Avid Pro Audio Community

Avid Pro Audio Community

How to Join & Post  •  Community Terms of Use  •  Help Us Help You

Knowledge Base Search  •  Community Search  •  Learn & Support


Avid Home Page

Go Back   Avid Pro Audio Community > Legacy Products > 003, Mbox 2, Digi 002, original Mbox, Digi 001 (Mac)

Reply
 
Thread Tools Search this Thread Display Modes
  #61  
Old 06-18-2003, 08:47 PM
Chaasm71 Chaasm71 is offline
Member
 
Join Date: Feb 2003
Location: Vancouver Canada
Posts: 68
Default Re: Digi 001 discontinued?!

Oh and yes, you should actually sample slightly above the Nyquist frequency. If you sample right at the Nyquist frequency, then you may get a flat line for signals at the Nyquist frequency. So, sample at Nyquist + 1! [img]images/icons/wink.gif[/img]

Cheers!

Charlie
Reply With Quote
  #62  
Old 06-18-2003, 08:49 PM
muspro muspro is offline
Member
 
Join Date: Dec 1999
Posts: 90
Default Re: Digi 001 discontinued?!

Hey Chaasm71,

I do not mean exactly as in your example. Those subtlies I can live with. I just meant there should be no aproximations, rounding or close enough representations.

Kinda like my diagram. This would be unexceptable.

I just need to find out how my example gets turned into a digital file that results in an "EXACT" copy by the DA.

Thanks Again!
Reply With Quote
  #63  
Old 06-18-2003, 08:56 PM
Chaasm71 Chaasm71 is offline
Member
 
Join Date: Feb 2003
Location: Vancouver Canada
Posts: 68
Default Re: Digi 001 discontinued?!

Two dots (the digital data) does not look like a sign wave. It is the DA converter that is clever enough to recognize that those two dots are actually a sign wave, and to reconstruct them as such. Again, not visually obvious, but look back to the circle example, and flex your analogy muscles! [img]images/icons/wink.gif[/img] If you wanna get all mathy, we can do that too...

Charlie.
Reply With Quote
  #64  
Old 06-18-2003, 09:02 PM
Chaasm71 Chaasm71 is offline
Member
 
Join Date: Feb 2003
Location: Vancouver Canada
Posts: 68
Default Re: Digi 001 discontinued?!

Duardo, thanks for the link to Strang's paper on DSP. I've read a book by him on linear algebra, and he's pretty good at explaining things, though he has a weird writing style. Anyhow, I've printed out the postscript file from www-math.mit.edu/~gs/papers/newsigproc.ps, and I'll give it a read. I'll see what he has to say.

Cheers!

Charlie.
Reply With Quote
  #65  
Old 06-18-2003, 09:27 PM
Duardo Duardo is offline
Member
 
Join Date: Oct 2001
Posts: 978
Default Re: Digi 001 discontinued?!

Wow, look at everything that happened while I was writing my last response.

Quote:
Anyway here's the real flaw to the whole 96k improvement:
the same converter that you use at 96k is the one that you use at 48k, in other words it is still an improved converter capable of the finer resolution.
<font size="2" face="Verdana, Arial">No, not necessarily. It may be so with some converters, but it doesn't have to be if the filters are right. Sure, it's capable of finer resolution, but we're not capable of perceiving that finer resolution. Otherwise we could go to 192, to 384, to 1 mHz and beyond...there's got to be a point where we can't perceive a difference. And it was determined where that point is long ago.

Now all we have to do is build our converters so that they take full advantage of that. I don't thing that we're there yet.

Quote:
Think of it like PPQ, a sequencer with finer resolution still sound somewhat different even at the lower resolution.
<font size="2" face="Verdana, Arial">Yes, but it's not the same thing at all. It's also not the same thing as the pixellated photograph example everyone uses. If anything, pixels would relate to bit depth. Frequency response would be analogous to a picture that was capable of reproducing infrared and ultraviolet light. What's the point?

Quote:
The higher resolution is both a hardware and software driven aspect.
<font size="2" face="Verdana, Arial">Yes, and certain software applications may well sound better at higher sampling rates. Even at lower rates many of them will upsample, process, and then downsample.

Quote:
All this to say that the 002 should sound better than the 001 since the converters had to be improved to deal with 96k, 96k should sound different from 48k since it's the converters in their best mode.
<font size="2" face="Verdana, Arial">I haven't A/B'd them so I don't know if that's true for the 002. It may sould better, but it's not necessarily the case that it "should". But the main improvements in the converters...mostly having to do with dynamic range...have nothing to do with the sampling rate.

Quote:
As far I'm concerned, the higher rates make it easier to build the filter, that's about it. Maybe lower the noise floor.
<font size="2" face="Verdana, Arial">They may make it easier to build filters, but with today's oversampling converters that doesn't have to be much of an issue. It can lower the noise floor if noise shaping is used, but that's not what we're talking about here...

Quote:
There are so many other aspects to the conversion that straight sampling theory kind of goes out the window-no?
<font size="2" face="Verdana, Arial">Well, sort of...obviously, do whatever sounds best...if your converter sounds better at 96 khz, by all means, record at 96 kHz. But it can't hurt to be educated, and maybe understand why somethign may or may not sound better than something else.

Quote:
Muspro, I think your point of confusion centers on the unintuitive fact that you only need two points on a sine wave of a known frequency and amplitude to define it provided that those two points lie less than 1/2 a period of the sine wave of eachother (this is just Nyquist's theorem stated in terms of period rather than frequency!).
<font size="2" face="Verdana, Arial">Yes...and it is very counter-intuitive. It only makes sense that more slices=more accurate representation. But once you understand that that's only true to a point, and that point is easy to define, suddenly it makes sense.

Quote:
That is, the rest of the wave can be inferred mathematically given these two points and the condition on the sampling frequency. So, I agree that two points don't look much like a sine wave, but if you are good at math (like Mr Nyquist was!), then two points is all you need!
<font size="2" face="Verdana, Arial">Exactly...and D/A converters are very good at math.

Quote:
We have all assumed forever that 20 Khz was the absolute highest frequency any human past, present and future could ever hear. I for one just came to that little bit of information reading books, not through actual practices. So here is the deal. What if we can hear higher than 20 Khz? And if we can't, says who and how did they come about that conclusion?
<font size="2" face="Verdana, Arial">I don't think we have assumed that, there have been numerous studies conducted that have come to that same conclusion...not to say that there aren't further discoveries ahead that may change the way we look at things...

Quote:
It's been widely reported by many engineers, specially mastering engineers that they can clearly perceive and or hear audio above 20 Khz.
<font size="2" face="Verdana, Arial">Most of the examples you hear about this sort of thing have to do with engineers hearing that something's off and finding out through measurement that there's an odd bump or dip up around 40 kHz or 50 khz or so. Again, one can look at those stories and say "oh, so, if they can hear up there then we must need to capture that information"! When what they're probably hearing is the phase shift caused by that boost or dip in the audible range.

It's easy to do a quick demonstration of that...there are many analog EQ's out there that go well beyond 20 kHz, and when you have them boosted at their extreme high settings (say, 30 kHz or whatever) you can clearly hear the difference. So people assume they're hearing 30 khz. Easy way to disprove that is to record what you're doing at 44.1 khz. You'll still hear it...so you're obviously hearing the effects in the audible band.

Quote:
As ironic or contradictory as it may sound, adding frequencies beyond 20 Khz permeates the sound with that 'something' which gives the playback a more natural, airy, fuller sound. As if the listener had a sort of sixth sense we are now just starting to perceive or realize.
<font size="2" face="Verdana, Arial">Yeah, that's a thought that's not likely to ever be proven.

Quote:
I posted a very basic diagram on my web site:
www.Star-Studios.com
It is under the proofs or digital pics link on the left hand side.
It represents a frequency that is 1/2 the sample rate. In my example I use 40kHz sampling (to make the math easy) and a 20kHz sine wave.
Duardo says he can make the two digital samples (blocks) look EXACTLY like the original wave. Better yet his DA knows what the original wave looked like based on the 2 samples of digital information.
<font size="2" face="Verdana, Arial">I never said anything about blocks. In your example you'd take your samples where the sine wave intersects the three vertical lines. What you've drawn there is obviously not a sine wave, but the point is that from theose three samples (not two) the D/A would know what the original "looked" like.

Nice looking studio, by the way.

Quote:
This is an extreme example but show what happens when you use sample rates that are close to the frequency that you want to record.
<font size="2" face="Verdana, Arial">There's no problem with sampling rates (where the Nyquist frequency is) close to the highest frequency you want to record. The only place that could be an issue would be where the highest frequency was exactly half of the sampling rate, which would never happen because a) it would have already been filtered out and b) why would you want to record a 22.05 kHz sine wave?

Quote:
It becomes very hard to approximate when you consider the analog waves do not line up perfectly with the sampling frequence as in my example.
<font size="2" face="Verdana, Arial">It is not hard at all for the D/A converters to do this!

Quote:
Notice how 1/2 of wave passes through the zero point in the middle of the sample! You would have to double the sample rate (WOW! novel idea) to represent it more accurately.
<font size="2" face="Verdana, Arial">It doesn't matter where it passes through the zero crossing. You've got one sample above, one sample below, and based on the samples before those and the fact that you cannot have frequencies above the Nyquist frequency there is only one point where the wave could possibly have crossed the zero crossing.
Let me know what you come up with.

Quote:
It is sufficiently complicated that I don't think it can be explained in detail to someone without there having a strong math background.
<font size="2" face="Verdana, Arial">Actually, it can be simplified pretty easily...although yes, the proof is much more complex.

Quote:
However, to make an analogy, suppose I have a picture of a perfect circle and you want to sample it (capture all the info you can about it). All there really is to learn about this circle is the co-ordinates of its center and its radius. So, if you sample enough to figure out the co-ordinates of its center, and its radius, do you learn more if you measure the locations of hundreds of additional points on the circle? Nope. Its the same with sine waves, though not as graphically intuitive. Hope this helps.
<font size="2" face="Verdana, Arial">Good analogy...very simple, but at the core that's exactly what we're talking about...esepcially with our theoretical sine wave.

Quote:
You may see a square wave in your computer, but as soon as it hits the voice coil in your speaker, guess what? It takes time to accelerate and decelerate, it is no longer a square wave!
<font size="2" face="Verdana, Arial">Actually, the filters in the D/A converter take care of that before it hits your speaker. But yeah, in most cases the speaker doesn't help...

And if you did have a theoretically perfect recording and playback chain, that was flat up to 100 kHz and beyond, there's still one transducer that would "round off" those square waves and filter out those overtones...your ear.

Quote:
Great info Chaasm71! I am with you. I still don't understand (as in my example), how a DA can derive the missing information. It isn't represented at all by the digital information.
<font size="2" face="Verdana, Arial">No, it's not represented by the digital information. The digital information is sufficient for the D/A to interpolate all the missing information.

Quote:
I am talking about EXACT reproduction of the original analog material. Are you saying this is what is being done? Exactly! I still have a hard time believing that.
If that were the case, it sounds more like a lossless digital codec not a DA conversion.
More info please!
<font size="2" face="Verdana, Arial">Yeah, that is pretty much what's being done...obviously, our analog circuitry isn't perfect (and likely never will be) it won't be 100% exact...but it's as close as we can get now, and simply increasing the number of samples won't get us any closer.

As for it sounding like a lossless digital codec, isn't a codec used to compress (and uncompress) data? What we're talking about here is uncompressed digital audio.

Quote:
I'm not sure where the practical limit to the bit depth/"vertical resolution" is, but that's probably measurable as how much distortion we hear.
<font size="2" face="Verdana, Arial">In a nutshell, the more bits, the lower our quantization noise is pushed. Each bit pushes it down 6 dB lower, so the more bits you have, the more dynamic range you have. It's totally different than frequency response in many ways, but there are just as many misconceptions about that...that more bits gives you better "resolution" at all levels is the biggest one, and just like the misconception that higher sampling rates give you more "resolution" of <20kHz audio, it's just plain wrong. But we're not here to talk about that...what were we talking about? The 001 being discontinued?

Quote:
I'm sure we can only hear distortion down to a certain point, similar to how we can only hear freqs up to 20k ( or see light only in the visible spectrum, or taste chemicals down to so many ppm, etc.)
<font size="2" face="Verdana, Arial">We can actually hear distortion down at a pretty low level, but if the stuff we're listening to at, say, an average of 85 dB SPL has a dynamic range of 30 dB (in other words, 100 dB at its loudest and 70 dB at its quietest) whether the noise floor is at 40, 30, 20, or 10 dB SPL is irrelevant.

Quote:
I think I've read about studies that have tested for physiological effects form high freq sounds ( sorry, can't quote anything) and I could see how that info could alter our "perception" of a piece of music.
<font size="2" face="Verdana, Arial">Yeah, you probably have...the only halfway credible one I'm aware of is the Ooahashi (sp?) study, which showed that our bodies were capable of detecting ultrasonic sounds on a subconscious level...but not that anyone was able to consciously perceive a difference. If you have to hook electrodes up to me for me to be able to "perceive" information above 20 kHz I'm not sure if that's really something I need to worry about.

-Duardo
Reply With Quote
  #66  
Old 06-18-2003, 09:33 PM
NannerPuddin' NannerPuddin' is offline
Member
 
Join Date: Apr 2000
Location: Irvine, CA, USA
Posts: 278
Default Re: Digi 001 discontinued?!

I have no choice but to upgrade for 96k. I record albums of music for dogs. My clients can definately tell a difference.

NP
Reply With Quote
  #67  
Old 06-18-2003, 09:52 PM
Duardo Duardo is offline
Member
 
Join Date: Oct 2001
Posts: 978
Default Re: Digi 001 discontinued?!

This time I double-checked to make sure there weren't any more new responses before posting mine, so hopefully I won't re-tread any ground here...

Quote:
The difficult task here is that the waveform does not start on the sample.
<font size="2" face="Verdana, Arial">Every sample will correspond exactly with one point on the wave. Theoretically, the very first and very last two samples may be a little off since the curve that connects each dot is determined by the the "dots" that come before and after it (which is why in the scientific papers they always say that it really applies to an "infinite" wave), but that's not going to make an audibly perceptible difference.

I think when you realize that we're talking about "dots" and not "block" here it will likely make a lot more sense.

Quote:
If I give you two points and tell you that they lie on a sign wave of a given frequency and amplitude and also assure you that they are within half a period of the side wave to eachother, then you have no choice as to what that signwave looks like. It has been completely determined.
<font size="2" face="Verdana, Arial">Exactly, except I think you'd need at least four points to determine the curve between the middle two...but yeah, that's what it is.

Quote:
So is everyone saying that all things being equal, there is absolutly no benifit for sample rates higher than 44.1/48?
<font size="2" face="Verdana, Arial">No, not quite...there may be other benefits. Maybe you can use cheaper filters and still get the same results. If you're sampling something and then manipulating it digitally there may be advantages to that (say you sample something and want to drop it an octave or two...if you've captured all of those inaudible harmonics, when you shift it down in pitch suddenly they're audible...could be a good thing or a bad thing). But as far as accuracy of frequencies below half the sampling rate, that's right.

Quote:
Considering only resolution, the same converter sounds EXACTLY the same at 44.1, 48, 96 and 192! Then why the hell do we have the high SR.
<font size="2" face="Verdana, Arial">Well, there are lots of reasons...we're not necessarily considering only resolution.

Quote:
I have had so many people tell me their HD3 system sounds better at high SR! Is it just because the converters perform better at higher SR?
<font size="2" face="Verdana, Arial">They may...I haven't A/B's those converters at higher SR's myself...but I'd bet that most of the people who have told you that haven't either.

I do think, however, that if Digidesign really felt that there was a real audible advantage at the higher sampling rates, they would have come out with a 96k-capable system long ago, rather than waiting until the market was absolutely screaming for it. But I could be wrong.

Also, a lot of times people's opinions may be clouded by what they want to hear. Personally, I don't want higher sampling rates to sound better. Don't understand why people do. Why would you want to cut your processing power and track count in half and double your hard disk space? I suppose sounding better would be a good reason. But if it were possible to sound just as good at 44.1kHz, why wouldn't you want that?

Quote:
Oh and yes, you should actually sample slightly above the Nyquist frequency. If you sample right at the Nyquist frequency, then you may get a flat line for signals at the Nyquist frequency. So, sample at Nyquist + 1!
<font size="2" face="Verdana, Arial">Yes, that's true as well...that's pretty much taken care of by anti-aliasing filters, though, so it doesn't really concern us. The next paragraph from that MIT paper I quoted basically said the same thing.

Quote:
Duardo, thanks for the link to Strang's paper on DSP. I've read a book by him on linear algebra, and he's pretty good at explaining things, though he has a weird writing style.
<font size="2" face="Verdana, Arial">Interesting. I just kind of pulled him up at random on a Google search...I think his came up third or fourth. It was the first one that said what I was looking for in language that I could understand.

The Nyquist theory doesn't just apply to audio, it applies to digital conversion in general...my brother-in-law is an electronic engineer in the X-Ray division at GE, and a while back we were discussing Nyquist for some reason. I think it's because we're both kind of geeky.

-Duardo
Reply With Quote
  #68  
Old 06-18-2003, 10:55 PM
muspro muspro is offline
Member
 
Join Date: Dec 1999
Posts: 90
Default Re: Digi 001 discontinued?!

Hey Guys,

Here are some examples from text books, magazines and web sites that support that higher sample rates = better resolution and more accurate reproduction of the analog signal. Are all of these authors and engineers wrong? If they are, I need clear proof that higher SR have no apparent effect on the resolution of the digital audio by creating a more accurate representation of the original signal. And any noticable sound quality improvment is due to another factor.

The 3D audio link has information about a listening test comparing 44.1, 48, 96 and 192. It is a good artical that I read a while back and is part of the basis for many of my statements.


Digital Audio Basics
Your computer can only store numbers using a limited number of digits or precision. Continuously varying sound is called an analog signal. Once the computer grabs the sound, it doesn’t have enough precision to store all the information about the sound in order to perfectly reproduce it. What the computer has stored is called a digital signal representation.

Your sound card captures information about an analog sound signal by measuring its intensity at a given instant. This corresponds to one single point on the waveforms we’ve been looking at. In order to capture an entire waveform, the measurement process must be repeated at a high rate, usually thousands of times a second. Since the hardware has limited speed and memory capacity, there are only so many points it can capture. Any information between those points is lost forever.

To play back a sound, we just reverse the process and convert the digital samples back to an analog signal. Of course, the new signal will probably retain some of the staircase effect, so the reproduction won’t be perfect.

-----

Sampling Rate
Number of samples or snapshots taken of a particular signal in a given amount of time (usually one second). Higher sampling rates result in more pieces of the true signal. When a sufficient number of pieces are generated, the pieces meld together to form a very close approximation of the original.

-----

DVD technology holds data encoded at a 96 kHz sampling rate with a quantisation of 24 bits resulting a vastly superior audio reproduction format compared to standard CD.

-----

Resolution and sound quality

The division of the sound into discrete samples means that there is some loss of information. The process fails to capture fluctuations in the signal which happen in the time interval between one sample and the next. Similarly it fails to capture fluctuations which are smaller than its measurement steps. So the resolution of the sampling process determines how much information about the sound is captured and how much is lost.

-----

http://www.3daudioinc.com/3daudio_hi-res.html
Reply With Quote
  #69  
Old 06-18-2003, 10:56 PM
M Lawrence M Lawrence is offline
Member
 
Join Date: Sep 2001
Location: N.Y.C.
Posts: 349
Default Re: Digi 001 discontinued?!

Quote:
the A/D Conversion and new Pre's blow the 001 out of the water!!!
<font size="2" face="Verdana, Arial">the cranesong hedd192's converters blow the 001's outta the water. i think perhaps the perceived diff between the 002 & 001 is just wishful thinking (it's certainly grossly exaggerated in the above statement).


ml
Reply With Quote
  #70  
Old 06-18-2003, 11:28 PM
DIGIDUDE26 DIGIDUDE26 is offline
Member
 
Join Date: Mar 2003
Location: ca
Posts: 298
Default Re: Digi 001 discontinued?!

A lot of this subject could make a non-math guy (myself) go freekin nuts.........although I do enjoy super technical discussions like this one, there is a point for each contender of the piss contest to realize (if they know what they're talking about which looks as though), that they are correct with their aspect. You guys are talking about 2 different issues that are related to the same fact, and thats why this urination continues lol. If you guys would just add the two discussions together you will have the basis of digital technology (related to audio).

lol, but anyways don't let me get in the way, please continue.........."LET'S GET IT ON!" Round 3 ...ding, ding
__________________
002r, Dual 1.25 G4, 1gig DDR, Dual 17 Studio Display, 12 Powerbook, HR-824's, NS-10m's Good Times!
Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
003 Discontinued? bootc Pro Tools 10 58 12-05-2011 06:15 PM
Is 003 really discontinued? ripekeai Pro Tools 9 15 03-29-2011 06:05 PM
Digi 003 Rack Discontinued? Tritono9 003, Mbox 2, Digi 002, original Mbox, Digi 001 (Win) 12 06-04-2010 06:39 PM
Digi 001 discontinued? Dino Bonanno 003, Mbox 2, Digi 002, original Mbox, Digi 001 (Win) 1 01-13-2004 07:22 PM
001 being discontinued??? fabpab 003, Mbox 2, Digi 002, original Mbox, Digi 001 (Win) 5 11-10-2002 04:18 AM


All times are GMT -7. The time now is 01:29 PM.


Powered by: vBulletin, Copyright ©2000 - 2008, Jelsoft Enterprises Limited. Forum Hosted By: URLJet.com