Understanding Sample Rate

[beerandmusic]

6 minutes ago, crenca said:

 

I'mmm baaaccckkkk

 

No, because of the nature of sound, more samples does not mean more accuracy.  Once you reach a certain point, any further samples add nothing, zero, nada, to the sound's accuracy.  This truth has been measured and can be described - like other physical phenomena like the speed of light, or energy of a hydrogen atom, etc.

 

Beerandmusic, your still thinking of sound as a quanta, a series of "infinite" events in "infinite" time.  You are thinking of sampling as a real sample of the music - like the sample is a "capture" of the sound itself just like when a person goes out into their front yard and collects a "sample" of the grass.  You also think that these samples are put back together again in a series, like a puzzle, so that the more pieces you have the better or "more accurate" the picture is.   None of this is true.

 

(tangent:  You see folks, is it not unfortunate that the term "sampling" has been used - laypersons have a completely different understanding of what this term means)

 

Question beerandmusic:  I am 6'1 in height.  If I "sample" my height once a year, or once a day, or once a second, does the "accuracy" of my measurement change?  If I set my cruise control in my car at 60 miles per hour (and it is reasonably accurate - it always keeps my car within 2 miles per hour of 60 miles per hour), if I "sample" my speed 100 times a second, would me speed be more "accurate" then if I sample my speed about once a second...or once a minute...or just a few times between point Albuquerque and San Diego?

 

A true sampling story that occured on the internet one day:

 

beerandmusic, what is your name?

 

Crenca, my name is "beerandmusic"

 

But beerandmusic music, I only know you "digitally", not in the real world, so I need to be more accurate  - what is your name?

 

Crenca, I already told you, my name is "beerandmusic"

 

Yes, but this is digital sampling, so for accuracy I will need to ask  you several times a second - what is your name?

 

Crenca, obviously you don't understand, your initial sample is an accurate sample of my name, and doing more samples does not give you a more accurate understanding/measurement of my name.

 

Why?  Because some engineers told you?  Logic dictates that because we know each other digitally, the more samples I get from you the more accurate I can hear you, see you, and know you - a sample is just a sample after all, and the more the better because I only captured a few instances of your name, I must need more for accuracy - what is your name?

 

But Crenca, names don't work that way, you have a fundamental misunderstanding of the nature of names

 

beerandmusic, I have already told you how sampling works...more is better because it leads to more accuracy - what is your name?.

 

 

 

when you sample the same thing, the accuracy doesn't change, but if i changed my name the sampling matters for accuracy.

i will ponder the rest.

 

[beerandmusic]

27 minutes ago, kumakuma said:

 

 

 

 

this is actually very good....for debate and understanding which i will elaborate on in future.

 

I am raising your scale from a negative 5 out of a possible negative ten to a negative one for this submission....but you have a long ways to go before you can catch jabbrs or miskas current rating of 10 (wink).

 

you may not give a hoot what i think of you, but i felt bad for inferring previously that you are a worthless piece of trash (grin)....so it is more for me than for you (wink)

[beerandmusic]

6 minutes ago, crenca said:

 

Very good!  NOW we are getting somewhere.  Tell me, are you a god or a man?  If you are a god, then you have the ability to change your name at an infinite rate (if you so choose) and since I am a man I could not sample your name fast enough to accurate reproduce your name.  However, if you are a man, then you are limited by the laws of the universe - you are a creature, and you are not infinite.  Thus, there is a limit as to how fast you can change your name. Tell me, what is this limit?  

 

How fast can you change your name???  

 

I ask because I can then tell you (even better, I can prove it too you) how fast I need to sample, to accurately, fully, without any errors or "gaps", reproduce your name...

 

i changed my name the moment that you did not sample....i changed it back the next moment you did not sample....you have the correct current name but you have no knowledge of the moment i changed my name....you are missing the details....but it doesn't matter now, because it is in the past.

[beerandmusic]

15 minutes ago, crenca said:

 

Ah, you are a god then.  You can change your name at an infinte rate.  I am but a man, and I live in the world which you created - this orderly world where men can only change their names up to a certain frequency.  True, bats and other animals can change their names at a greater frequency, but even they have their limits.  Man, being a creative and intelligent creature (made in your image of course) through his art can make machines that can change their name even faster (much faster) than creatures, but alas, even these machines have  their limits.

 

What is it like to be a god, such that you can not be sampled accurately?  Tell me please, I am but a lowly man...

 

even man can create sound at an "infinite rate"...the plucking of a guitar for instance.

the complex waveform has an infinite number of frequencies, at an infinite number of time slices,  within it's own frequency range (between 50 and 300hz) that man just is not able to record accurately.

[beerandmusic]

13 minutes ago, crenca said:

 

Man?  no.  Demigod?  Perhaps.  Do you speak of the winged Cupid, plucking on his strings?  These strings produce frequencies of an infinite rate?!?!  Oh, to be a god and hear the infinite sounds of the heavens!!  Tell me noble beerandmusic, your highness, what is it like to hear the infinite sounds of the gods and their heavens!!??  Please tell me, for I am but a man, and when men pluck on our crude and limited strings, only finite frequencies come forth from them....OH, MY SOUL my soul,  I lament at my mortality, my finiteness, my limitations.  PLEASE, please noble and great beerandmusic, tell me what do the sounds of the gods sound like??

"infinite rate" was taken in your context.

 

But i have heard "things" that you would never believe in the material world you know....but i won't go into because it is off topic (lol)...only sharing because you asked....it was in deep meditation, and a state I have never been able to get back to...

 

it was in my quest for trying to undestand what God would want praise and adoration....when i discovered it wasn't for God, it was for unity of man.

 

 

[beerandmusic]

2 minutes ago, esldude said:

Fourier analysis says any waveform no matter how complex can be broken down into a series of sine waves that combine to the complex waveform.  There is no level of waveform complexity that cannot be broken down into an equivalent group of sine waves

 

this i agree with

[beerandmusic]

as long as people stop responding, i will let it rest and ponder...for future consideration.

[beerandmusic]

1 minute ago, crenca said:

 

I hear you noble beerandmusic!  I too believe in the gods!!  I have seen their shadows on the walls, seen the shapes out of the corners of my eyes.  But, alas, I am but a man and have also seen that a stone is just a stone as well.  When I walk home, after a weary day, the distance is always the same.  The gods, yes yes they are infinite, but the world that they created for us is not arbitrary.  They are also not facetious or cruel - they help us lowly men by creating order and limits.  Our names and the notes that our instruments create, alas, they are finite as well, and can only be manipulated so fast.  Thank the gods!  This means that they can be measured!!  If men were infinite, then men would have no names, and music would have no frequency, just as you point out noble beerandmusic!!!

 

 

don't cut yourself short.

[beerandmusic]

19 minutes ago, crenca said:

 

I hear you noble beerandmusic!  I too believe in the gods!!  I have seen their shadows on the walls, seen the shapes out of the corners of my eyes.  But, alas, I am but a man and have also seen that a stone is just a stone as well.  When I walk home, after a weary day, the distance is always the same.  The gods, yes yes they are infinite, but the world that they created for us is not arbitrary.  They are also not facetious or cruel - they help us lowly men by creating order and limits.  Our names and the notes that our instruments create, alas, they are finite as well, and can only be manipulated so fast.  Thank the gods!  This means that they can be measured!!  If men were infinite, then men would have no names, and music would have no frequency, just as you point out noble beerandmusic!!!

 

P.S. Thanks for spending so much time patiently with me.  Where it may not be apparent that I have gained anything, i believe I have....and the results may not be immediately forthcoming, but in the overall big picture, it has and will continue.  you are a good man, and better than most.  I am not saying this as some may suggest that i am full of myself or that my opinion even matters...i am merely acknowledging your time and effort.

 

[beerandmusic]

8 hours ago, kumakuma said:

 

waveforms do not have an infinite number of frequencies, they have one frequency at any arbitrary point in time

 

 

 

7 hours ago, mansr said:

Wrong. Frequencies don't exist at points in time. Waveforms have a set of frequencies (possibly only one) over an interval of time.

 

if frequencies don't exist at a point in time, what is being measured at sample time?

it may not be a frequency that can be output without time, but it is plotting a point of of the composite frequencies in the complex signal?

 

 

 

[beerandmusic]

3 minutes ago, kumakuma said:

 

Did you watch the video I recommended about how complex waveforms are created from simple waveforms?

I watched about 1 minute of it and was very interesed....i was planning on finishing watching that when i first got up....i should have before starting to read and respond (wink)...will go watch it now, before reading more....

[beerandmusic]

Ok, i watched it...i didn't realize it was only 3 minutes or i would have watched it all at once (grin)..i was hoping it was much longer, but it does show a LOT in just 3 minutes.

 

Looking at 1/100th of a second in the last example, is something worth pondering and how accuracy can be obtained by sampling...but if sampling isn't just taking one sample at just one point, then i to need to understand more....but right now, i don't see how more samples can't create more accuracy, given the complexity of the signal.

 

i agree, i have much more to understand...i will probably just take a little at a time unless i become bored.

 

my current understanding (as dense a many may think i am), is relatively the same.

That increasing sample time does 2 things...

 

1. as related to nyquist, allows the highest frequency range to be measured.  In the example i choose to use of 3 instruments playing from 20 to 3000hz, the sample rate only needs to be 6k (or double the maximum 3K i which to produce)

 

2. i believe unrelated to nyquist (unless someone can share a link that suggest nyquist also determines the accuracy by increased sampling), that a higher sample rate improves accuracy by allowing for more audio data, having NOTHING to do with the highest frequency rate i wish to reproduce.

 

3. but perhaps it is only for pcm technology that 44.1 sample rate is all that is needed, but that DSD technology allows for more accuracy for different reasons than sample rate.

 

 

[beerandmusic]

I think something just clicked in my head.

 

maybe the theorem is suggesting that since only so many transitions can take place during any certain time, that is how the sample rate is calculated, and thus the theorem actually suggests not only a maximum frequency range (which i cared nothing about, that people kept trying to explain), but also that it is the sample time needed to calculate all "measurable" possibilities of how many transitions can take place in a given time.....

 

i still have to ponder more, but if that is the case, then i have to go back to the belief that DSD uses a different technology which allows a more accurate reproduction.

[beerandmusic]

very good

 

 

[beerandmusic]

9 hours ago, mansr said:

Wrong. Frequencies don't exist at points in time. Waveforms have a set of frequencies (possibly only one) over an interval of time.

disregard my last question...

watching this my assumption posed in my last question to you was correct...it is capturing a plot of the amplitude of the composite frequency, which is in agreement with your statement that frequencies don't exist at a point of time, but in essence it is measuring an amplitude plot of the waveform.

 

https://ru-clip.com/video/BNVVq-iVPy8/digital-audio-explained-samplerate-and-bitdepth.html

 

 

[beerandmusic]

9 hours ago, Don Hills said:

 

Nevertheless, provided those frequencies are all less than half of the sampling rate, they can be accurately sampled and reproduced.

 

I am going to attempt to go down this road again with another example in hopes for better understanding.

 

Of course totally hypothetical to extremes to help me better understand....

 

Assume 10000 different tone generators (starting and stopping at different picoseconds) all at different frequencies between 10hz and 10.1hz (e.g. similar to 10.00000001, 10.00000002, 10.0000003, but not necessarily linear in difference etc)

Granted we would not be able to discern differences, but with only a needed 20 samples per second, would we not be able to more accurately capture with a higher sample rate?

Isn't it possible that the 20 samples per second that we capture were amplitudes that corresponded to frequencies 10.000000000021 though 10.00000000040, when in actuality during that same second the composite included those frequencies once, but all the other frequencies occurred more often, and that if we captured 40 samples per second we may have captured plots corresponding to frequencies 10.000000050 through 10.000000089 which was more accurate to reality?

 

 

edit to add...if 10hz is not considered audible, raise it to 20hz with a sample rate of 40 times per second....i only used a low number to help my understanding...

 

 

[beerandmusic]

OK, going to bed....

Guess what...

I accept that nyquist and what most have posed here...but i am on to something else....

 

something to do with higher sampling required for processing and filtering purposes....

 

that's it for now, my wife is making me go watch tv with her.

[beerandmusic]

 

[beerandmusic]

4 minutes ago, audiventory said:

 

Accuracy is defined by analog filter quality. The filter is interpolator of information between samples.

 

OK, my summary of the thread to date.

 

This thread was spun off someone's comment that an SACD will sound the same as a CD.

I did not believe that, and i still do not believe it. 

I admit, I am possibly one of the biggest novices here when it comes to audio engineering.  I really did not know why, nor do I understand now why, but my opinion has not changed, that an SACD can sound better than a CD. 

I may have gone down the wrong path with sampling rate for the wrong reason, but there are some audio engineers that suggest that higher sampling rate does afford a better SQ because of processing and filtering, again which I know nothing about.

 

I also admit that I still do not understand sampling rate, but i believe I now have a "better" understanding of it.

I believe the nyquist theorem within it's criteria.

 

I currently "believe" that higher sampling rate allows for more accuracy and better SQ for other reasons, but that DSD likely sounds better than lower resolution PCM for other reasons, than the path I went down.

 

In other words, I know a little more but still very little (grin), but my opinion that SACD can sound better than CD's has not changed.

 

Anyway, thanks for anyone that was patient with me.

 

[beerandmusic]

9 minutes ago, mansr said:

The composite waveform includes all the frequencies all the time. Sampling at a rate higher than twice the highest frequency is enough to capture them all perfectly. However, with more closely spaced frequencies, we need to sample for a longer duration in order to tell them apart. A single sample tells us nothing about frequency, a few samples tells us a little, and many samples tell us a lot. Now I repeat, because this is important, those samples must be spread over a long duration. Once the sample rate exceeds twice the highest frequency component, adding more samples within a fixed duration does not improve accuracy.

 

Ok, thanks...i understand more and accept this.

[beerandmusic]

3 minutes ago, audiventory said:

 

SACD, CD, high res and other digital formats have no sound quality as itself. They give some abilities to achieve sound quality.

But playback processing and electrical schemes define quality. There is big field for improvements (distortion decreasing).

Example: CD (due lower sample rate and, thus, analog filter issues) have lesser sound quality improvement abilities than SACD.
But, In general case, fine built and ajusted CD player may sound better (lesser distortions) than simpler SACD player.

 

yes, and an mp3 player can sound better than the best cd player if the recording is lousy on the cd.

The point was that it is "possible" for an SACD to sound better than a CD, that the current technology itself is allows for more data, and I read SNR, bit depth, and other things as well.  That any previous obstacles have been overcome.

 

Is it your opinion that CDs "in general" sound better than SACDs "in general" with comparable hardware and recordings?

 

 

[beerandmusic]

8 minutes ago, firedog said:

 No one doubts hi res, or even upsampling to hi res and then using a hi res filter for better filtering is a possible way to improve results.

 

are you sure everyone?  Does mansr believe it?

Does most everyone truly believe that sampling higher than nyquist is a possible way to improve SQ?

I think that is what that video is saying that mansr and elsdude say is a bunch of baloney?

 

[beerandmusic]

4 minutes ago, firedog said:

That’s not what I wrote.

please clarify then what you mean by no one doubts....

 

 

[beerandmusic]

3 minutes ago, Spacehound said:

Only those who don't know how it works. Which is probably the majority.

 

what say you?

I doubt you can hang with the mansrs or jabbrs of the site....but please try to share anything besides a mock or a troll statement.

 

Do you know how it works?  Can you explain the sampling and filtering thing to obtain more accuracy?

Or do you have any input of any substance?

 

[beerandmusic]

13 minutes ago, firedog said:

See above. I was referring to different filtering possibly sounding better. 

you mean by Oversampling right?

 

e.g. from wikipedia: Oversampling improves resolution, reduces noise and helps avoid aliasing and phase distortion by relaxing anti-aliasing filter performance requirements.

 

or some other method?