The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
if higher sample rate doesnt matter then why .... Effects Pedals, Units & Accessories
Old 9th September 2011
  #61
Lives for gear
 

Quote:
Originally Posted by timlloyd View Post
That's a good summary!

I agree. Most people dont understand else they would not keep claiming that peaks are real.

Their real problem appears to be that they cannot predict the analog values they will get from a digital signal.

But when you are getting back the exact original signal how can their be any peaks? That is just nonsense.

The problem is they dont know what the original signal would have been after they diddled the samples. that is not a peak problem. That is user error.
Old 9th September 2011
  #62
Lives for gear
 

Quote:
Originally Posted by psycho_monkey View Post
oldanalogueguy - half of what you post doesn't make sense (either grammatically or from a technical viewpoint). If you'd like to communicate your thoughts, please do so, but at the moment it's just random references to technical words.

The rest of what you post is factually incorrect. Intersample peaks, as bgrotto points out, ARE fact, they ARE what happens in digital audio, they're NOT necessarily a problem, and they ARE totally irrelevant here!

Likewise, your whole spiel about Nyquist....er, what?! I didn't get one modicum of sense out of your last 5 posts....and my degree was partly in this subject, so don't trot out the " if you don't understand, I can't be bothered to explain" argument - I understand this sh1t, I (like many others it seems) just don't have a clue what you're trying to say!

The more you post, the more it just seems to be that you don't understand digital audio AT ALL! Stick to analogue maybe?! At least talking about it...

Plus the whole 88.2 to 44.1 "simple division" - much cleverer people than either of us have repeatedly stated that this is a myth and with modern day SRC it doesn't matter. IF you have proof to the contrary, please provide it!

Actually, I wonder...there was a guy interviewed in Resolution Magazine ybe April 2010 that swore his method of SRC was more transparent...where to convert 48 to 44.1 he'd upsample to 88.2 then use a bitsplitter...Tony Faulkner was the guy's name...wasn't you was it?! You're obviously not American....
intersample peaks do NOT exist
unless
the d/a was improperly designed
or the digital signal is not a valid nyquist set of samples.

I agree that they are not a problem.
They do not happen except as noted above.

no idea what your other two grafs said.
Old 9th September 2011
  #63
Lives for gear
 

Quote:
Originally Posted by filipv View Post
No matter what sample rates you want to convert from-to, you can always upsample to the lowest common denominator, right?
no.
you cannot upsample without problems.
you do not , and cannot , know what values they should have been. All you can do is add noise, errors, etc.

downsamplign in multiples of 2 are trivial and cause no problems. downrezzing to odd multiples can cause artifacts.
Old 9th September 2011
  #64
Lives for gear
 

Quote:
Originally Posted by oldeanalogueguy View Post
I agree. Most people dont understand else they would not keep claiming that peaks are real.

Their real problem appears to be that they cannot predict the analog values they will get from a digital signal.

But when you are getting back the exact original signal how can their be any peaks? That is just nonsense.

The problem is they dont know what the original signal would have been after they diddled the samples. that is not a peak problem. That is user error.
Digital prediction of the analogue voltage peaks that will result at the output of a d/a just requires up-sampling (or interpolation). Not it's not 100% accurate, that's not under discussion currently.

You insist on using the word "peak" in an irritatingly ambiguous way, reminding me of some of your previous posts. "Peak" in audio just means the maximum deviation from a 0 value. It doesn't mean that distortion has occurred.

I don't know what you mean when you keep saying things like "diddled the samples". Please explain?

You seem to be saying that intersample "clipping" is nonsense. It's not nonsense, it can happen, just not very often.

Intersample "peaks" most certainly are not nonsense, as anyone with a rudimentary understanding of sampling theorem will know. You seem to have that, which is one reason why your posts are confusing ...

So, "peak" = maximum positive or negative value of a bipolar signal

... are you still going to argue that a reconstructed analogue signal cannot have higher peak amplitude than the digital samples which were used to reconstruct it from?



What's going to happen when that signal is reconstructed? The peak level will exceed the max value of the samples. That will happen between two sample values ... "inter", originally from Latin, means (among other subtly different things) "between" ...

Last edited by -tc-; 9th September 2011 at 11:51 PM.. Reason: clarity ... and spelling
Old 9th September 2011
  #65
Quote:
Originally Posted by oldeanalogueguy View Post
intersample peaks do NOT exist
unless
the d/a was improperly designed
or the digital signal is not a valid nyquist set of samples.

I agree that they are not a problem.
They do not happen except as noted above.

no idea what your other two grafs said.
I think Tim's probably beat me to it but:

- intersample peaks are very much real. They happen when you have 2 points at 0dBFs but there is a peak between them. This much is basic digital audio theory, and if you don't agree then time to get reading, because you really don't understand the basics.

- however, the second part of your argument IS valid. A good DA will still reconstruct the waveform correctly. A poor one won't. But since there are many poor DAs around (I hear the kick clipping on the Andy Wallace-mixed "The Puzzle" by Biffy Clyro for example, when listening on headphones and iPod. THAT is intersample peaks in action, with a poor DA) we have to make allowances for it.

As for my other 2 "grafs", well now you know how we all feel about your posts! I don't know which 2 out of the 6 you're referring to -the one where I asked if that was your identity, the one where I said no-one understands what you're trying to say, the one where I asked you to back up your outdated 88.2-44.1 myth...? Please elaborate and I'll try to clarify your confusion!

I find you a total enigma...you've obviously got some knowledge somewhere, but it doesn't ever seem to make it into the online world with any sort of clarity as to what you're trying to say!
Old 9th September 2011
  #66
Lives for gear
 

Quote:
Originally Posted by psycho_monkey View Post
They happen when you have 2 points at 0dBFs but there is a peak between them.
Not only then; every time there is not a sample point taken at the peak amplitude of the input waveform, there will be an "intersample peak" which is higher than the two samples immediately before and after it.

Correct me if I'm wrong, but the phrase describes this occurrence even if the result doesn't exceed the calibrated 0dB point, in which case it could be more accurately called a "potential intersample clip".
Old 10th September 2011
  #67
Quote:
Originally Posted by timlloyd View Post
Not only then; every time there is not a sample point taken at the peak amplitude of the input waveform, there will be an "intersample peak" which is higher than the two samples immediately before and after it.

Correct me if I'm wrong, but the phrase describes this occurrence even if the result doesn't exceed the calibrated 0dB point, in which case it could be more accurately called a "potential intersample clip".
Of course, bad explanation monkey; I was only talking about the ones which might cause problems of course!

"bad explanation monkey"...is that like a bad angel that sits on your shoulder and writes analogies that don't work, like the godawful horse-car hybrid on the Hd native thread?!
Old 11th September 2011
  #68
Lives for gear
 

Quote:
Originally Posted by psycho_monkey View Post
I think Tim's probably beat me to it but:

- intersample peaks are very much real. They happen when you have 2 points at 0dBFs but there is a peak between them. This much is basic digital audio theory, and if you don't agree then time to get reading, because you really don't understand the basics.

- however, the second part of your argument IS valid. A good DA will still reconstruct the waveform correctly. A poor one won't. But since there are many poor DAs around (I hear the kick clipping on the Andy Wallace-mixed "The Puzzle" by Biffy Clyro for example, when listening on headphones and iPod. THAT is intersample peaks in action, with a poor DA) we have to make allowances for it.

As for my other 2 "grafs", well now you know how we all feel about your posts! I don't know which 2 out of the 6 you're referring to -the one where I asked if that was your identity, the one where I said no-one understands what you're trying to say, the one where I asked you to back up your outdated 88.2-44.1 myth...? Please elaborate and I'll try to clarify your confusion!

I find you a total enigma...you've obviously got some knowledge somewhere, but it doesn't ever seem to make it into the online world with any sort of clarity as to what you're trying to say!

intersample peaks are a myth
people confuse digital with analog

there is NOTHING between digital the samples value.
NOTHING

when you convert to analog
back from a signal you just sampled
of course teh analog is higher than the digital samples
in most places in along the continuous analog domain
however
at the sample points it is exactly the same.
THERE IS NO PEAKING

that is nyquist in action
you get back the original signal

the only way not to have " peaks" by that thinking is for every sample to be at the maximum value of 0dBFS

THE REAL PROBLEM is that people want to predict the analog result while still in the digital domain. cant be done.
they play games and futz up approximations and then claim to have a peak problem. they have a dont understand nyquist and fourier problem while confusing digital and analog.

the problem is compounded by people drawing digital domain values on top of analog signals when they make pictures to "prove" their misconception.

worst case example
sample sinx at 6x rate and be unlucky enough to sample at degrees 0 60 120 180 240 300 360 (zero of next cycle) yada yada

you get digital values of 0 .866 .866 0 -.866 -.866 and repeats for following cycles

between the samples which ARE RECREATED EXACTLY WITH ZERO PEAKS OR DIPs the analog sin function does its thing and MAXES at +/- 1 just like the original. THERE IS NO PEAK WHEN YOU GET EXACTLY WHAT YOU AHD BEFORE.

A proper D/A circuit would recreate teh 0dBFS at a level at least 1.25 dB below 1 volt. Then you could add analog gain/faders to make taht signal as big (or small) as you want.

But if you kept your digital at least -6dBFS or lower you would not have any d/a problems anyway with any design. and you could still twist that analog gain up to make your ears bleed.

so where does that extra alleged peaking come from?

promise to stop confusing analog and digital and i will tell you.
(actually i have told you several times but nobody reads what i say they just parrot the internet wizdumb as if their religion had been attacked.)
Old 11th September 2011
  #69
Lives for gear
 

Quote:
Originally Posted by timlloyd View Post
Digital prediction of the analogue voltage peaks that will result at the output of a d/a just requires up-sampling (or interpolation). Not it's not 100% accurate, that's not under discussion currently.

You insist on using the word "peak" in an irritatingly ambiguous way, reminding me of some of your previous posts. "Peak" in audio just means the maximum deviation from a 0 value. It doesn't mean that distortion has occurred.

I don't know what you mean when you keep saying things like "diddled the samples". Please explain?

You seem to be saying that intersample "clipping" is nonsense. It's not nonsense, it can happen, just not very often.

Intersample "peaks" most certainly are not nonsense, as anyone with a rudimentary understanding of sampling theorem will know. You seem to have that, which is one reason why your posts are confusing ...

So, "peak" = maximum positive or negative value of a bipolar signal

... are you still going to argue that a reconstructed analogue signal cannot have higher peak amplitude than the digital samples which were used to reconstruct it from?



What's going to happen when that signal is reconstructed? The peak level will exceed the max value of the samples. That will happen between two sample values ... "inter", originally from Latin, means (among other subtly different things) "between" ...
now we are getting somewhere

the real problem is that you cannot predict analog values from the digital domain. you have to do the real d/a to find uot.

up samplign or interpolation introduces significant error.
cant be done good enough.

peak to me means the deviation upward from the orignal signal due to some mysterious peaks that got added somehow somewehre supposedly between the digital sample values.

I never said that distortion occurred.
what i said was taht non linear processing in the digital domain can create an infinite bandwidth signal in the digital domain that unless you low pass filter in the digital domain first that when you send it to the analog domain via d/a the higher freqs in the digital domain do get folded back during the creation of the analog signal and that can cause peaking over what you expected.

define:
diddled:
the use of non linear processing, or otherwise creating a non band limited signa, and hence not nyquist compliant signal in the digital domain.

i never said there was intersample CLIPPING
i am responding to the alleged intersample PEAK problem

peak can have two meanings
peak = the maximum value of an original signal
peak also means the deviation from that signal to larger values due to the a/d/a process and is caused by the digital not being nyquist compliant

i am saying that this is exactly nyquist in action and is not any alleged peak problem: .. are you still going to argue that a reconstructed analogue signal cannot have higher peak amplitude than the digital samples which were used to reconstruct it from?
of course the values between the samples can be higher and usually are. that is what nyquist guaranteed. you get back teh origianl signal. you dont flatten the signal to avoid alleged peaks.

nyquist is not a problem
exact recreation of a signal is not a problem
alleged peaks are not a problem

the problem is people who confuse digital domain with analog domain. and those who try to predict analog values in the digital domain without doing the d/a to see what it really is. so when their wild assed guesses are wrong they blame some alleged peak problem. sorry but its stupid user error.

rotflamo
there are no peaks being generated in that diagram
the digital values always are the same
the analog values will exactly match those digital samples
of course the analog values will generally be higher than the values at the reconstructed points matching the digital sample values. that is nyquist in action. THERE IS NO PROBLEM !!

unless you think that getting the right answer is a problem.
what would you want to see in the analog domain to say that everything is fine and there is no problem?

the problem is either
bad d/a design ( see other msg today)
or someone diddled the data in the digital domain
( also see other msg today)

THERE IS NO DIGITAL INTERSAMPLE PEAK PROBLEM
AND THERE ARE NO SAMPLES IN THE ANLOG DOMAIN
JUST THE PROPER ANALOG SIGNAL
Old 11th September 2011
  #70
Lives for gear
 
Arksun's Avatar
Quote:
Originally Posted by oldeanalogueguy View Post
intersample peaks are a myth
people confuse digital with analog

there is NOTHING between digital the samples value.
NOTHING

when you convert to analog
back from a signal you just sampled
of course teh analog is higher than the digital samples
in most places in along the continuous analog domain
however
at the sample points it is exactly the same.
THERE IS NO PEAKING

that is nyquist in action
you get back the original signal

the only way not to have " peaks" by that thinking is for every sample to be at the maximum value of 0dBFS

THE REAL PROBLEM is that people want to predict the analog result while still in the digital domain. cant be done.
they play games and futz up approximations and then claim to have a peak problem. they have a dont understand nyquist and fourier problem while confusing digital and analog.
It seems this entire time in the thread it is you that has confused the meaning of the words "Intersample Peaks" to mean digital in other peoples minds, which we know has always ALWAYS refered to the peaks that occur after conversion. But for some oddly bizarre reason you thought everyones take of intersample peaks was peaks that existed in the digital realm like some kinda floating magic point between the real points.

Wasn't helped when you wrote that and i quote "intersample peaks are non-existant" , of course they exist, after conversion to analog. Perhaps if you'd clarified in that first post saying "intersample peaks in the digital only realm don't exist" this whole silly argument wouldn't have started up in the first place!.

I do however disagree regarding sample rate conversion. With todays modern quality sample rate convertors, a straight division from 88 to 44 doesn't yield any superior conversion compared to say 96 to 44. The initial upsampling isn't the issue as its a perfect multiple to divide back down again and the noise levels stay well well below hearing thresholds (-150db and below), its the quality of the filters themselves that have the most dramatic impact on how it sounds, the position and sharpness of slope and how that affects any foldback, alias, pre/post ringing amount etc etc.
Old 11th September 2011
  #71
Lives for gear
 

Quote:
Originally Posted by Arksun View Post
It seems this entire time in the thread it is you that has confused the meaning of the words "Intersample Peaks" to mean digital in other peoples minds, which we know has always ALWAYS refered to the peaks that occur after conversion. But for some oddly bizarre reason you thought everyones take of intersample peaks was peaks that existed in the digital realm like some kinda floating magic point between the real points.

Wasn't helped when you wrote that and i quote "intersample peaks are non-existant" , of course they exist, after conversion to analog. Perhaps if you'd clarified in that first post saying "intersample peaks in the digital only realm don't exist" this whole silly argument wouldn't have started up in the first place!.

I do however disagree regarding sample rate conversion. With todays modern quality sample rate convertors, a straight division from 88 to 44 doesn't yield any superior conversion compared to say 96 to 44. The initial upsampling isn't the issue as its a perfect multiple to divide back down again and the noise levels stay well well below hearing thresholds (-150db and below), its the quality of the filters themselves that have the most dramatic impact on how it sounds, the position and sharpness of slope and how that affects any foldback, alias, pre/post ringing amount etc etc.
there ARE NO SAMPLES IN THE ANALOG DOMAIN.
so there are NO INTERSAMPLE PEAKS IN THE ANLOG DOMAIN.

I repeat:
HOW CAN GETTING THE EXACT SIGNAL IN THE ANALOG DOMAIN BE A PEAK OF ANY SORT. IT IS EXACTLY THE SIGNAL.
IT IS NOT PEAKED, IT IS NOT DROOPED, IT IS THE REAL THING.

THE CONFUSION COMES FROM YOU ALL CONFUSING THE DIGITAL DOMAIN WITH ANALOG DOMAIN.
Old 11th September 2011
  #72
Lives for gear
 

OAG, I've literally just woken up. I'm going to watch the rubgy and then try to understand your post again, because I'm not really understanding you after a single read-through

I shall be back
Old 11th September 2011
  #73
Lives for gear
 

Quote:
Originally Posted by oldeanalogueguy View Post
THE CONFUSION COMES FROM YOU ALL CONFUSING THE DIGITAL DOMAIN WITH ANALOG DOMAIN.
Just a prologue ... nobody here is confusing the digital and analogue domains. However, there is a lot of accidental miscommunication ... if we clear all of that up, I wouldn't be surprised if the disagreements vanish.

... back to the rugby!
Old 11th September 2011
  #74
Lives for gear
 
Red Black's Avatar
Hmm . . I just saw a bunch of messages I'd missed (trying to follow from my email) and thought I should just remove my comments . . I was just adding to the momentum of the merry-go-round :]

Good luck!

Last edited by Red Black; 11th September 2011 at 04:09 PM.. Reason: merry-go-round
Old 11th September 2011
  #75
Lives for gear
 
Arksun's Avatar
Quote:
Originally Posted by oldeanalogueguy View Post
there ARE NO SAMPLES IN THE ANALOG DOMAIN.
so there are NO INTERSAMPLE PEAKS IN THE ANLOG DOMAIN.

I repeat:
HOW CAN GETTING THE EXACT SIGNAL IN THE ANALOG DOMAIN BE A PEAK OF ANY SORT. IT IS EXACTLY THE SIGNAL.
IT IS NOT PEAKED, IT IS NOT DROOPED, IT IS THE REAL THING.

THE CONFUSION COMES FROM YOU ALL CONFUSING THE DIGITAL DOMAIN WITH ANALOG DOMAIN.
Firstly, no need to shout. Secondly, is English not your native language?.

What do you think intersample peaks means, it means quite literally, when a peak of a waveform exists between samples, thats what the 'inter-sample' part means, it does NOT mean there is an intermediate point which is a sample.

So once again you are arguing against a viewpoint that only exists in your head.

We all happen to be on the same page here dude, its just that for some reason you think everyone else is on a different page
Old 11th September 2011
  #76
Lives for gear
 
chrisdee's Avatar
Quote:
Originally Posted by ArnieInTheSky View Post
768KHz sample rate, that'll take a snap shot of 384KHz. What note is that? Does it sound to my ears as infrared looks to my eyes? I don't see it... but it is there...
Perhaps our descendants one day will be able to hear up to 768 kHz.
Think about how glad they will be to find that we thought of them all those years ago.heh
Old 11th September 2011
  #77
Gear Nut
 

/delurking...

IMHO, this thread is full of misused vernacular and mixed metaphors to dive as deep as possible into remote scenarios.

Sure, "Intersample Peak" in the analog world *is* a meaningless term. However, we are talking about this in the context of a DAC. A DAC straddles the line - there's no confusing analog and digital domains going on when we are talking about the nexus of the domains. In the context of a DAC, digital and analog are not "separate" until the hit the next opamp or cathode. Er, I mean coupling cap.

Intersample peaks happen ALL the time - um, every second peak, more or less. Honest question - if the reconstruction filters' analog peak causes clipping on the DAC output, isn't the DAC is just a bad DAC with too little internal headroom?

If on the rare occasion we do encounter it with your ears, most of us instinctively go for a quick nudge of the fader and the problem is gone. We don't even think about it. Comparing that one little fader touch to the amount of typing here is kinda funny!

I should prolly reread this thread, but I really don't want to....

Dave
Old 11th September 2011
  #78
Gear Guru
 
u b k's Avatar
 

Questions for no one in particular...

On a scale of 1 to 100, how much awesomeness do your current recordings embody?

How much more awesomeness would they embody if you could suddenly record at 10x the sample rate and nothing else changed?

How much more awesomeness would they embody if you spent a week recording and mixing with another engineer who is at least as good as you?

Forest, meet trees.


Gregory Scott - ubk
Old 11th September 2011
  #79
Lives for gear
 

Quote:
Originally Posted by DaveDaveDave View Post
I should prolly reread this thread, but I really don't want to....
So should I, but as you say ... not going to follow up on that clarification either, I thought I could, but I can no longer be bothered to reiterate.

I'm done with wandering around the forest for a little while ... need to get out into the open where I can move more quickly ...
Old 11th September 2011
  #80
Lives for gear
 

Quote:
Originally Posted by u b k View Post
Questions for no one in particular...

On a scale of 1 to 100, how much awesomeness do your current recordings embody?

How much more awesomeness would they embody if you could suddenly record at 10x the sample rate and nothing else changed?

How much more awesomeness would they embody if you spent a week recording and mixing with another engineer who is at least as good as you?

Forest, meet trees.


Gregory Scott - ubk
two different issues!

the quality of my awesomeness or lack of it would certainly sound better with higher sample rates. i use the max rates (192/32) that the gear that i can afford can work at. storage is cheap. 3TB was on sale this week for about $100.

not being a golden eared green magic marker using stereophile , my budget would probably make me content to stay at those sample rates even if there were higher rates that i *might* be able to buy gear to do. but they would still sound better if i did spring the money for higher rates.

i am sure the other engineer could learn something if i worked with them

as you imply, or at least i infer, people here would be far better off forgetting about gear and worrying more about performance and how well they use the gear they have.
Old 11th September 2011
  #81
Lives for gear
 

Quote:
Originally Posted by Arksun View Post
Firstly, no need to shout. Secondly, is English not your native language?.

What do you think intersample peaks means, it means quite literally, when a peak of a waveform exists between samples, thats what the 'inter-sample' part means, it does NOT mean there is an intermediate point which is a sample.

So once again you are arguing against a viewpoint that only exists in your head.

We all happen to be on the same page here dude, its just that for some reason you think everyone else is on a different page
clearly intersample peaks means peaks between samples
samples are in the digital domain
there is nothing between them
so they cannot peak

there are no samples in the analog domain
so no samples means no peaks between samples there either

you are still confusing the two domains

in the analog domain i have the exact proper signal
nothing has peaked
it is reproduced exactly as it should be

the fact that you cannot predict that signal while
in the digital domain does not mean there are peaks
it means you are confuse and not using the gear or theory correctly
Old 11th September 2011
  #82
Lives for gear
 

Quote:
Originally Posted by chrisdee View Post
Perhaps our descendants one day will be able to hear up to 768 kHz.
Think about how glad they will be to find that we thought of them all those years ago.heh
nothing to do with hearing that high
it has to do with hearing the audio better with more fi
Old 11th September 2011
  #83
Lives for gear
 

Quote:
Originally Posted by DaveDaveDave View Post
/delurking...

IMHO, this thread is full of misused vernacular and mixed metaphors to dive as deep as possible into remote scenarios.

Sure, "Intersample Peak" in the analog world *is* a meaningless term. However, we are talking about this in the context of a DAC. A DAC straddles the line - there's no confusing analog and digital domains going on when we are talking about the nexus of the domains. In the context of a DAC, digital and analog are not "separate" until the hit the next opamp or cathode. Er, I mean coupling cap.

Intersample peaks happen ALL the time - um, every second peak, more or less. Honest question - if the reconstruction filters' analog peak causes clipping on the DAC output, isn't the DAC is just a bad DAC with too little internal headroom?

If on the rare occasion we do encounter it with your ears, most of us instinctively go for a quick nudge of the fader and the problem is gone. We don't even think about it. Comparing that one little fader touch to the amount of typing here is kinda funny!

I should prolly reread this thread, but I really don't want to....

Dave
exactly
the people talking about intersample peaks confuse the digital and analog domains and befuddle themselves by making up nonsense terms aka vernacular

on output consider:
on the left is the digital domain
on the right is the analog domain
between them and keeping them separate is the d/a circuit

there is no voltage level, no peaks, no nothing in the analog domain until the d/a makes it happen.

a proper d/a will not cause "peaks" it gives you the exact same signal you should have. no filters cause any clipping.
(where do these crazy ideas some people throw in come from anyway?)

a proper d/a will give you a lower voltage so you will not "peak" except if you make invalid assumptions and draw the digital sample values in the output analog domain. the output analog can be , and should be, designed to be below the original input analog sample peak values that are sent to the digital domain by the a/d. you can then amplify that perfect unpeaked signal to be as loud as you want.

so not only are there no peaks of any kind on the output analog signal (unless the d/a is badly designtd) they are below(or equal) the original digital peaks which were below the original input analog signal peaks.
Old 11th September 2011
  #84
Quote:
Originally Posted by oldeanalogueguy View Post
clearly intersample peaks means peaks between samples
samples are in the digital domain
there is nothing between them
so they cannot peak

there are no samples in the analog domain
so no samples means no peaks between samples there either

you are still confusing the two domains

in the analog domain i have the exact proper signal
nothing has peaked
it is reproduced exactly as it should be

the fact that you cannot predict that signal while
in the digital domain does not mean there are peaks
it means you are confuse and not using the gear or theory correctly
I believe your understanding would be well served by a careful reading and absorption of the information in the explainer article on this page: Solid State Logic | Music

Quote:
When a digitally-recorded sound is played-back it must be converted into an analogue representation at some point. This is performed by the DAC (digital to analogue converter). Commonly used over-sampling DACs apply a ‘reconstruction’ filter to convert the periodic sample levels back into a discrete signal. In simple terms they ‘fill in the gaps’ using interpolation so that digital artefacts are reduced to a minimum.

Regardless of the reconstruction filter’s design, in certain circumstances it is possible for a reconstruction filter to produce signals greater than 0dBfs even if the samples used to create these signals are all within the 0dBfs limit.
There's more and it would probably be well worth your time absorbing it.
Old 11th September 2011
  #85
Gear Nut
 

duplicate - fixed image url tags...

Last edited by DaveDaveDave; 11th September 2011 at 09:01 PM.. Reason: duplicate - fixed image url tags...
Old 11th September 2011
  #86
Gear Nut
 

Quote:
Originally Posted by oldeanalogueguy View Post
exactly
the people talking about intersample peaks confuse the digital and analog domains and befuddle themselves by making up nonsense terms aka vernacular
Sure, but "intersample peaks" is a real term, and is commonly understood.

Intersample peak:




another:


another 13 intersample peaks:


Vernacular is how most words are born.

Quote:
there is no voltage level, no peaks, no nothing in the analog domain until the d/a makes it happen.
paraphrased: DAC makes voltage and peaks, er.. signal. Yes, we agree on something

60 or so peaks:


Quote:
a proper d/a will not cause "peaks" it gives you the exact same signal you should have.
A proper D/A WILL cause peaks - because without peaks there is no signal, no sound, just D.C. I can't hear D.C.

I don't mean to give you a hard time, but a signal is a series of waveforms, which by definition has peaks. No quotes. Peak as in pi/2 or 3pi/2. As in lim(dX/dY)=>inf.
Therefor, I don't really understand what you're talking about when you say peak. Putting a word in quotes doesn't change it's meaning.

Quote:
(where do these crazy ideas some people throw in come from anyway?)
A nice mix of understanding, misunderstanding, a little something-to-prove and sometimes POOR USE OF VOCABULARY! I do all of the above it all the time, I just try not to post it out loud Feel free to ding me on it if you catch me.

Dave
Old 11th September 2011
  #87
Lives for gear
 

Quote:
Originally Posted by DaveDaveDave View Post
Feel free to ding me on it if you catch me.
Do the same for me please
Old 11th September 2011
  #88
Quote:
Originally Posted by oldeanalogueguy View Post
exactly
the people talking about intersample peaks confuse the digital and analog domains and befuddle themselves by making up nonsense terms aka vernacular

on output consider:
on the left is the digital domain
on the right is the analog domain
between them and keeping them separate is the d/a circuit

there is no voltage level, no peaks, no nothing in the analog domain until the d/a makes it happen.

a proper d/a will not cause "peaks" it gives you the exact same signal you should have. no filters cause any clipping.
(where do these crazy ideas some people throw in come from anyway?)

a proper d/a will give you a lower voltage so you will not "peak" except if you make invalid assumptions and draw the digital sample values in the output analog domain. the output analog can be , and should be, designed to be below the original input analog sample peak values that are sent to the digital domain by the a/d. you can then amplify that perfect unpeaked signal to be as loud as you want.

so not only are there no peaks of any kind on the output analog signal (unless the d/a is badly designtd) they are below(or equal) the original digital peaks which were below the original input analog signal peaks.
So basically, what you're saying is the entire world of digital audio should reconfigure itself to match your terminology?

No one is confusing anything. I see your point, but in your description the problem comes in the DA. Intersample peaks may not actually "exist" at any point, but it's a convenient way to describe things.

If that's not good enough for you, stick to analogue! Everyone else understands the terminology fine....
Old 12th September 2011
  #89
Lives for gear
 
duckoff's Avatar
 

Best case scenario this whole discussion is down to semantics with some very poor syntax & lots of typos......
Old 12th September 2011
  #90
Moderator
 
narcoman's Avatar
 

Quote:
Originally Posted by oldeanalogueguy View Post
now we are getting somewhere

the real problem is that you cannot predict analog values from the digital domain. you have to do the real d/a to find uot.
yes you can. It's the sinc(x) function. It's description of points on the sine curves STILL exist when described by discrete samples. It's how SRC is done without converting back to analogue (it is never divide by two - never has been and never will be). The sinc filter can tell us the value (amplitude) of the waveform at ANY time (even between samples). This is mathematically the same as reconstructing the sample points as an analogue waveform.

That's what a DA functions as. It's also where the (rare) intersample peaks occur. They aren't a problem, but they are a facet of sampling. Few samples fall on the peaks of waveforms and maintaining those peaks is vital to Nyquist-Shannon. That's what the maths proves - the information isn't lost. The analogue signal, recreated from the samples, RELIES on intersample peaks!! The discontinuities between a discrete sampled waveform and it's continuous analogue counter part are not lost when sampled. They are still 100% reconstructable within the bandlimited boundaries of the sample rate.
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump