For all you Vinyl junkies . . .

[amey01]

It is impossible to create a curve by filling in squares on a sheet of graph paper but you can create the illusion of a curve.

........but we don't listen to the "squares on the piece of paper" - that is the job of the DAC to convert that into analogue!

The DAC superimposes the analogue waveform over the "squares on the paper" - and within the frequency limits to which we are referring, what is superimposed is PERFECT.

What comes out of the DAC is the "perfect" representation - perfect if not for a whole host of other issues!

 

[amey01]

...If we listened to pure sine waves we could reconstruct them perfectly, but music is a combination of sine waves...

A combination of sine waves and what? It is pure sine. 20,000Hz (with two samples per cycle) can theoretically be regenerated perfectly, by a perfect DAC.

20,000Hz (22,050 actually) is our worst case scenario Bernard. It doesn't get any more complex than that because anything above gets filtered out by the anti-aliasing filter. Of course the filtering (and that 20,000Hz limit itself) are downfalls, but that's not what this argument is about. I am responding to Craig's comments "there are minute gaps in the music" - there are not; and "digital looks like squares on graph paper" - it does not.

And of course Wardsweb's comment that digital is flawed because there are different sized pits - well, that's how digital encoding works - look up EFM "eight fourteen modulation".

 

[User211]

Read that link I posted above, guys, and you will understand it... or at least you should do...

 

[amey01]

Actually, Adam is right - he didn't express it brilliantly, but he is basically right.

Thanks Justin - sometimes I need to try to be a bit more eloquent! I guess it comes from being at work and posting between meetings!

 

[Craig]

........but we don't listen to the "squares on the piece of paper" - that is the job of the DAC to convert that into analogue!

The DAC superimposes the analogue waveform over the "squares on the paper" - and within the frequency limits to which we are referring, what is superimposed is PERFECT.

What comes out of the DAC is the "perfect" representation - perfect if not for a whole host of other issues!

You're right, we don't listen to paper. That was just an analogy.

A squarewave representation of sound is a less accurate representation than a sinewave. In order to convert a squarewave to a sinewave you have to fill in the missing information and that is accomplished in various ways by the digital to analog process. Upsampling, resampling, are just two ways to further smoothout the signal but all it is doing is creating smaller squares (or higher frequency) the same as a higher rez digital source.

Higher resolution digital comes closer to a sinewave as the conversion process better refines the signal as it is altered into an analog signal. The missing information that must be filled in to create a curvy sinewave is a best guess by the algorithm used to create a sinewave and hence, analog signal. If the digital signal were perfect and complete there would be no need to convert to an analog sinewave since it would already be there.

Yes, the digital signal is converted to analog to do that requires a sinewave but to convert squares to curves requires those missing areas be filled in with best guess information such as with a CD being converted the DAC.

A digital signal can converted to analog very accurately but is still only making an educated quess needed to fill in the gaps of missing info to create a sinewave for use in reproduction of sound. Its very close but not perfect or exact in computer science terms.

.

 

[Bernard]

Higher resolution digital comes closer to a sinewave as the conversion process better refines the signal as it is altered into an analog signal. The missing information that must be filled in to create a curvy sinewave is a best guess by the algorithm used to create a sinewave and hence, analog signal. If the digital signal were perfect and complete there would be no need to convert to an analog sinewave since it would already be there.

I agree with you completely.

Forgetting Nyquist and the mathematics behind all this stuff, let's look at it simplistically: If the sampling is such that a sample occurs before a peak, and the next consecutive sample occurs after the peak, you have no way of knowing when in time the peak occurs (i.e. where between the samples), and you also have no way of knowing the amplitude of that peak, ergo, you cannot create the waveform exactly, as it is not a perfect sine wave. I am not saying it sounds like crap, but just saying that the waveform cannot be recreated perfectly.

Justin, I looked briefly at the link but did not plough through it all. You are right - the people here are polite compared to the guys on that site.

 

[amey01]

let's look at it simplistically: If the sampling is such that a sample occurs before a peak, and the next consecutive sample occurs after the peak,

I've been thinking about this for a long time - but I don't know how it works. Worse, what if the samples occour at 0 - how is the signal even recorded? Reached the limits of my digital theory - anyone else know?

I'm guessing this is part of the job of the ADC. That is, to get those samples in at the correct point. Perhaps by some sort of timeshift? Worst case would be 1/88,200 second out which I'm sure is far less than any other medium. Don't forget, the ADC gets to see the full analogue signal, no the digital representation. It's job is to ensure those digital representations are correct.

All I know is that it does work. Dunno how!

 

[amey01]

You're right, we don't listen to paper. That was just an analogy.

A squarewave representation of sound is a less accurate representation than a sinewave. In order to convert a squarewave to a sinewave you have to fill in the missing information and that is accomplished in various ways by the digital to analog process. Upsampling, resampling, are just two ways to further smoothout the signal but all it is doing is creating smaller squares (or higher frequency) the same as a higher rez digital source.

Higher resolution digital comes closer to a sinewave as the conversion process better refines the signal as it is altered into an analog signal. The missing information that must be filled in to create a curvy sinewave is a best guess by the algorithm used to create a sinewave and hence, analog signal. If the digital signal were perfect and complete there would be no need to convert to an analog sinewave since it would already be there.

Yes, the digital signal is converted to analog to do that requires a sinewave but to convert squares to curves requires those missing areas be filled in with best guess information such as with a CD being converted the DAC.

A digital signal can converted to analog very accurately but is still only making an educated quess needed to fill in the gaps of missing info to create a sinewave for use in reproduction of sound. Its very close but not perfect or exact in computer science terms.

.

You're neglecting the existance of the DAC - it's job is to create analogue sound.

As long as it gets enough information to recreate that analogue sound then that's exactly what it does.

To use an analogy myself, it's like lossless compression. There are bits missing, but there is enough information there to recreate the oringinal perfectly. Lossless doesn't sound worse because an algorithm has to fill in the missing bits, does it? This is what digital audio does at sub-Nyquist frequencies - the samples are enough to recreate that analogue wave perfectly. No gaps, no guessing. Educated guessing or otherwise.

I quote the image from Justin's link to explain myself. The first shows only two samples per cycle. The second shows multiple samples per cycle. Both show a DAC-generated sinewave (in black) that is THE SAME!

b)'s analogue signal does NOT contain more resolution because it had more samples!!!

So whether you have two, 50 or 64,305 samples per cycle you are going to get the same sinewave.

Think about it - if these errors existed, we would be generating an altered frequency response!! Surely you know that is not happening! Now, with 16-bit audio, there IS guessing in the amplitude side of things - ie: there are only one of 65536 amplitudes the signal has to match. But I digress.

 

[amey01]

Pre-empting further questions - I must also clarify from my above post. I am not saying that 44.1kHz sampling is all that is necessary. I have stated many times - 44.1k digital has some very serious and real issues. Higher sampling rates produce very real benefits - both theoretically and to my ears at least.

But these are everything to do with higher frequency response and less filtering / different filtering required. They are NOT to do with less "guessing" or higher resolving ability in the 20-20,000 Hz range.

MMM - intense debate about digital theory is a very strange topic for the "Analogue" section of the site!

 

[Rich]

MMM - intense debate about digital theory is a very strange topic for the "Analogue" section of the site!

:ROFL: Yeah, when I posted that link, the last thing I expected was that the thread would turn into a debate about digital audio reproduction. But then, that is the beauty of an open forum. You never know what twists and turns the conversations will take, but you are always quite likely to learn something along the way. Thanks to Justin for posting that link. It was a very interesting read.

Personally, I have never really heard any big difference between really good analog and really good digital. That, and the maintenance issues of analog, are why I gave up on vinyl many years ago. Some of the best sound I heard at RMAF was from streaming digital sources and some of the worst was vinyl (and vice versa).

As for the original posted link, I would love to find a picture that shows the needle in the groove.

 

[FOUNTAIN]

As for the original posted link, I would love to find a picture that shows the needle in the groove.

I wonder if they could do a short video with a needle in the groove through a microscope? Or maybe a series of pictures would provide the same result. That would explain a lot if we could actually observe the needle tracking along the walls. Of course I'm not sure this is possible so a picture would be quite nice as well.

Glen

 

[Craig]

A video of a stylus/needle in a groove would be pretty cool. That would be challenging since it would need to be taken with a high-speed microscope video camera in order to capture a tiny needle vibrating in the groove.

What was that show on TV that did high-speed high-def videography to capture actions that were too fast to see such as a ballon popping or what really happens in super slow-mo when a droplet hits the water? Is that show still on?

 

[RichTeer]

What was that show on TV that did high-speed high-def videography to capture actions that were too fast to see such as a ballon popping or what really happens in super slow-mo when a droplet hits the water? Is that show still on?

That would be Time Warp. I don't know if it's still on, though (apart from re-runs).

Nice to see someone using correct terminology; "needles" went out with wax cylinders!

 

[User211]

Thanks to Justin for posting that link. It was a very interesting read.

It really cleared my head and led to some interesting discussions between my collegues and I at work, let alone in that thread.

Sometimes, forums are actually useful... as long as people don't get too irate about things, can continue to discuss the subject in a civil manner, and can accept they are just plain wrong when they are wrong.

Never let pride get the better of you. If you are wrong - admit it! It simply means you've learnt or even re-learnt something.

 

[Craig]

That would be Time Warp. I don't know if it's still on, though (apart from re-runs).

Nice to see someone using correct terminology; "needles" went out with wax cylinders!

"Time Warp"...Yep, that's the show. Neat stuff.

Funny how the term needles is still around.

 

[Craig]

........but we don't listen to the "squares on the piece of paper" - that is the job of the DAC to convert that into analogue!

The DAC superimposes the analogue waveform over the "squares on the paper" - and within the frequency limits to which we are referring, what is superimposed is PERFECT.

What comes out of the DAC is the "perfect" representation - perfect if not for a whole host of other issues!

This a discussion on Sampling Theory:
An approximation is not perfection but yes it is a very good one and highly unlikely the human ear could detect any difference.

Even if you sample at twice the bandwidth it is still just a sample of the original sound and the output from a DAC is only an approximation of that sampling. An approximated output from sampling cannot be claimed to be 100% accurate or perfect unless you use the word perfect loosely and not literally. Mathmeticians are proud of their work and like to claim their sampling theory formulas are "perfect" and maybe the formulas are but ultimately the output of a DAC is not a perfect or exact representation. Although, close enough to sound great.

That doesn't mean vinyl sounds any better than digital because there are more obstacles to vinyl than digital.

 

[amey01]

This a discussion on Sampling Theory:
An approximation is not perfection but yes it is a very good one and highly unlikely the human ear could detect any difference.

Even if you sample at twice the bandwidth it is still just a sample of the original sound and the output from a DAC is only an approximation of that sampling. An approximated output from sampling cannot be claimed to be 100% accurate or perfect unless you use the word perfect loosely and not literally. Mathmeticians are proud of their work and like to claim their sampling theory formulas are "perfect" and maybe the formulas are but ultimately the output of a DAC is not a perfect or exact representation. Although, close enough to sound great.

That doesn't mean vinyl sounds any better than digital because there are more obstacles to vinyl than digital.

Well, now it is coming down to how losely we are gong to use that word "perfect".

What is recorded will never be perfect because the air does not transmit the sound perfectly, the mic doesn't pick it up perfectly, the cable doesn't take it to the mic preamp perfectly, etc etc etc.

So perfect is a very powerful word and one could not argue that anything is ever perfect.

But in terms of that sinewave and gaps - yes, I'm afraid I still hold it is perfect. In terms of the amplitude, then as I stated earlier, no - the value can only be one of 65,536 values so it is really a gross approximation. But as you say - good enough to sound good.

 

[Craig]

Whatever the mic picks up good or bad is what gets recorded and thus establishes the standard from which to measure and compare. But yes, it really comes down to agreeing on a threshold or degree of precision in order to determine what perfection really is.

The one inherent advantage of an analog over a digital source is that avoids the whole A to D then D back to A process in the first place. And the magnified photos and graphs are great illustrations of those fundamentals.

 

[FISH_MAN]

that's a great post Rich ! I know I love the sound of my records even the weaknesses often come across as strengths most of the time.

 

[FOUNTAIN]

A couple of these pictures are repeats from Rich's link, but there are some new ones as well. If somone has 3D glasse,s there is one pic that will let you view it in 3D. Way cool!!!!!


http://www.synthgear.com/2010/audio-gear/record-grooves-electron-microscope/

Glen