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 11th October 2011
  #841
Lives for gear
 

Quote:
Originally Posted by narcoman View Post
Ah - so you're talking about something you've built.

And I'm guessing you've built it that way because you have some tech knowledge that leads you to the conclusion that your system samples "better".

I'm okay with that even though I don't support it. The reason all the major manufacturers make higher sample rate systems is to move the brickwall filters out of the hearing range and to increase the bandwidth (for reasons I don't think are worth it).



I've heard of no-one who agrees with this but if it's something you posit then fair enough but it isn't supported by any current texts to my knowledge - again, if you know any then it'd be good to share.
things i have built/designed as well as things i use... Dont know current texts.... but certainly in the old texts and old engineers who built this stuff for control system use
Old 11th October 2011
  #842
Lives for gear
 
dcollins's Avatar
Quote:
Originally Posted by oldeanalogueguy View Post
nonsense. if you knew engineering as applied to nyquist you would know why higher sample rates result in better a/d/a quality
Did you really go to college?

DC
Old 11th October 2011
  #843
Moderator
 
narcoman's Avatar
 

..and on an unrelated note - I found a very young kitten last night under my car. In the middle of no-where.

Took it home.... any cat experts?
Old 11th October 2011
  #844
Lives for gear
 

Don't feed it cow's milk. Most cats are lactose intolerant and it could cause diarrhea - could potentially die of that depending on how young it is. If it's less than a month or so you'll need special kitten milk.
Old 11th October 2011
  #845
Lives for gear
 
stinkyfingers's Avatar
 

Old 11th October 2011
  #846
Lives for gear
 

heh


/Peter
Old 11th October 2011
  #847
Quote:
Originally Posted by narcoman View Post
..and on an unrelated note - I found a very young kitten last night under my car. In the middle of no-where.

Took it home.... any cat experts?
Call the cats protection league for advice!

But studios and cats go together very well...
Old 11th October 2011
  #848
Lives for gear
 

If this cat is still with the CPL, you'll be in good hands ... paws ...

Old 11th October 2011
  #849
Moderator
 
narcoman's Avatar
 

It's about 4 or 5 weeks old. Bought kitten milk, but it preferred water. Seems much happier than yesterday now!!!

Business partner has loads of cats so all in good hands....
Old 11th October 2011
  #850
Lives for gear
 
Bristol_Jonesey's Avatar
Quote:
Originally Posted by narcoman View Post
..and on an unrelated note - I found a very young kitten last night under my car. In the middle of no-where.

Took it home.... any cat experts?
Agree about not feeding it milk - cats like water - just like us.
Old 11th October 2011
  #851
Lives for gear
 
doug hazelrigg's Avatar
Quote:
Originally Posted by swim View Post
Some mis-confusionage among even some of you "highly educated" debaters.

You do effectively gain about 1 bit for every 4x increase in sample rate. That is a fact. Even Lavry conceded that much.

Also, contrary to what some have indicated, increased bit depth does not affect only the accuracy of low amplitude information, it provides increased "resolution" / accuracy / precision at all signal amplitudes.

Note that acquiring individual samples with greater accuracy in the first instance is better than trying to use dither to achieve a statistical effect over time.

For the most part, you guys also seem to be ignoring the fact that modern quantizers are already sampling in the megahertz range, and then decimating.

Some also seem to be under the impression that samplers are actually running each sample through the exact mathematical proof of the sampling theorum to obtain each sample's sinc function, and that the sampling theorum's requirements are somehow being truly met by real world hardware [they are not].
I don't disagree with anything you've written above
Old 11th October 2011
  #852
Lives for gear
 
doug hazelrigg's Avatar
Quote:
Originally Posted by oldeanalogueguy View Post
nonsense. if you knew engineering as applied to nyquist you would know why higher sample rates result in better a/d/a quality
Uh, true, but NOT due to the sampling theorem per se. Again, using higher sample rates does yield better quality because

>it allows gentler filtering
>it allows more accurate measurement of peak values

The sampling theorem says NOTHING, ZERO, about aliasing or transition band filtering. So the phrase "engineering as applied to nyquist" is quite silly. Neither Nyquist nor Shannon envisioned digital audio, you know. But you still seem to be saying that using fs>2B yields better "quality" in the audible range. It doesn't. In fact, the Nyquist theorem itself, which you continually cite, proves this. Conclusively. tutt

Ignorance is one thing, and nothing to be ashamed of, but to continually expose it so brazenly is appalling
Old 11th October 2011
  #853
Lives for gear
 

Quote:
Originally Posted by doug hazelrigg View Post
Again, using higher sample rates does yield better quality because

>it allows gentler filtering
Up to a point, yes. Upping it even more does nothing to the audible range.

Quote:
>it allows more accurate measurement of peak values
Not for frequencies below Nyquist.


/Peter
Old 11th October 2011
  #854
Lives for gear
 

Quote:
Originally Posted by swim View Post
Some mis-confusionage among even some of you "highly educated" debaters.

You do effectively gain about 1 bit for every 4x increase in sample rate. That is a fact. Even Lavry conceded that much.
In practical modern PCM audio, where do I see that 1 bit gain?

Quote:
Also, contrary to what some have indicated, increased bit depth does not affect only the accuracy of low amplitude information, it provides increased "resolution" / accuracy / precision at all signal amplitudes.
Increased bit depth provides a lower noise floor. When that noise floor is inaudible no further audible gain is possible. In modern converters the inaudible analog noise floor dominates.

Quote:
Note that acquiring individual samples with greater accuracy in the first instance is better than trying to use dither to achieve a statistical effect over time.
Don't know what you're trying to say here, that non dithered 24bit is better than dithered 16bit?


Theorem, not theorum!


/Peter
Old 11th October 2011
  #855
Lives for gear
 
doug hazelrigg's Avatar
Quote:
Originally Posted by Audiop View Post
Up to a point, yes. Upping it even more does nothing to the audible range.
Agreed.

Quote:

Not for frequencies below Nyquist.


/Peter
Disagree. Somewhere back in this thread I posted the potential error at various audible frequencies for several different sample rates.
Old 11th October 2011
  #856
Lives for gear
 
doug hazelrigg's Avatar
Now I'm wondering if ultra-high sample rates used in engineering and medicine are also an example of marketing BS... manufacturers and hospitals have deep pockets and critical requirements that might be easy prey for exploitation!
Old 11th October 2011
  #857
Lives for gear
 
dcollins's Avatar
Quote:
Originally Posted by doug hazelrigg View Post
Now I'm wondering if ultra-high sample rates used in engineering and medicine are also an example of marketing BS... manufacturers and hospitals have deep pockets and critical requirements that might be easy prey for exploitation!
Unlikely, as those industries actually have to show their work.

It depends on the signals being analyzed. Maybe it's ultrasound, or something.

More dots does equal a better 'scope display though, as you usually don't (but can) view the waveform through a reconstruction filter.

DC
Old 11th October 2011
  #858
Lives for gear
 

There's no deep pockets in medical in most of the world (in Sweden many people die becasue of lack of money in the health care system) and as mentioned, they have to show that the stuff works, not only claim it does.

When it comes to medicine/substances controlled double blind test are standard for a reason.


/Peter
Old 11th October 2011
  #859
Lives for gear
 

Quote:
Originally Posted by doug hazelrigg View Post
Agreed.



Disagree. Somewhere back in this thread I posted the potential error at various audible frequencies for several different sample rates.
So you actually prooved the Shannon-Nyquist sampling theorem wrong?

I'm glad someone finally showed those old farts how things really work. ;-)


/Peter
Old 11th October 2011
  #860
Lives for gear
 
BillSimpkins's Avatar
Physicists can use extreme sample rates for analysis. For instance, computer analysis of a butterfly flapping it's wings or earthquake tremors. Probably also come in handy for computer modeling gear responses while developing plugins.
Old 11th October 2011
  #861
Lives for gear
 
doug hazelrigg's Avatar
Quote:
Originally Posted by Audiop View Post
So you actually prooved the Shannon-Nyquist sampling theorem wrong?

I'm glad someone finally showed those old farts how things really work. ;-)


/Peter
No, the theorem is still intact. The peak values issue is unrelated to frequency or bandwidth
Old 12th October 2011
  #862
Lives for gear
 

Quote:
Originally Posted by doug hazelrigg View Post
Uh, true, but NOT due to the sampling theorem per se. Again, using higher sample rates does yield better quality because

>it allows gentler filtering
>it allows more accurate measurement of peak values

The sampling theorem says NOTHING, ZERO, about aliasing or transition band filtering. So the phrase "engineering as applied to nyquist" is quite silly. Neither Nyquist nor Shannon envisioned digital audio, you know. But you still seem to be saying that using fs>2B yields better "quality" in the audible range. It doesn't. In fact, the Nyquist theorem itself, which you continually cite, proves this. Conclusively. tutt

Ignorance is one thing, and nothing to be ashamed of, but to continually expose it so brazenly is appalling
as i noted
it is the ENGINEERING APPLICATION of the theorem

nonsense
the math theorem says nothing about practical realisation using circuits

within the error limits of voltage resolution
and clock jitter then higher bit depth and higher sample rates do improve the a/d/a in actual practice
Old 12th October 2011
  #863
Lives for gear
 

Quote:
Originally Posted by dcollins View Post
Did you really go to college?

DC
college
grad school
instructor
lecturer
professor
asst dean
Old 12th October 2011
  #864
Lives for gear
 

Quote:
Originally Posted by doug hazelrigg View Post
No, the theorem is still intact. The peak values issue is unrelated to frequency or bandwidth
partially true --
**if** the theorem were correctly applied

but when the hypothesis is not true then peaks can and do result in the final d/a because the digital signal is no longer band limited
Old 12th October 2011
  #865
Lives for gear
 

Quote:
Originally Posted by doug hazelrigg View Post
No, the theorem is still intact. The peak values issue is unrelated to frequency or bandwidth
But the theroem says that a bandwith limited signal will be reproduced.

It does not say that it will be reprodcued with squashed peaks that happened to fall between two samples.

Intersample peaks anyone?


/Peter
Old 12th October 2011
  #866
Lives for gear
 

Quote:
Originally Posted by swim View Post
In the audio band.
Ok, I assumed you would not give an explanation. The audio band is represented by a 24bit wordlength of the PCM samples. Does 4x sample rate add a bit so we end up with 25bit samples? Of course not.

And the actual precision of every sample is not limited by digital resolution but by the analog noise in the input and output stages. You can not gain meanigful sample accuracy by sampling with higher speeds when the 24bit wordlength and (more important) the analog noise dominates the errors.

Quote:
Increased bit depth provides more accurate capture, on a sample by sample basis, at all amplitudes. Sample values far above the "noise floor" are a more accurate representation of the input values. In addition to that fact, and as a secondary result, the noise floor is also lower.

It is not "one or the other". It is both.
The signal "far above the noise floor" also are affected by the noise. It's not like the noise only lives at weak signals. The noise is there always on every sample and IS an inaudible thing even tough it is what set the actual limit on the precision of the samples.

And no, the dominating analog noise floor is not lowered by going for higher sample rates.

Quote:
By way of example, but not necessarily limitation, I am saying that it is superior to capture the samples with greater precision / accuracy in the first instance, and also to maintain that approach in subsequent calculations / processing.
But you wrote:
Quote:
Note that acquiring individual samples with greater accuracy in the first instance is better than trying to use dither to achieve a statistical effect over time.
And it's still unclear what you are getting at with this. No one suggest sampling at 16bit with dither instead of sampling at 24bit. OTOH you effectively get some free dither in a 24bit AD since the analog noise floor has significantly higher level than the signal.

Quote:
Your notion that everything is useless unless you deem it "audibly significant" [apparently at some arbitrary stage of the process which choose not to reveal], is a false premise.
I don't say it's useless to sample at higher rates in order to avoid aliasing in plug ins for example.. or that 24bit sampling is a waste just because 16bit is perfect for a consumer format in 99.99% of all situations. Don't put words in my mouth, ask for my stance or opinion instead. Also it's not about ME deeming it audibly insignificant. I rely on the knowledge about human anatomy and hearing and what scientifical tests has taught us. Funny enough my own tests and controlled (non published) tests by other professionals I know in person reflects the findings of the scientifical community as a whole.

Quote:
Even if you could prove the absurd theory that a certain method's impact could never engender, or contribute to, an "audible" result [which you could never truthfully prove], it remains that it is simply not necessary for something to be audible for it to desirable or to be of benefit.
I can't proove that there are no flying elephants either (on earth or elsewhere), but do you think it would be meaningful to finance expeditions to search for flying elephants?

Extraordninary claims demands extraordinary proof. If someone claims to be hearing something which to the scientifical world is "known" to be inaudible I think it's fair to ask for something to support it. And it's not to be a pain, it's to LEARN about this and if necessary change the text books.

If you claim to be able to walk on water, be prepared to receive requests for evidence.

Of course some margin is useful in many situations.


/Peter
Old 12th October 2011
  #867
Lives for gear
 

Quote:
I don't say it's useless to sample at higher rates in order to avoid aliasing in plug ins for example..
Yay...another one gets it
Old 12th October 2011
  #868
Lives for gear
 

Quote:
Originally Posted by swim View Post
I don't really think this requires explanation, as it is pretty much common knowledge. You gain about 6dB for each 4x increase, and that translates to effectively 1 bit.
Indeed, hence very high oversampling for sigma delta. But the issue is the decimation, which must occur to derive an output which can be used. IOW, AFTER decimation, how does 192k compare to 96k, 88.2k, 48k, 44.1k?

There are two key differences: bandwidth and filtering.

The answer depends on how much bandwidth is required and audibility of the filter. That is the intent behind the topic of this thread, is it not?

As far as I know, this thread is not about the merits of oversampling as implemented in common sigma delta converters. It's about the output.

Or am I missing the point?

Sean
Old 12th October 2011
  #869
Lives for gear
 

Quote:
Originally Posted by Captain Proton View Post
Yay...another one gets it
Eeehr it's hardly news is it?

AFAIK it's been known for many years that working with high sample rates (captured or upsampled) can have benefits in some processing. It's also been known for many years that some converters performs better at higher rates, and some converters performs better at lower.

The problem is (in many discussions that is) that some people with a lack of understanding draw flawed conclusions from this.


/Peter
Old 12th October 2011
  #870
Lives for gear
 
projektk's Avatar
 

I feel 96 KHz is the most we need. Also your DAW decides your sample rate. I don't have HD but if I did I'd still avoid 192 in Pro Tools because its simply less tracks in my session. I'm no 96 track Guy so 96 kHz is fine for me, I find my AT2020 sounds a lot less muddy at 96 compared to 48 as well.

Sent from my LG-P925 using Gearslutz.com
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