The No.1 Website for Pro Audio
88.2 - 96kHz Vs 44.1-48kHz (a thread to end them all!!)
Old 11th March 2015
  #1
Lives for gear
88.2 - 96kHz Vs 44.1-48kHz (The facts about standard rates)

This is a post that I just made it a recent thread. One of a plethora of high definition audio threads out there. Hopefully this thread can help end this silly debate over sample rates. And hopefully dissolve a few misconceptions about recording above 44.1kHz.

If you can hear audio far above 20kHz, high sample rates are for you. Unfortunately, very few humans can hear much above 16kHz. And yes, I know the argument about how high def will make the frequencies we do hear sound better. This is completely untrue. 1000Hz at any sample rate should sound the same. In fact 20-20,000Hz should sound the same at any sample rate. If the sound changes at different sample rates, you probably don't have very good converters.

I think this misconception of high def sample rates is due to people using photography as an analogy. Meaning the higher the resolution, the more detailed photograph. Digital audio dosen't work like that. Think of audio as light instead. Light has a spectrum that we can see. And like audio, there is a spectrum above (infrared) and below (ultraviolet) that we can't see. Much like how most people can't hear past 16-18kHz or under 20Hz. A quality converter samples the audible frequency as close as possible to the source, regardless of the rate. If 1000Hz is a blue ball and a camera takes a picture of it, a quality camera will produce a picture of a blue ball like the source. Now if there was a setting on the camera that could capture light all the way up into the infrared spectrum (96kHz) and you took a picture of the blue ball again, it should be exactly the same color—even though there are additional infrared frequencies we can't see in the picture.

Another misconception is analog tape speeds compared to sample rate speeds. With tape, it's achieving the same noise floor of about 12 or 13 bit digital audio. Speeding up tape to 30ips will lower the noise floor to achieve a better S/N ratio... Maybe 14 bit. Also the frequency roll off shifts up. Meaning, at say 15ips, the lowend roll off will start at 30-40hz. At 30ips it will start rolling off at 50-60hz (depending on the machine and calibration). With tape speeds, it's always a trade off between: 15ips—bigger low end, softer highs and noisey or, 30ips— less sub low end, crisper highs and less noise. From 44.1khz to 96khz, we don't get this frequency shift like tape. With digital, the frequency responce is flat from end to end of any given sample rate. You just are just recording inaudible frequencies. Analog tape is a very crude way to record an audio signal. It's noisey, it's frequency responce is all over the place and harmonic distortion evrywhere... But it is super vibey and the artifacts are pleasing!

Here's the kicker... Guitar amps start rolling off at 5khz, most all dynamic microphones start rolling off at 16khz at best, most condenser mics start rolling off at 18-20khz, same with most preamps. What are you guys recording up there?!?!

I've never heard a hit song that was ruined by a bad choice of sample rate!!

Last edited by Flippy Floppy; 14th March 2015 at 08:08 PM..
Old 11th March 2015
  #2
Lives for gear
 
White Falcon's Avatar
Quote:
Originally Posted by Flippy Floppy View Post
I've never heard a hit song that was ruined by a bad choice of sample rate!!
This kinda sums it up.

Last edited by White Falcon; 11th March 2015 at 07:45 PM..
Old 11th March 2015
  #3
Lives for gear
 
Whack Doofa's Avatar
Sadly I doubt this thread really will end them all, but here's hoping.

From what I have seen over the last 5 or 6 years most people have come to the same conclusion and 90% of the material I see or get is recorded at 44.1kHz, unless it's TV stuff.

There are differences of course but most engineers (myself included) seem satisfied with 44.1kHz.
Old 11th March 2015
  #4
Lives for gear
 
lobsterinn's Avatar
In terms of listening to recorded music, I'm totally fine with 44.1.

When producing, I've found my converters sound slightly better at 88.2, and I've found some key plugins sound better processing there as well (yes, I was bored one day and tested this). It's definitely not a difference I'm terribly concerned with, but when given the option why not go with what sounds best? Drive space is cheap!
Old 11th March 2015
  #5
Lives for gear
 
ARIEL's Avatar
To clarify then , any mid range frequencies say at 1000 HZ is captured equally by 44k and 96k . The 96k does not double the sample capture of that frequency compared to 44k
Old 11th March 2015
  #6
Lives for gear
 

Quote:
Originally Posted by Flippy Floppy View Post
You just are just recording inaudible frequencies.
Not really. The fact is that from 20-20k most modern converters will be more accurate at 88.2/96 than at 44.1/48.
Old 11th March 2015
  #7
Lives for gear
 
lobsterinn's Avatar
@ Flippy Floppy :

I think you are missing a key argument: processing at higher sample rates puts aliasing artifacts up above the range of human hearing. Aliasing = non-harmonic noise, aka "digital hash" "pro tools haze" etc.
Old 11th March 2015
  #8
Lives for gear
 
mike vee's Avatar
Everything I do sounds better at 96, period. Recording, plugins, VSTis, my converters and my Integra 7. I can easily hear it.

But I honestly use 96K most of all for the lower latency.
Old 11th March 2015
  #9
Lives for gear
Quote:
Originally Posted by ARIEL View Post
To clarify then , any mid range frequencies say at 1000 HZ is captured equally by 44k and 96k . The 96k does not double the sample capture of that frequency compared to 44k
If I'm understanding you correctly, Yes, the extra data points of 96khz do not create a more clear 1 kHz sine wave. Digital samples are data points (kind of like dots) that are connected together with bit quantization before the DAC. When bit quantization is applied, the waveform is smooth not jagged like pixelation in a digital picture or s***** like your DAW waveform display.

With a 96 kHz sample rate, the extra data points, in a 1khz sine wave, are covered up by bit quantization. It does not make it any smoother. The extra sample rates just make it possible to hear frequencies higher than standard same rates because of the Nyquist theorem.

Last edited by Flippy Floppy; 12th March 2015 at 04:05 AM..
Old 11th March 2015
  #10
Lives for gear
Quote:
Originally Posted by lobsterinn View Post
@ Flippy Floppy :

I think you are missing a key argument: processing at higher sample rates puts aliasing artifacts up above the range of human hearing. Aliasing = non-harmonic noise, aka "digital hash" "pro tools haze" etc.
Isn't that what over-sampling is for?
Old 11th March 2015
  #11
Gear Maniac
 
DaveNJ's Avatar
 

I get the audible frequency range stuff, but haven't there been studies about how we "feel" audio beyond the 20Hz to 20kHz range? Couldn't audio still be like light where we can be affected by light we can't see?
Old 11th March 2015
  #12
Gear Addict
 
haysonics's Avatar
 

Quote:
Originally Posted by Flippy Floppy View Post
Isn't that what over-sampling is for?
Yes, oversampling fixes the issue of 100Hz (the .1 of 44.1) not being enough bandwidth for a filter to be inaudible.

Oversampling is also employed in a number of effects processing so that higher sample rates are not needed. Some older plugins that did not employ oversampling with 44.1 processing and so will sound better at 96.

If folk hear a difference with a converter at different rates then there is a design problem with the converter. There are two possible reasons for this:

1 - there are some extra audible issues that relate to high sample rates that need to be dealt with. Some converters do a better job of that than others. Folks may hear and like the distortion that a subprime converter introduces at 96
2 - there may have been a deliberate design fault employed with the 44.1 section so as to justify an expensive high rate converter

There are a couple of engineers offering rewards to anyone that can detect the difference between lower and higher sample (and bit) rates (in double blind testing using a correctly designed converter) if you trawl the net.

A lot of the myth has come about due to the early days of cd (before oversampling) and also the increasing level of loudness that has been employed since the 90s. The latter is apparent when comparing masters of vinyl, cd and sacd. A number of remastered CDs (like Nevermind and Brothers In Arms) sound horrible as they have been mastered badly.
Old 11th March 2015
  #13
Quote:
Originally Posted by haysonics View Post
Yes, oversampling fixes the issue of 100Hz (the .1 of 44.1) not being enough for a filter to be inaudible when recording.

Oversampling is also employed in effects processing so higher sample rates are not needed there either.

If you can hear a difference with a converter at different rates then there is a design problem with the converter. It may even be a deliberate fault to justify an expensive high rate converter
If you can hear a difference it may be problem with your projections too
But both statements are without factual backing - they are both opinions
In the end no matter how objective your claim is-
It all returns to being 'known' in subjective experience.
Both a richness and challenge of our world

Truth is 44.1 is fine if your convertors are good.
But if you like higher rates - by all means enjoy them!
Old 11th March 2015
  #14
Lives for gear
 
Sotsirc's Avatar
 

Quote:
Originally Posted by Flippy Floppy View Post
If I'm understanding you correctly, Yes, the extra data points of 96khz do not create a more clear 1 kHz sine wave. Digital samples are data points (kind of like dots) that are connected together with dither. When dither is applied the waveform is smooth not jagged like pixelation in a digital picture or s***** like your DAW waveform display.

With a 96 kHz sample rate, the extra data points, in a 1khz sine wave, are covered up by dither. It does not make it any smoother. The extra sample rates just make it possible to hear frequencies higher than standard same rates because of the Nyquist theorem.
What do mean the samples are connected together with dither?

Are you saying the waveform of a 1kHz tone will be jagged if you're not using dither?
Old 11th March 2015
  #15
Gear Addict
 
adogg4629's Avatar
 

The funny thing here is that there is a mix environment where 20hz frequencies are used all the time, as are 15khz and up. In feature films the sub effect track will reproduce this for that low end effect rumble and often some library or field effects have high frequency harmonics that are nearly impossible to reproduce with musical instruments in studio. Also, we are still mixing at 48/24 partly because the best spec'd out listening environments rarely exist that would make higher sampling rates worth it, and most movie theaters can barely reproduce current film mixes without distorting.

AD
Old 11th March 2015
  #16
Lives for gear
 
lobsterinn's Avatar
Quote:
Originally Posted by Flippy Floppy View Post
Isn't that what over-sampling is for?
Yes, but oversampling is not a perfect solution (latency, etc.), and lots of otherwise great plugins don't offer it.
Old 11th March 2015
  #17
Gear Addict
 
haysonics's Avatar
 

Quote:
Originally Posted by mahasandi View Post
But both statements are without factual backing - they are both opinions
In the end no matter how objective your claim is-
It all returns to being 'known' in subjective experience.
Both a richness and challenge of our world
Just download a free ABX (double blind test) program and the facts will become apparent. Yes, opinions are subjective and facts are objective but what is "known" changes as new facts emerge. We can train our interpretation (subjective experience) to be more in line with objective reality through ABX testing.
Old 11th March 2015
  #18
Lives for gear
 
junya-eskimo's Avatar
Old 11th March 2015
  #19
Lives for gear
 
Lipps's Avatar
 

Why do so many people have a problem with others recording at higher sample rates? Record at 44.1 and move on!
I'll be at 96k "recording bats", "wasting drive space" and "CPU resources".
Old 11th March 2015
  #20
So, all else being equal, what would be preferable -- high end converters at 48k, or cheaper converters at 96k? I need to add some channels to my setup and I'm wondering if I'd be better off buying a used Lucid or Mytek etc that only does 41.1/48, or a newer cheap converter that does 88.2/96.
Old 11th March 2015
  #21
Lives for gear
 

Quote:
Originally Posted by Lipps View Post
Why do so many people have a problem with others recording at higher sample rates? Record at 44.1 and move on!
I'll be at 96k "recording bats", "wasting drive space" and "CPU resources".
It's well documented that 60-70 kHz is optimal as far as conversion quality goes. It's also well documented and a proven fact that DAWs and plugins can benefit oversampling as high as 96k......

If you can't hear the difference then just record at 44...or 32k for that matter.. certainly 16khz is about as high as an older analog recording was capable of
Old 11th March 2015
  #22
Lives for gear
Quote:
Originally Posted by Sotsirc View Post
What do mean the samples are connected together with dither?

Are you saying the waveform of a 1kHz tone will be jagged if you're not using dither?
Sorry, I'm talking about bit quantization. Dither is related to bit depth. But both together will create a nice smooth sine wave from you DAC.

My bad. Thanks for noticing
Old 12th March 2015
  #23
Gear Head
 

Sound is an energy equation. The original energy equation has no sample rate parameter, it is infinite. In audio processing, the energy equation is altered by all processes on the sound, including the rate at which it is sampled, which becomes a parameter of the energy equation as it moves from infinite to finite. The higher the sample rate, the closer the new energy equation will be relative to the original energy equation.
Old 12th March 2015
  #24
Lives for gear
 
Jazz Noise's Avatar
This thread is bait. We've never had any conclusive fact finding from the 88-192Khz facts from a scientific point, and these soul searching meanderings are tedious. Oversampling, aliasing, filter topology and others have been discussed in other threads by people far more qualified than most to have any input and with enough data to put most people in their box. But the fact is no matter what you say, some people will always cling on to their version of reality.

Op knows this and is looking to get himself a nice big thread. Don't take the b8, m8.

Old 12th March 2015
  #25
Lives for gear
 

Quote:
Originally Posted by Jazz Noise View Post
This thread is bait. We've never had any conclusive fact finding from the 88-192Khz facts from a scientific point
There is plenty of science that supports the need for higher sample rates.
88-192? well maybe not that high but higher than 44/48.... The problem is people want to revert back to Nyquist Theorem verbatim and cling to the myth that 44k is *always* sufficient..when in fact its foundations are not based on modern DAW system requirements. When Shannon/ Nyquist were doing their work, they were not considering future requirements for high quality audio workstation, plugins and VI... Those technologies did not exist in the 1940s.. sort of like when you know 640k is all the memory anyone would ever need?? yeah that was Bill Gates who stated that ... Remember when the inventor of the first 64 bit processor said "no one will ever need a personal computer"?? well times change and requirements change with them. Nyquist Theorem is only a guideline it is not some interstellar law across the universe. It is common knowledge 65kHz or so is optimal for A/D. Any DAW or plugin developer will tell you 96k is an advantage internally to a DAW system. That also is well documented.

Some of you guys should just stick with your ADATs and your Mackie 2408s... and let professional recording engineers and qualified computer engineers and EEs worry about what is actually required to push the threshold of audio quality.

Last edited by therock; 12th March 2015 at 03:45 AM..
Old 12th March 2015
  #26
Lives for gear
 
Silent Sound's Avatar
A thousand years from now archeologists will look back on this thread and label it the turning point of humanity. Just think, we are only a few posts away from solving this debate and bringing the world into a utopian paradise! I'm so excited to read and cherish each and every post! We're so lucky to be alive in the year 1 B.C.C. (Before Conversion Consensus)!
Old 12th March 2015
  #27
Lives for gear
 
Sotsirc's Avatar
 

Quote:
Originally Posted by tridelica View Post
Don't think 44.1 is good enough. The standard should be 96/24, for recording, processing, and playback, and be done with the debate. Future technology improvements will relegate 44.1 to the dustbin of history.
It seems prudent to have a delivery resolution that can possibly outstrip the subtleties of human hearing, like DVD audio. Forget CDs. You don't want a car that can go just the speed limit, you want headroom, so the resolution of your files is more future-proof.
Well, 44.1 was created with headroom in mind (22 kHz). By future-proof you mean when evolution endows us with super hearing? There is a reason humans can't hear beyond 20 kHz, there is nothing useful for us up there.
Old 12th March 2015
  #28
Gear Addict
 
haysonics's Avatar
 

Quote:
Originally Posted by musicbird54321 View Post
Sound is an energy equation. The original energy equation has no sample rate parameter, it is infinite. In audio processing, the energy equation is altered by all processes on the sound, including the rate at which it is sampled, which becomes a parameter of the energy equation as it moves from infinite to finite. The higher the sample rate, the closer the new energy equation will be relative to the original energy equation.
You can't get "closer" to something that is infinite but if you want infinite then just record using analogue equipment. What's interesting is that digital (with a 44.1 sample rate parameter) is transparent in ABX testing against source material whereas analogue is not.

Quote:
Originally Posted by therock View Post
Some of you guys should just stick with your ADATs and your Mackie 2408s... and let professional recording engineers and qualified computer engineers and EEs worry about what is actually required to push the threshold of audio quality.
Or just download a free ABX program, pull back the curtain, take the red pill, and then put money saved by avoiding expensive esoteric gear towards buying more gear ! And whatever else, don't listen to the "professional recording engineers" as they just need to justify the expensive kit. Listen to scientists instead :
http://www.trustmeimascientist.com/2...ist-challenge/
Old 12th March 2015
  #29
Lives for gear
 

Quote:
Originally Posted by Sotsirc View Post
Well, 44.1 was created with headroom in mind (22 kHz). By future-proof you mean when evolution endows us with super hearing? There is a reason humans can't hear beyond 20 kHz, there is nothing useful for us up there.
sample rate is not only synonymous frequency "range"... it also has a lot to do with the filtering process and its design implementations as well. You have to keep in mind sampling theorem goes back way further than the advent of the semiconductor. Modern digital devices didn't come out till 30 years after Nyquist's work, which was based on Telegraphy encoding. A far cry from hi quality audio we use it for today.

Last edited by therock; 12th March 2015 at 06:45 AM..
Old 12th March 2015
  #30
Lives for gear
 
Sotsirc's Avatar
 

Quote:
Originally Posted by therock View Post
sample rate is not only synonymous frequency "range"... it also has a lot to do with the filtering process and its design implementations as well. You have to keep in mind sampling theorem goes back way further than the advent of the semiconductor. Modern digital devices didn't come out till 30 years after Nyquist's work, which was based on Telegraphy encoding. A far cry from hi quality audio we use it for today.
Yes, filtering might potentially affect the top end. But with a well built system, 44.1 is good enough for playback I will argue. It does make sense to use higher sampling rates in mixing (plugins), but for playback of a finished 2 track master arguing about fidelity is somewhat silly.
Topic:
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