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 12th September 2011
  #91
Lives for gear
 

Quote:
Originally Posted by narcoman View Post
It's also where the (rare) intersample peaks occur. ... Few samples fall on the peaks of waveforms ... The analogue signal, recreated from the samples, RELIES on intersample peaks!!
They're not rare at all, as you say ...

///

How about we all simply refer to this as the "true peak" of a signal instead.
Old 12th September 2011
  #92
Lives for gear
 

Quote:
Originally Posted by psycho_monkey View Post
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....
no
i am saying they should use proper english
and not invent bogus terms

it is bad enough that
samplers dont sample - they are samples
sequencers dont sequence
controllers dont control - they enter note data to finale or sebelius
yada yada
(at least not today - although they did once way back when)

digital samples have shape they have no voltage value
claiming that there are intersample peaks is bullbleep

the analog signal level depends on the d/a
the shape will match the digital values
but the digital has nothing to do with peaks
(even the invented word ones)
that is strictly a function of teh d/a

stop confusing teh digital and analog domains
you cannot move the sample values from digital
and draw them on the analog in any meaningful way
Old 12th September 2011
  #93
Lives for gear
 

Oh g*d someone lock this thread, it's like a bad merry-go-round
Old 12th September 2011
  #94
Lives for gear
 

Quote:
Originally Posted by duckoff View Post
Best case scenario this whole discussion is down to semantics with some very poor syntax & lots of typos......
some of that
but still confusion between digital and analog domains

digital has no voltage value only waveshape
so peaks are meaningless

the voltage is determined by the d/a
i can design the d/a to be 40dB below the original analog signal that was sampled and sent right back to d/a
where would the peak PROBLEM be then?

there is no problem
just twist the knob to the right and make the recreated analog signal as loud as you want
Old 12th September 2011
  #95
Registered User
Who's butchering the English language here??

In what way do modern samplers not sample, or modern sequencers not sequence?

In fairness - there are some modern software "sample players" that do not record audio, mainly because they are imbedded in a DAW that offers all the audio recording functions you could wish for and more.

But you made a blanket statement that implies that all modern samplers (whether hardware or software) do not sample, and that is clearly untrue. If they take a digital audio recording, that is a file of samples. We talking about modern soft sample players as having 'sample accurate timing', because they do.

And in what way do modern sequencers not sequence? They record midi event data, and play them back in a sequence.

Seems to me that you are just being an ass for the sake of being an ass.

May I politely suggest you loosen your shirt and pull you head in.
Old 12th September 2011
  #96
Registered User
Quote:
Originally Posted by oldeanalogueguy View Post
some of that
but still confusion between digital and analog domains

digital has no voltage value only waveshape
so peaks are meaningless

the voltage is determined by the d/a
i can design the d/a to be 40dB below the original analog signal that was sampled and sent right back to d/a
where would the peak PROBLEM be then?

there is no problem
just twist the knob to the right and make the recreated analog signal as loud as you want
How can you say peaks are meaningless? In a modern DAW you have Peak meters. Totally digital.

The word "peak" implies the highest value. Nobody is confusing digital with analog here.

Sorry the world does not revolve you.
Old 12th September 2011
  #97
Registered User
I can, however, agree with the point I think you are making:

IF there is sufficient headroom built into the analog section of a D/A - the waveform can be built without clipping or distorting, even if the digital values were slightly clipped.

The other thing that I think many people are missing is that VERY often the digital audio stream is lowered by a digital gain reduction ... so there is even more headroom available than it would appear ...
Old 12th September 2011
  #98
Lives for gear
Quote:
Originally Posted by oldeanalogueguy View Post
some of that
but still confusion between digital and analog domains

digital has no voltage value only waveshape
so peaks are meaningless

the voltage is determined by the d/a
i can design the d/a to be 40dB below the original analog signal that was sampled and sent right back to d/a
where would the peak PROBLEM be then?

there is no problem
just twist the knob to the right and make the recreated analog signal as loud as you want
Peaks and clipping are obviously two different things. It seems all you've really said is that if you're using both analog and digital, no signal exists on the other side of the ADC/DAC until you send a signal through it; once you've gone from digital to analog you can turn your monitoring volume up as loud as you want.
Old 12th September 2011
  #99
Lives for gear
 

Quote:
Originally Posted by Kiwi View Post
I can, however, agree with the point I think you are making:

IF there is sufficient headroom built into the analog section of a D/A - the waveform can be built without clipping or distorting, even if the digital values were slightly clipped.

The other thing that I think many people are missing is that VERY often the digital audio stream is lowered by a digital gain reduction ... so there is even more headroom available than it would appear ...

if the digital were clipped then you have an infinite bandwidth in the digital domain. if you do not LP filter the digital before the d/a then you will get foldback aliasing of the digital signal as it is d/a processed. no LP filtering after the initial d/a will get rid of that. those foldbacks will likely line up where the signal got hot (drum hit?) and can add up to a big peak in the analog domain. yes that could clip even with a proper d/a design because the signal sent to it was not nyquist compliant.
Old 12th September 2011
  #100
Lives for gear
 

Quote:
Originally Posted by Kiwi View Post
How can you say peaks are meaningless? In a modern DAW you have Peak meters. Totally digital.

The word "peak" implies the highest value. Nobody is confusing digital with analog here.

Sorry the world does not revolve you.
peak meters in digital domain are only meaningful in the digital domain and only measure the digital peaks.

there is no intersample peak PROBLEM in the analog domain.

peak has two meanings (at least)
highest is one
higher than a reference value is a second
Old 12th September 2011
  #101
Lives for gear
 

Quote:
Originally Posted by Kiwi View Post
Who's butchering the English language here??

In what way do modern samplers not sample, or modern sequencers not sequence?

In fairness - there are some modern software "sample players" that do not record audio, mainly because they are imbedded in a DAW that offers all the audio recording functions you could wish for and more.

But you made a blanket statement that implies that all modern samplers (whether hardware or software) do not sample, and that is clearly untrue. If they take a digital audio recording, that is a file of samples. We talking about modern soft sample players as having 'sample accurate timing', because they do.

And in what way do modern sequencers not sequence? They record midi event data, and play them back in a sequence.

Seems to me that you are just being an ass for the sake of being an ass.

May I politely suggest you loosen your shirt and pull you head in.
all the samplers that i have seen
are either SAMPLES themselves
OR A PLAYER of samples
but none of them let me MAKE samples.

if you know a cheap one for sonar x1 let me know

what sequencer sequences ?
they just play back what you have.
it has already been put in sequence for playing.

you forget to dump on me for controllers dont control.
Old 12th September 2011
  #102
Lives for gear
 
Old Goat's Avatar
 

Isn't that what a sequencer does? Play samples that you have, that you put in sequence?

Ah, ****, there's five minutes I can't get back.
Old 12th September 2011
  #103
Registered User
Quote:
Originally Posted by oldeanalogueguy View Post
all the samplers that i have seen
are either SAMPLES themselves
OR A PLAYER of samples
but none of them let me MAKE samples.

if you know a cheap one for sonar x1 let me know

what sequencer sequences ?
they just play back what you have.
it has already been put in sequence for playing.

you forget to dump on me for controllers dont control.
[dump]You don't MAKE samples - you TAKE samples. And any sampler that records audio is obviously taking samples - aka "sampling".

Gigasampler and EMU Emulator X2 are two soft samplers that actually record audio and therefore are correctly named as samplers. Most other samplers are officially considered to be sample players, but I see no problem calling them samplers for short.

What the hell do you mean by "what sequencer sequences?" They all sequence, Jedi Dumb Arse. If you are recording midi notes, one after another, and playing them back, one after another, that is - by defination - a sequence of notes. You are sequencing with your sequencer ... get over it. Whether recording, or playing back, it's all a sequential stream of notes and data.

And controllers don't control?? Then WTF are you doing when you move the Mod Wheel and the pitch goes down, or you move you Expression pedal and the volume goes up, or press a pedal and a Leslie emulator changes from fast to slow speed ...

In my book, you are most certainly "controlling" stuff using MIDI control data initiated by control devices ...

Feel free to invent your own definition of the English language - it's a living language after all. But don't expect everyone to understand what the hell you are talking about if you don't join in with the common useage of words ...
[/dump]
Old 12th September 2011
  #104
Quote:
Originally Posted by oldeanalogueguy View Post
no
i am saying they should use proper english
and not invent bogus terms

it is bad enough that
samplers dont sample - they are samples
sequencers dont sequence
controllers dont control - they enter note data to finale or sebelius
yada yada
(at least not today - although they did once way back when)

digital samples have shape they have no voltage value
claiming that there are intersample peaks is bullbleep

the analog signal level depends on the d/a
the shape will match the digital values
but the digital has nothing to do with peaks
(even the invented word ones)
that is strictly a function of teh d/a

stop confusing teh digital and analog domains
you cannot move the sample values from digital
and draw them on the analog in any meaningful way
You've had the benefit of digital design engineers trying to explain it to you. You were pointed to a very cogent, straightforward explanation from the makers of SSL consoles and software -- who sort of have a rep for having more than half a clue what's going on. Intersample peaks are a real phenomenon of the DAC process, whatever your feelings about the logic or lack thereof behind their nomenclature.

Somehow I don't think this is a matter of the experts all being wrong and you being right.
Old 12th September 2011
  #105
Gear Guru
 
u b k's Avatar
 

Quote:
Originally Posted by oldeanalogueguy View Post
digital has no voltage value only waveshape
so peaks are meaningless

the voltage is determined by the d/a

I don't agree with this. The d/a creates the voltage, yes, but it does so by decoding the digital information in accordance with a very clear set of established standards.

The voltage level is encoded within the binary word, just as the waveshape is. If this weren't true, we'd never be able to reconstruct the original waveform with any degree of fidelity; but we can, both shape and voltage come out as they went in (unless something is flawed in the design or calibration of the a/d or d/a).



Quote:
Originally Posted by oldeanalogueguy View Post
i can design the d/a to be 40dB below the original analog signal that was sampled and sent right back to d/a
where would the peak PROBLEM be then?

there is no problem

Well, the problem would be that you arbitrarily inserted a -40dB amplifier stage in your d/a, thus completely breaking with standards and ruining any hope I have of good gain staging.

To the matter at hand, maybe I'm missing the point of the discussion, but from my perspective it is both fallacious and counterproductive to blame either analog or digital for intersample peaks and the clipping that results. There is no clipped waveform in the digital word, in fact there is no waveform whatsoever, only binary data stored accurately, so how can this be a digital issue? Likewise, the analog circuit could easily recreate the original waveform if only the word were trimmed back before conversion.

So as I see it, intersample peaks and clipping are a problem that only exists when the worlds of analog and digital meet under real world conditions. They do have to meet somewhere, and without industry standard ref levels we'd all be screwed and cd players would've never been viable, let alone daws and interfaces... wait, maybe that'd be a good thing.

Regardless, no matter where you set the ref level there will still be a ceiling, so this problem is unavoidable, and the blame lies squarely at the border of the two worlds.


Gregory Scott - ubk
Old 12th September 2011
  #106
Gear Head
 

Quote:
Originally Posted by oldeanalogueguy View Post
digital samples have shape they have no voltage value
They carry the information of the voltage relative to the D/A converter playing them back and that's all that matters.

Quote:
Originally Posted by oldeanalogueguy
the voltage is determined by the d/a
i can design the d/a to be 40dB below the original analog signal that was sampled and sent right back to d/a
where would the peak PROBLEM be then?
No you couldn't. Well, maybe you could but your S/N ratio would be terrible and no one would use your converters. Even the best analogue consoles in the world cannot achieve that kind of dynamic range, even the highest end gear tops out at about 124db, or 100db + 24db headroom.

Intersample peaks are normal. However, it seems that you are referring to intersample "clipping" or "overs" as "intersample peaks", and then claim that such clipping does not happen. Well it does. And it has nothing to do with not being "nyquist compliant", whatever that means.

Of course samples dont exist in the analogue world, everyone knows that. The reason "intersample peaks" or "intersample overs" are used to describe the phenomenon is because they describe what will or what has happened to the audio when it is converted from digital numbers to analogue voltage.

True, 0dbfs is a digital measurement and is not necessarily explanatory of the corresponding voltage level in analogue, but to get the best dynamic range out of the converters, 0dbfs has to be pretty darn close to the close to clipping in analogue. [Not to mention gain staging with other equipment]. In the best converters, you are turning 24bit digital streams (144db) over to a 124db analogue circuit. There's not much room to mess around there. If a converter manufacturer leaves 1bit extra for headroom, that still gives you +6db for any overs, which should be plenty in most cases.

If you brickwall limit or clip "in the box" with processing that does not compensate for intersample clipping (which many do) and peak at 0dbfs, there is a good chance - depending on the program material - you will be creating consecutive samples at 0dfs which will correspond to "overs" in the D/A. This is not a problem, as it can be manually compensated for. It is also not a manufacturing fault, since those peaks can theoretically be huge enough to overcome the internal headroom in most converters - again depending on the program material. If you don't understand this by now, I don't think anyone can help you.
Old 12th September 2011
  #107
Moderator
 
narcoman's Avatar
 

Quote:
Originally Posted by timlloyd View Post
They're not rare at all, as you say ...

///

How about we all simply refer to this as the "true peak" of a signal instead.
weird - can't even understand why I put that !!! They're not rare at all.
Old 12th September 2011
  #108
Moderator
 
narcoman's Avatar
 

Quote:
Originally Posted by oldeanalogueguy View Post
peak meters in digital domain are only meaningful in the digital domain and only measure the digital peaks.

there is no intersample peak PROBLEM in the analog domain.
No one is saying it's a problem. They are not represented by value as samples themselves (how could they be?).

Peak meters in the digital domain measure the sinc function reconstruction so do, in fact, measure the peak of the reconstructed wave - in other words they measure the peaks between the samples. It's not a problem but it is a facet. As long as you have a calibrated system the peak meter in the digital domain will give you an exact match for the voltage reading in the analogue domain. They give a warning in reconstruction and may have a problem in the analogue domain as the peaks can be up to 6dB over peak sample value. If you're running into headroom on an analogue system it is possible to run into distortion and further clipping with such an issue. In fact this was a real problem with older CD players built with little headroom.

When you do a "peak" search in PT on a waveform it does NOT go through the file and look for the largest sample value. It looks for the highest "virtual amplitude" which includes "between the samples".
Old 12th September 2011
  #109
Lives for gear
 

Quote:
Originally Posted by narcoman View Post
Peak meters in the digital domain measure the sinc function reconstruction so DO, in fact, measure the peak of the reconstructed wave - in other words they measure the peaks between the samples.
Not all of them can do this unfortunately.
Old 12th September 2011
  #110
Gnu
Here for the gear
 

Ugh. To perhaps drift this closer to the original topic, I don't think it's been mentioned that there are indeed very practical uses for such high-resolution converters, but they have nothing to do with music. Ultrasonic recording is essential in research, particularly in studying animal ranging. 384kHz recording has been in use for some time in the research world. Even if we don't have an ADC to go with that 768k DAC yet, it's still useful and a good start. I don't think many of us are gonna be recording bats and porpoises in our project studios anytime soon, but I hear that dolphins can't do anything without a click track.
Old 12th September 2011
  #111
Quote:
Originally Posted by Gnu View Post
I don't think many of us are gonna be recording bats and porpoises in our project studios anytime soon, but I hear that dolphins can't do anything without a click track.
"what's that Flip? You thought the bass player was a touch early in bar 54, and you'd like to have a bit more of the vocals in your headphones? Atta boy!"
Old 12th September 2011
  #112
Moderator
 
narcoman's Avatar
 

Quote:
Originally Posted by timlloyd View Post
Not all of them can do this unfortunately.
True!!! It's what they should be doing though!!
Old 12th September 2011
  #113
Lives for gear
 

Eventually they will ...
Old 12th September 2011
  #114
Moderator
 
narcoman's Avatar
 

DO you mean eventually as some spacial taylor series as, being digital, they can only ever "approximate".... you know - joining up the dots....heh
Old 12th September 2011
  #115
Lives for gear
 

Yes, but when I manage to get the details of the martin series worked out, I'll give it to all the digi-meter-makers for free, cz then all our metering will sound better. I'm just that nice.
Old 12th September 2011
  #116
Gear Nut
 

Quote:
Originally Posted by duckoff View Post
Best case scenario this whole discussion is down to semantics with some very poor syntax & lots of typos......
Tru Dat!
Old 12th September 2011
  #117
Lives for gear
 
DaveUK's Avatar
Teh?
Old 12th September 2011
  #118
Gear Nut
 

Quote:
Originally Posted by child of Gaia View Post
They carry the information of the voltage relative to the D/A converter playing them back and that's all that matters.
...In the best converters, you are turning 24bit digital streams (144db) over to a 124db analogue circuit. There's not much room to mess around there. If a converter manufacturer leaves 1bit extra for headroom, that still gives you +6db for any overs, which should be plenty in most cases.

If you brickwall limit or clip "in the box" with processing that does not compensate for intersample clipping (which many do) and peak at 0dbfs, there is a good chance - depending on the program material - you will be creating consecutive samples at 0dfs which will correspond to "overs" in the D/A. This is not a problem, as it can be manually compensated for. It is also not a manufacturing fault, since those peaks can theoretically be huge enough to overcome the internal headroom in most converters - again depending on the program material. If you don't understand this by now, I don't think anyone can help you.
Thanks for this COG - I for one did learn something in this thread after all.

Now I'm curious what the maximum possible overage is and how likely it is. s Where'd I put that old slide rule...
Old 12th September 2011
  #119
Lives for gear
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...
You're looking at the wrong end of the spectrum - think higher frequency, not lower. So maybe UV or even X-Rays. I want a gamma ray A/D. In surround sound. And at these prices, the damn popcorn better be free.

And, yes, I'm teasing after a fashion. If we want to pretend that we can equate D/A frequency with the frequency of a photon, then even far IR frequencies are on the order of hundreds of GHz on up to THz. clearly, A/Ds of the future will convert sound to light, and perhaps we will then all see that light.
Old 12th September 2011
  #120
Lives for gear
 
Surbitone's Avatar
Quote:
Originally Posted by oldeanalogueguy View Post
upsampling will not improve anything
bad upsampling is worse than not doing it
I knew you were going to say that. I know.

Also, I didn't mention anywhere about an improvement, you made the assumption that I made that assumption. I find it really hard to believe you're an 'old analogue guy'.
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