top
#441463 - 04/08/01 06:06 AM Re: George, Watch this!!!....(96k)
Lee Flier Offline
10k Club

Registered: 09/13/00
Posts: 15398
Loc: Atlanta,GA,UNITED STATES
Nika, you need to trim the white space out of your graphics so they don't go way out to the right.

The example is helpful, but I still think people won't get it because you only showed harmonics that are over 20K. In practice, usually when you filter out harmonics over 20K you are still going to be left with a complex waveform. What I think you need to show is how the "dips" from the harmonics will move the bisection angle of the samples. If you draw a line between the two sample points of a wave with audible harmonics, the line will fall at a different angle than if it were a pure sine wave, regardless of at what point the samples fall, as long is the sampling rate is constant.

--Lee
_________________________
What The...?
http://www.what-the.com
http://www.myspace.com/whattherock

Radio KIRA
http://www.radiokira.com


NEW CD out April 7th! >> Get it here , or here ! <<

Top
#441464 - 04/08/01 06:17 AM Re: George, Watch this!!!....(96k)
Curve Dominant Offline
MP Hall of Fame Member

Registered: 10/29/00
Posts: 4223
Loc: Philadelphia USA
originally posted by Curve Dominant:
>>Anti-aliasing in computer graphics is a counter-intuitive use of blurring... Does it have a parallell with anti-aliasing filters in audio convertors...my best guess is: no.<<

For the record, I was not the one who established the parallel to visual imaging, but rather disputed it. That having been said...

We are "bouncing the rubble" as my friends in the defence industry would put it. OK, I'm a simple guy who relies on simple math. But simple math establishes that 44.1K is adequate resolution for digital recording, or no?

curvedominant
_________________________
Eric Vincent (ASCAP)
http://www.curvedominant.com

Top
#441465 - 04/08/01 06:26 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 Lee Flier:
Nika, you need to trim the white space out of your graphics so they don't go way out to the right.


Lee,

Remember when I said I'm not a graphics person?? I tried for 45 minutes to figure out how to do that and failed. If someone wants to fix them for me I'll give them my email address and they can send them back to me. I'll repost them then.

Thanx, though. If I get another three hours I'll try to show what you are describing. It's hard to find the right material.
_________________________
For more...

Top
#441466 - 04/08/01 06:46 AM Re: George, Watch this!!!....(96k)
Curve Dominant Offline
MP Hall of Fame Member

Registered: 10/29/00
Posts: 4223
Loc: Philadelphia USA
Nika:

It totally makes sense, as it should have for anyone who was actually READING this thread, as opposed to "skimming" through it, or taking pieces of it out of context and reacting with glib comments.

>>What I think you need to show is how the "dips" from the harmonics will move the bisection angle of the samples. If you draw a line between the two sample points of a wave with audible harmonics, the line will fall at a different angle than if it were a pure sine wave, regardless of at what point the samples fall, as long is the sampling rate is constant.<<

I think (?) what Nika is illustrating is that "a pure sine wave" as an issue is a fallacy, or a redundacy, or a non-issue...I'm guessing here, so flame away. There is no "pure sine wave" in any ensemble recording, so how could that possibly be a factor? If A/D or D/A convertors only functioned on "pure sine waves," we would be in the dark ages still, or no? I may be a newbie, but I am not completely mentally retarded yet.

curvedominant
_________________________
Eric Vincent (ASCAP)
http://www.curvedominant.com

Top
#441467 - 04/08/01 06:59 AM Re: George, Watch this!!!....(96k)
alphajerk Offline
MP Hall of Fame Member

Registered: 03/06/00
Posts: 7950
Loc: asheville nc usa
what im saying is its still not the original wavform, the complex one with the 23khz harmonic. that or its too late for me to pay attention. i wonder if you compress both those waveforms and THEN filter the complex one if the results will be the same?

theoretically both visual and aural anti-aliasing are accomplishing similar tasks which is what i was refering to. "it simply cuts off the very high frequencies which it cannot handle. This cripples the sound a lot, but it is required to prevent even more serious damage to the sound, which would make the sound unrecognizable. This is a low-pass (cut the 'high' frequencies, let the 'low' frequencies pass through) anti-aliasing (smoothing, blurring) filter"-http://bundy.hibo.no/~thammer/HammerSound/audiobasics/audiobasics.html#DA_Digitizing they also claim "If you record at 96000Hz, you will have more than four samples for each 20000Hz period, so the chance of losing high frequencies will decrease dramatically"

and the speed of light has little to do with the limited frequencies of light we can see. remember, what we cant see visibly can give us some color too, and we filter that stuff out too because too much will give us cancer.... hmmmmm.

i would entertain capturing the complex form above the filtered wave. the added information could prove useful later on in the process.

again, its late and im listening to a bootleg tape of this incredible band in the 3rd generation embedded in hiss and im loving the music.... imagine that?
_________________________
alphajerk
FATcompilation
"if god is truly just, i tremble for the fate of my country" -thomas jefferson

Top
#441468 - 04/08/01 08:18 AM Re: George, Watch this!!!....(96k)
ChrisJ Offline
Senior Member

Registered: 02/10/01
Posts: 107
Loc: Anywhere,WY,UNITED STATES
Executive Summary:
One would do well to avoid making statements of absolutes in the field of physics and engineering.
--------

Perhaps some comments in this thread have been made in the heat of the moment. If so then I wish my comments to be taken as food for thought rather than a specfic response to a specific post. If on the other hand the comments were made in earnest then I would invite discussion of the points that I will raise below.

Perhaps we would do well consider the limitations of human knowledge before speaking in absolutes. Specifically, I wish to point out that in most cases there are assumptions that underlie what are commonly referred to as "laws" of physics and engineering. The laws are no more true or valid than the underlying (sometimes unexamined) assumptions.

> Nyquist himself was wrong?

Yes and No. The theorem is very useful. No doubt about it. However the theorem is not absolute. It is a mathematical *idealization* of a more complex reality. The theorem is true only to the extent that the idealization assumed corresponds to reality.

In the discipline of physics one uses mathematical "laws" or theories as a compact way of summarizing a consistent set of observations. The theory has only to "save the appearances." That is any theory (however different its mathematical formulation may appear) is to be considered equally valid with another if both are able to reproduce the observations with comparable accuracy. Only when theories embody differing predictions of observable phenomena may one objectively distinguish two theories and thus be able to decide if one is "better" than the other.

Having said that, I would suggest that part of the problem with discussing topics such as the one at hand is that there appears to be an ongoing need for further research and education in psychoacoustics. The community continues to engage in debate as to the usefulness of 96kHz sampling and 24 bit amplitude resolution. Arguments often are based on the assumption that human hearing is the canonical 20-20 kHz. This is an approximation that has limitations that are not well explored to the best of my knowledge. Some people may be able to hear higher frequencies. Many people certainly reach limitations at lower frequencies. There is some appreciation of this fact. However, the community would do well IMO to re-examine the empirical basis for this common assumption. Who did the measurements? Where are they published? Who has corroborated the measurements? Most importantly what are their limitations?

Another less often discussed issue is that the canonical "20-20 kHz" frequency response seems to be based upon the detection of a steady state sine wave. This implies a certain interval of integration (averaging time) which almost never is specified. Thus the results are rather meaningless. For example, some people claim to detect tones at levels lower than -96 dBFS for 16-bit digital systems. They may be telling the truth. Why? Because the commonly accepted "rule" that one gains 6.02 dB per bit of amplitude resolution implicitly defines the averaging interval as one sample time. Depending upon the type of noise present one may be able to average the noise down by averaging over many sample times. Some types of dithering make use of this principle. Thus it is possible to create a dithering system that allows detection of signals lower than -96 dBFS even though theory says we can only represent 16 bits * 6.02 dB/bit = 96 dB of "range". What is usually left unsaid is that 96 dB means 96 dB *per sample*. The underlying assumption is that the relevant averaging interval is one sample. In many cases it is but certainly not in all cases.

What would a meaningful averaging time be? I can't answer that in general but surely this is an area for further research in the field of psychoacoustics. Perhaps the Haas effect's ~30 msec is a more appropriate averaging time than one sample for some applications. My purpose here is not to give a definitive answer but rather to raise what I consider to be essential questions.

Returning to Nyquist, one would do well to remember that the Nyquist theorem (which states that one must sample at twice the rate of the band that one wishes to represent e.g. 40k samples/sec is needed to accurately represent a bandwidth of 20 kHz) *assumes* that one has an ideal band limited signal --that is to say a brickwall filter that goes from unity gain to zero gain in an arbitrarily short interval of frequency.

In the real world one cannot construct such a filter. In the analog realm even as one approaches this ideal the phase and delay effects become drastic near the filter edge. This may cause audible effects. Since the filter does not have a perfect stopband, signals above the nominal cutoff frequency are aliased back into the passband. The fact that filters may have some rolloff before the ideal cutoff fequency means that samples will no longer be independent but will rather be mathematically correlated.

So what is my point? My point is that Nyquist's theorem or any other physical law is no better than its underlying assumptions. Thus it behooves one to understand the a priori assumptions being made and to make an effort to verify that those assumptions hold true before one makes absolute statements.

> Newton has a major flaw in his works?

It depends on how one defines "major". Newtons laws are most helpful for applications in which:

1) velocities are a very small fraction of the speed of light
2) gravitational fields are weak
3) dimensions are large compared to atomic dimensions

One must realize that these are critical underlying assumptions. One would do well to remember that it took the scientific community centuries to appreciate the importance of these assumptions. So again, we would do well realize that "laws" should only be considered valid within the regimes of parameters space in which they have been empirically verified. Based on the early sucess of Newtons laws, too many people assumed that the universe was a deterministic entity. They failed to appreciate the limited range of conditions under which Newton's laws had been tested. In so doing they jumped to unjustified philosophical conclusions. Quantum mechanics certainly has been a sobering influnce in this matter, but more importantly even before the flowering of quantum mechanics people like Poincare served the discussion by pointing out that (even within the above listed restrictions on velocity, gravity, and dimension) any meaningful definition of "deterministic" would require an unrealizable detailed knowledge of the a priori state of the system under consideration. The success in recent decades of Chaos theory have come from building on the insights of Poincare and others.

If one desires to describes systems in which any of these three conditions are not met then yes, Newton's law have major flaws. This is not just an academic issue. Electrons in the circuitry we use travel at speeds which are of the same order of magnitude as the speed of light. The computers that we use to access this board are designed based upon quantum mechanical principles. So for those interested in electronics and computers, yes there are major, relevant limitations to the applicability of Newton's laws. This does not diminish in any way Newton's brilliance nor does it in any way diminish the magnitude of his contribution. Oft times the difficulties do not lie with the originator of the law but rather with those that press those laws into service in regimes beyond that which was intended.

> Einstein answered "I would have pity on the lord.
> The theory is correct."

I do not know where this is quoted from or if it is just a parable used to make a point. However, let us remember that Einstein for all his brilliance was still a man with limitations. In his early work on the theory of gravitation Einstein included a so called cosmological constant. Later in life he spoke of this as his "greatest mistake." The fortunes of this aspect of his theory have waxed and waned as further discoveries have shed light on the issue (e.g. Hubbles discovery of an expanding universe and more recent evidence for an accelerating universe). If one means by the use of the term "correct" that the theory accurately accounts for the presently available observations then one may say that a theory is "correct". It is beyond the competence of the scientific method to say that something is "correct" if one means by this that there have been, are, and can never be any conditions under which the law does not hold.

> The theory is correct."
Again, I believe it is more accurate to say that a *physical* theory is a valid approximation that holds within the limitations of the underlying assumptions and to the extent that it has been empirically verified.

> the theory is not just correct here. It is mathematical law.

I wish here to make explicit my assuption that we are interested in physical theories which will lend insight into music and the recording of music. Granted that assumption, I believe it is of critical importance to make a distinction between physical theories (or laws if you wish) and mathematical theorems. Mathematics by its nature guides us in drawing logically consistent conclusions from a *given* a priori set of assumptions. The discipline of mathematics in and of itself does not "get behind" the assumptions themselves. For example, for two millenia the world got along fine with Euclidean geometry which is based on 5 foundational assumptions. Around the 1800s the last of these assumptions, namely that two parallel lines never meet, was re-examined. Mathematicians deduced that if this assumption was not required then one could devise so called non-Euclidean geometries. Perhaps these geometries would still be largely unnoticed if it were not for the fact that they proved to be most useful dealing with theories of gravitation and the curvature of spacetime.
So to summarize, mathematical proofs are not a guarantee of the infallibility of a physical theory. Mathematical consistency is a necessary but not sufficient criteria for a fruitful physical theory.

> And as to 96k, be VERY prepared to be disappointed.

You are probably quite right.

The oft used aphorism that the system is "only as good as the weakest link" in the chain applies here. There is a need for more system level engineering.

"What is the limiting factor in the system?"
"If that factor is addressed what is the next limiting factor?"
and so on.

I have a rather modest setup at present. No mics over $500. Mixing board under $1000. etc. etc. Yet, I am able to hear the effect of our small room on the recorded sound. If I were to purchase a complete system with 24/96 throughout it still would not address my acoustical problems.

As I have discussed with you in previous threads, Nika, there is reason to suspect that the chief benefit of 96K sampling rates would be to improve the spatial imaging. Perhaps there might be some benefits in reduced aliasing.

However, from a systems point of view one must consider that the vast majority of mics in use today do not have usable sensitivity above 20 kHz. Earthworks and B&K measurements mics are an exception. Although such devices may exist, who has monitors that have accurate response above the mid 20 kHz (Tannoy super-tweeters?). Even assuming that these issues are dealt with one then has to re-examine acoustical issues. What kind of acoustics would be required to enable one to discern the presumably small improvements in spatial imaging? In parallel, one would need to re-examine the present practice of amplitude based panning. Placement on a real soundstage involves not only amplitude but also phase/delay.

Although I have not done an adequate systems level study, I suspect that before 96K sampling would be useful one would have to adresss the aforementioned three areas: transducers, acoustics, and mixing practices.

Thanks for the stimulating thread.

Peace,
Chris


This message has been edited by ChrisJ on 04-08-2001 at 06:58 AM

Top
#441469 - 04/08/01 09:11 AM Re: George, Watch this!!!....(96k)
Chip McDonald Offline
MP Hall of Fame Member

Registered: 01/19/00
Posts: 4792
Loc: Augusta, Ga. USA
I've admittedly skipped explanations, half-asleep and tired, but.... I'm probably missing something obvious, but:


The process reduced to mechanical terms is that a measurement of voltage is taken at a certain increment of time.

If one were to make a ultra-primitive converter - say 1 bit at a 1 second rate, what would happen if you tried to use that converter to represent an event that happened *approximately* every second?

Every second the "converter" would take it's measurement: on or off.

If the thing being sampled has the exact same clock as the converter, what the converter records will be accurate.

If the converter's clock is 180 degrees out of phase with the sample's periodicity, the sample disappears.

All examples of the "bit depth/rate" discussions I've seen are presented in a rather stark manner. The examples always assume a simple waveform. They also assume a periodicity that perfectly corresponds to the clock rate of the converter.

44.1 yields the Nyquist pass through of roughly 20k, assuming that 20k sine wave is in sync with the converter's clock.

Using my earlier converter, if you varied the periodicity of the sample, the converter would turn an on/off sample of a 1 second periodicity into
something like:


0 1 0 0 1 0 0 0 1 0 0 1 0 1 0

Instead of:

0 1 0 1 0 1 0 1

Which is essentially jitter: it doesn't have to be two *electronically* digital devices, but a timing displacement.

So I see two problems. An overtone that is a multiple of the Nyquist limit that has it's cycle start 180 degrees out of phase with the converter's clock, thereby "disappearing" by occuring when the converter is not sampling. Or, a situation where the frequency of the overtone is modulating and and crosses a multiple of the converter's clock. So in other words, it doesn't matter if the overtone happens beneath the Nyquist frequency, but if the timing of it coincides slightly out of sync with the clock rate, or if it modulated across the timing of the clock rate, wouldn't there be artifacts involved?

I would think this would only occur en masse in complex signals of a lot of high frequency content: cymbals, ambient sounds, distorted guitar sounds, and would be less noticable on other things.

I also think the model of the way the mind perceives sound is not fully defined. I think our perception of *musical* sound isn't an instanteous linear event but a parallel assessment of a memory "buffer". However impossibly fast it may seem, I think our *conscious* perception of sound occurs after an autonomic function occurs to gather the things more pertinent to instantaneous survival: phase relationships of sound to localize things. *Then* conscious frequency analysis of harmonic content. We're used to thinking about one process leading to another, when our minds are probably extraordianarily massively parallel, with evolutionary reasons for doing one thing before another.

SO,

I think when we evaluate what the high frequency content of what we're hearing sounds like, we're doing it post-buffered. We probably have some ridiculously fast way of deducing phase relationships, and our conscious perception of sound is probably distorted by that sub-conscious pre-processing. Because of the parallel nature, timing errors like I describe above might become evident as a "general feeling" because of this parallel processing - while apparently exceeding the "reasonable" expectations of the system - our hearing apparatus.


HMMMMM.

While our ears are only empirically capable of approximately 20-20k bandwidth, if our hearing apparatus is tuned to a degree that it *should* be, couldn't our mind be using some algorithmic process to extrapolate the impression of higher frequencies by interpolating differences in harmonic content from one ear to the other?


Hmm.

Wouldn't a stereo sample recorded with one channel 180 degrees out of phase clock timing-wise on one side yield a combined doubling of apparent sampling rate? Neither sample side alone would reveal anything, but what one side missed the other side could get.

Hmm.


What if you made a dual/multiple converter that rotated the clock rate

Nevermind, I'll stop, I know I'm babbling and probably not making sense. This should be amusing to read tomorrow.

http://www.mp3.com/chipmcdonald
_________________________
www.chipmcdonald.com
(tagline inlieu of having a representational page of downloadable music for the moment...) / "big ass windbag" - Bruce Swedien

Top
#441470 - 04/08/01 09:27 AM Re: George, Watch this!!!....(96k)
ChrisJ Offline
Senior Member

Registered: 02/10/01
Posts: 107
Loc: Anywhere,WY,UNITED STATES
Quote:
anti-aliasing is totally different in audio than it is in graphics,

--Lee


Lee, Thanks for your post.

I'm not an expert on computer graphics. I certainly don't know the details of how commercial applications do their anti-aliasing. That being said, perhaps a simple analogy may be helpful.

In audio we speak of frequencies meaning how frequently as a function of time does the wave change. In graphics one may gain some insight by thinking in terms of a "spatial" frequency. That is how many times does the graphic change per unit of space. For example, a graphic composed of pixels will often have changes occuring at a regular spacings equal to a pixel size.

Mathematically one can use the same tools to analyze either the audio or the graphic "signal". In the audio examples that have been discussed the hypothetical audio system does not have the frequency resolution to represent a signal that varies faster than 20 kHz. Any information contained in such very fast variations is "averaged" over or lost.

In graphics, high spatial frequencies correspond to the sharp edges --that is places where the picture changes rapidly. Applying a low pass spatial filter removes these high spatial frequencies and smooths out the boundaries between regions of the picture. The idea is that a picture may look unnatrual because there is a lot of high (spatial) frequency information at pixel boundaries but not within the pixel. As you pointed out the problem is that the systems we use to represent the picture are rather limited and introduce artifacts.

It seems that this would be analogous to a low pass anti-aliasing audio filter at least in broad terms. The mathematics at least is very similar.

Well perhaps I am just stating what is already common knowledge. If so please excuse the distraction.

Top
#441471 - 04/08/01 09:37 AM Re: George, Watch this!!!....(96k)
alphajerk Offline
MP Hall of Fame Member

Registered: 03/06/00
Posts: 7950
Loc: asheville nc usa
now thats thinking outside the box. i had thought about several of your points but too lazy to verify them so i didnt mention them. glad you did, connected them better than i could of anyways.

the brickwall limiter is my major concern. tannoy now has the supertweeter out capable of reproducing those frequencies... or so they say. also i understand square waves will benefit from higher cutoff due to foldback. and i wonder if it is less susceptable to jitter as well, or should i say the jitter results in less of a LF distortion.

i havent heard it yet so i cant say one way or the other, all i can do is point out the possible improvements.

nika, delete that last pic so i dont have to scroll across the screen to read.
_________________________
alphajerk
FATcompilation
"if god is truly just, i tremble for the fate of my country" -thomas jefferson

Top
#441472 - 04/08/01 09:45 AM Re: George, Watch this!!!....(96k)
alphajerk Offline
MP Hall of Fame Member

Registered: 03/06/00
Posts: 7950
Loc: asheville nc usa
damn chrisJ, did i leave my brain over at your place. thats the exact meaning of the visual = aural anti-aliasing analogy i was trying to imply. i find things you can see illustrate/describe what you hear quite well. at least in its simplist form.
_________________________
alphajerk
FATcompilation
"if god is truly just, i tremble for the fate of my country" -thomas jefferson

Top
#441473 - 04/08/01 01:50 PM Re: George, Watch this!!!....(96k)
Lee Flier Offline
10k Club

Registered: 09/13/00
Posts: 15398
Loc: Atlanta,GA,UNITED STATES
Chris and alpha, the analogy makes sense as a way to visualize how anti aliasing works, but the fact remains that light waves are so small and pass so quickly that if you sampled them 44.1K times per second, or even 96K or anything remotely available to us today, you wouldn't even begin to have an analogous representation of a picture. Audio sampling is able to reproduce waveforms more or less exactly because the frequencies involved in sound waves are very low. That's the only point I was making with that, in explaining why what seems obvious for graphics doesn't quite work for audio.

alpha: it's true that removing the +20K overtones changes the slope of the audible part of the waveform. That's why the filter is the critical part of a converter - how well a converter works depends on how well the filter can interpret what the slope of the original waveform was.

As for the harmonics above 20K having some psychoacoustic effect on us that we perceive, while I agree that could happen in a live performance, in a recorded piece there is no way it could happen (as a couple of us now have pointed out) because that would depend on the microphone, preamp, mixer, recorder, outboard processing, distribution media, and the listener's stereo system all being able to handle +20K frequencies. In the real world, they don't. If any one of those things can't record or reproduce those frequencies, they're outta there. So even with analog, those frequencies have never been included in recordings. Now it is possible to include them with digital recording but that still assumes everything else in the chain can handle them. In any case no one complains that they are missing when they hear analog recordings. People mostly try to use that argument (perceived harmonics) when trying to explain why digital isn't as representative as analog. I agree that right now, it isn't, but not for that reason because it's not physically possible for most analog tape to record those frequencies either.

--Lee
_________________________
What The...?
http://www.what-the.com
http://www.myspace.com/whattherock

Radio KIRA
http://www.radiokira.com


NEW CD out April 7th! >> Get it here , or here ! <<

Top
#441474 - 04/08/01 01:58 PM Re: George, Watch this!!!....(96k)
Lee Flier Offline
10k Club

Registered: 09/13/00
Posts: 15398
Loc: Atlanta,GA,UNITED STATES
ChrisJ wrote:

Quote:
Returning to Nyquist, one would do well to remember that the Nyquist theorem (which states that one must sample at twice the rate of the band that one wishes to represent e.g. 40k samples/sec is needed to accurately represent a bandwidth of 20 kHz) *assumes* that one has an ideal band limited signal --that is to say a brickwall filter that goes from unity gain to zero gain in an arbitrarily short interval of frequency.

In the real world one cannot construct such a filter. In the analog realm even as one approaches this ideal the phase and delay effects become drastic near the filter edge. This may cause audible effects.


Right. THAT is why digital representation is not perfect, and why 96K can provide some benefit, especially when applying DSP.

Of course, that still presumes that anything over 20K makes it to the converter in the first place. If you are using most of the common microphones people use today, it won't.

--Lee


This message has been edited by Lee Flier on 04-08-2001 at 01:39 PM
_________________________
What The...?
http://www.what-the.com
http://www.myspace.com/whattherock

Radio KIRA
http://www.radiokira.com


NEW CD out April 7th! >> Get it here , or here ! <<

Top
#441475 - 04/08/01 02:04 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
Alpha,

I did the same test using compression that you mentioned. Same results. Reverb: same results. Delay: same results. Etc. Maybe some things won't when effected. Maybe 96k will help with pitch processing?? That's fine. But at this point, are we all in agreement that:

Providing the converters are electronically perfect, a converter at 44.1k can accurately reproduce signals at under 20k??

Thus, higher sampling rates will NOT improve the conversion of your 10k source that you suspected. Fair? Now, if we want to talk about the benefits of 96k that are valid, I'm all ears, but I thought that we should close down one conversation before we started up the next. It seems that several people are jumping in here trying to race to other potential benefits of 96k when we haven't closed down yet what the last 60 posts have been about.

If there is any disagreement, please speak up now so that we can finish one topic. If we're all agreed on that, the two obvious holes in that are: whether or not we can "hear" anything over 20k, and whether or not the converters are electronically perfect. I have discussed both of these matters in explicit detail in the other forum as well as an issue that has to do with the mixing of those signals. If we want to revisit those topics I'm all ears, and am anxious to discuss those. I'm curious to hear about tests that others have done in these fields.

How're we doing?

Nika.
_________________________
For more...

Top
#441476 - 04/08/01 02:07 PM Re: George, Watch this!!!....(96k)
Lee Flier Offline
10k Club

Registered: 09/13/00
Posts: 15398
Loc: Atlanta,GA,UNITED STATES
Quote:
Originally posted by Curve Dominant:
Nika:
I think (?) what Nika is illustrating is that "a pure sine wave" as an issue is a fallacy, or a redundacy, or a non-issue...I'm guessing here, so flame away. There is no "pure sine wave" in any ensemble recording, so how could that possibly be a factor?


Curve, it has nothing to do with whether it's a pure sine wave or not. If you sample any two points on a complex OR sine waveform whose fundamentals and harmonics all fell within the 20-20K range, the line between those two points is going to be different from that produced by any other complex OR sine waveform that falls within that range. Since any "dip" in the waveform that adds to its complexity STILL can't be any shorter than a 20K wave (assuming it were perfectly limited to 20K, which as we mentioned, in the real world it ain't), all those audible harmonics will still be sampled if the rate is at least 40K. Does that make more sense?

--Lee
_________________________
What The...?
http://www.what-the.com
http://www.myspace.com/whattherock

Radio KIRA
http://www.radiokira.com


NEW CD out April 7th! >> Get it here , or here ! <<

Top
#441477 - 04/08/01 02:09 PM Re: George, Watch this!!!....(96k)
dj3 Offline
Senior Member

Registered: 08/18/00
Posts: 106
Loc: suffern,NY,UNITED STATES
I suppose I am a graphics 'expert.' We use smoke and mirrors to fill in the gaps left by our technology, EVERYTHING that you see in print, broadcast and in the old CRT is made up of squares. In TV and crt's its little squares of red, green and blue, in print it is cyan, magenta, yellow and black.

Your eyes fill in the rest. We can oversample the heck out of stuff and you'll never see it because of the limits of our 'playback' technology.

The exception is photography, exposed in a darkroom, (not printed on your Epson) which is a 'continuous' medium...no dots.

D/A conversion is math. You define a triangle with the end points. A sinewave can be defined with points as well (if I recall correctly). One would guess that once you have enough resolution to define the points, you've defined the wave. Any more than that and you might have spent too much money.

Maybe someone will make a way to 'photograph' music...if we could just find way to make the music sit still for the portrait.


This message has been edited by dj3 on 04-08-2001 at 11:22 AM

Top
#441478 - 04/08/01 02:10 PM Re: George, Watch this!!!....(96k)
dj3 Offline
Senior Member

Registered: 08/18/00
Posts: 106
Loc: suffern,NY,UNITED STATES
dbl post

This message has been edited by dj3 on 04-08-2001 at 11:12 AM

Top
Page 3 of 42 < 1 2 3 4 5 ... 41 42 >