The quality has to do with the number of increments avalible and the sampling rates nyquist cut off.

Higher sampling rates restore some of the ultra high frequency harmonics that are higher in pitch than we can hear that gives each instrument more of its "distinctive" quality.

If you are doing CD format (know as redbook) finals, then they are at 16 bit 44.1K and their is therory that, (in my experience) recording at higher frequencies will change the shapes of the waves we do hear and trickle down to the redbook or CD standard.

The higher the frequency, the more "snippits per second" of information is stored and the higher fidelity assumed. This is most apparent when using multiple tracks in digital.

Face it. What sounds better to you?

I have had engineers swear by keeping everything at 44.1 since it will end up that way, assuming that their sample rate converters cause audible problems.

Do yourself a huge favor.

Do works in all of the sampling rates you have, and burn them to cd. Pick the one you feel is closest to the actual performance.

For a direct answer, with fine equipment, higher is ALWAYS better.

Yes, Dusty, yes there is. It is a difference in ultra high frequency response more than anything else. While the question if this makes a difference is debated quite hotly, I had the opportunity to do an A /B test of this once and I was convinced that 96 K sounded much better than 48. The difference in openess and stereo depth was very evident. Even when you dither down to 44.1 for CD's this can be significant through the production process as the existence of these high harmonics can excite lower more audible frequencies. I feel it is a significant improvement. If you can afford the file size and the throughput on your machine it is a worthwhile pursuit. Fats
------------------------------------------------------------------------
Tannoy, Dynaudio, Blue Sky, JBL, Earthworks, Westlake, NS 10's :D , Genelec, Hafler, KRK, and PMC
Those are good. …………………….. Pick one.
------------------------------------------------------------------------

LOL Fats!

I guess we were on the same wave!!!!

Bill & Fats, Why don't you post your files as a blind A/B and we'll have a poll and see how many guess the higher bit rate. It might be fun.

John,
My Internet computer and my ISP are woefully inadequate for this application. Secondly this test was performed a couple years ago with client material and I don't have the tapes, client took them. Third, IMO any Internet transfer sucks so bad I doubt that the quality would show. IMO it's something you have to do in a studio. This test was performed at my old studio KFRS in Fremont CA using an analog 2" multitrack through an Apogee 24 /96 PSX 100. (See old thread in Small Steps “Why Digital Still Sucks) I a/b'd the analog and then the different rates starting at 96 and coming down to 48 the 44.1. Each step showed a further loss in quality. Bill may be able to post something like this. He has a much better Internet interface than I, however I am still skeptical if the quality differences world be evident on mp3’s. ... Fats
------------------------------------------------------------------------
Tannoy, Dynaudio, Blue Sky, JBL, Earthworks, Westlake, NS 10's :D , Genelec, Hafler, KRK, and PMC
Those are good. …………………….. Pick one.
------------------------------------------------------------------------

Fats, you cannot stream with a modem but you can download the file and I have mp3's that are very close to wavs and some that were done from 24 bit that are beyond redbook. You have to have the file to play back as streaming with a modem is no good.

John,

> Why don't you post your files as a blind A/B and we'll have a poll <

I agree with that. But the big problem with most "tests" I've seen is they compare apples and oranges. The only way to truly compare, say, 44.1 and 96 KHz. is to route one preamp output to two separate tracks - or, more likely, separate DAWs - and record the same performance at the same time. Anything less is just guessing.

--Ethan

I've read that, beyond the NyQuist stuff, a clear advantage of higher sampling rates is more accurate stereo localization by a listener.

From the ArtistPro website:

It’s been determined that time delay differences of 15 microseconds between left and right ears are easily discernible by nearly anyone. That’s less than the time difference between two samples at 48kHz (about 20 microseconds). Using a single pulse, one microsecond in length as a source, some listeners can perceive time delay differences of as little as five microseconds between left and right. It is therefore, indicated that, in order to provide a system with exact accuracy concerning imaging and positioning, the individual samples should be less than five microseconds apart. At 96kHz (a popularly preferred sample rate) there is a 10.417-microsecond space between samples. At 192kHz sample rate there is a 5.208-microsecond space between samples. This reasoning suggests that a sample rate of 192kHz is probably a good choice.

I'm interested, as the first poster was, if anyone can hear the difference between 96 and 192. That is, should someone looking to buy a 24-bit system today bite the bullet and get 192, or will 96 become the standard for a few years?

The advantages of the other leaps in sample rates seem well established.

there's a difference of course...but not enough to worry about in that things will always continue to "progress" and the interelationship of all of the variables that go intop making great recorded music make the format choice just that...a choice...not the make or break issue.
"Jagged little Pill" weather you liker it or not was a sucessful record on many levels and recorded and mixed off of blackface adats.
"Sgt. Pepper's..." was only(!). four track analog.
So don't wait for the next big thing....do something now with the best that you have at hand...the best way YOU can...becuase you more than the medium dictate the results.

Originally posted by RecorderMan:
So don't wait for the next big thing....do something now with the best that you have at hand...the best way YOU can...becuase you more than the medium dictate the results.

Very well said RecorderMan. Even bleeding edge technology can't compensate for one that doesn't know how to get good results; or know what good results are for that matter

themidiroom

Agreed, if you can't make a recording sound pro at 48 then something else is the problem.

Here's an interesting bit of info on why higher sample rates are 'better' (beware - it's techinical):

When an analog to digital converter (ADC) is operating at 44,100 samples per second, it can't 'see' any wave form that is higher than 22,050hz (nyqyist). If it does encounter a situation where there is a frequency greater than 22,050hz, aliasing occurs which causes distortion across the frequency band. To prevent this from happening, a low-pass filter is installed in every ADC. The actual reason higher sampling rates sound better is because they can accurately convert the signals that would otherwise cause distortion.

What this means is that the higher the sampling rate of the ADCs the better, but once it's been converted the only thing that affects the quality is the dithering process (which can be virtually transparent depending on the algorithm).

The other thing to note is the bit depth. A 16 bit converter can only distinguish 65,536 different levels per sample, where a 20 bit one has 1,048,576 leves, and 24 bit has 16,777,216 different levels. Although you may be able to distinguish a small bit of difference between 16 bit and 20 bit, there is probably no noticable difference between 20 bit and 24 bit.

The caveat to all of this is that different converters have different quality, so one brand of 44khz/16bit converters may sound better than another brand of 192khz/24bits. It also depends on the clock source for the converters, but that's for a whole other thread. I hope I didn't confuse anybody too badly. :)

The biggest difference that is audible is the resolution of the recording. Going from 16 bit to 24 bit will show a higher degree of fidelity than the sampling rate.

A 48 KHz sampling rate translates into a response up to 24 KHz & A 96 KHz sampling rate translates into a response of 0-48 KHz, well beyond our hearing range. 192 KHz translates into a response curve up to to 96 KHz. I would go with the 96 KHz sampling rate captures more harmonics and is much easier on our hearing. It really depends on what you are recording also. Cymbals for example have a natural decay all the way up to 50 KHz. The digital realm employs sharp filters. For example a sampling rate of 48KHz has a smooth response to 22 KHz. There is a very steep curve from 22 KHz to 24KHzI believe analog does a much better job of capturing the recording and then dumping it or slaving it into the digital realm, especially with older digital equipment. A/D converters have improved exponentially over the years. If you have the space, record at the highest resolution & then dither it down to the required format such as 44.1 KHz, 16 bit.

One thing that never seems to get addressed when we talk/bitch about converters and sampling rates is the converter's front end and it's componentry. If you build an A/D converter out of 10% tolerance parts on the input side, it's not going to sound as good as something built with 1% tolerance components. Of course, those parts cost a lot more.

We all get excited when we hear class A circuits. No kidding. They're built to a high spec! Many of the analog pieces we lust after were carefully designed and tweaked so that they were either sonically pleasant or as transparent as possible with the technology of the time. Both seems to happen a lot with tube gear and analog tape.

I wager, though, that if you recorded with good pre's through a high quality D/A (i.e. Apogee, Prism, etc) clocked with a high quality master clock (Nanosyncs, Aardsynch, etc) you'd be very impressed with the warm and punchy sound of your rig.

I love tape. I have tape decks here and always try to figure out how to use them on things. I love the analog process moreso than holding on to it as the holy grail of sound. Sorry guys, analog has it's limitations. It's expensive, more difficult to edit, requires lots of maintenance, has a higher noise floor and less dynamic range, and is generally more physically delicate than a hard-disk system in some ways. But GOD does it sound good! Heh!

Digital is more of a wysiwyg (should be wyHiwyg, h=hear). What you put into it is largely what you get out of it. The systems were designed to be linear and have flat response across the audible spectrum. Analog certainly doesn't do that. That's why we love it so much. Same as an overdriven tube. Sounds yummy, but certainly isn't linear. I'm not arguing for one or the other here, just trying to present the strenghts and weaknesses.

Seems to me the time we spend tracking to analog tape and then transferring to digital would be better spent figuring out how to get the results we want from our digital rigs. Oh, and for those of you that cite "Sgt Peppers was just 4-track". Yeah. They used 2 4 tracks which was high tech back then. These were not portastudios, guys. They were some of the best machines in the world, with some of the best consoles/other gear in the world. It's not about the track count on that one. 1" 4 track has some FAT track width, track width=resolution.

I'm done babbling, was it coherent?

Dan Roth
Otitis Media
Audio - Video - Film

A couiple of observations.
1. If someone want to a/b an audio file you had best use something that has not experienced data compression (like an mp3) or you are going to be testing the mp3 encoder more than the source material.

2. As Dan said, the quality of the analog front end to the converter is the single most important sonic element in an A/D converter. The next most important thing is the stability of the master clock. A very stable clock will provide a much better sound stage.

3. So far there has been no documented tests or proofs that higher sample rate makes any difference in the audible quality of a digital recording. For one thing it is almost impossible 9to make a fair test) given that a 96ksps system will have a completely different set of analog filters in fron of it than a 48ksps system will. I have a 192i/o and great monitors. I cannot tell the difference so far.

I am not saying there is no difference but so far no one has been able to publish anything other than speculation and physics does not support there being a difference.

Again i am only referring to sample rate influence. It is also possible by the way that higher sample rate designs can degrade performance. This occurrs when the designer decides to use a cheaper or sloppier anti aliasing filter design for the 96ksps system.

I have personally done extensive analysis in the DSP realm of digital audio and I can tell you that an impule response sampled at 48ksps and 96ksps will have the EXACT same spectrum within the Nyquist pass band..

I believe the AES has acommittee looking into designing something that might be a fair testing process. Until then, the only reason I can see to buy higher sample rate converters is :
1. the cost the same as lower rate converters
2. if you need to provide content for DVD-A
3. you just gotta have the biggest,fastest widget on the block.

Steve

One thing to keep in mind is, if there's any chance you'll mix digitally, you should keep your sample rate at a multiple of 44.1k. IE 44.1, 88.2, or 176.4. That way when downsampling to 44.1 for CD no fancy sample rate conversion math is reqired, it's just a matter of throwing away every other sample. Sample rate conversion of non-integer ratios just never sounds good.

PS somebody earlier said something about "dithering" down to 44.1; just to keep our terminology straight, dither has nothing to do with sample rates- it's used sometimes in reducing BIT DEPTH.

All this theory makes sense. I think a question for a lot of us "semi-pros" here is will our lower end daws have an audible difference withthe higher sampling rates? (In my case a pair of delta 1010LT's for 16 tracks. Up to at least 96khz is available)

My second question is the clock. If we got a high quality clock (Say apogeee etc...) would it make an audible improvement as well? Or since the front end is so cheap to beign with it wouldn't really matter and money would be better spent elsewhere?

Thanks.

Originally posted by Paul Berolzheimer:
PS somebody earlier said something about "dithering" down to 44.1; just to keep our terminology straight, dither has nothing to do with sample rates- it's used sometimes in reducing BIT DEPTH.

Dithering doesn't just apply to bit depth. It's a generalized term meaning that a reduction in data (quality) is required. This happens when converting to a lower bit depth and/or sampling rate.

Wes,

> If we got a high quality clock (Say apogeee etc...) would it make an audible improvement <

I don't think so. But I don't think 96 KHz. is worthwhile either. :)

I'm still waiting for one of the 96 KHz. proponents to post a short Wave file clip so we can all do a blind A/B test. It's easy to do, and it's the only way to settle this for once and for all.

--Ethan

p0rk,

> It's a generalized term meaning that a reduction in data (quality) is required. This happens when converting to a lower bit depth and/or sampling rate. <

No, dither means specifically adding noise to a recording to increase the dynamic resolution. And it applies to any digital recording, not just when converting from one bit depth to another.

--Ethan

Ok still wrong.

Dither is noise applied at a level slightly above the LSB to randomize quantization error. Quantization error is the result of the LSB (least significant bit) being in error due to rounding up or down.

Quantization error sounds bad because it tends to be periodic ( repetitive) and correlated to the audio that is being sampled. The Dither process actually REDUCES the effective bit depth by randomizing the LSB but does so in a way that tends to sound more natural then the undithered signal.

Dither does not have anything to do with sample rate conversion.

By the way, if you sample at 96kHz, then thow away everyother bit, you have exactly the same result as if you had sampled at 48kHz. So why bother???? AND sample rate conversion involves a process of upsampling to a much higher sample rate or convoluting to an instantaneous impulse response with a much higher sample rate. So SRC from 96kHz to 48kHz is not easier or cleaner than converting from any other sample rate.

but the question is: added cost x benefits in overal quality.
Twice the space or more ( 96k/192k), bounce time ( thinking in digital)...
so let u say we invest and throw away all our digital gear that only works at 44k/48k to achieve 20% of superior quality?
So... why don´t they start some really big wordlenght stuff like 32/48 bits @ 192k/384k?
I am sure that in 3 years Digi and the others will start this trend again
:(

ACB
Your right my friend. Perhaps even sooner. Truth is that 24 bit @ 44.1 sounds great. I know 96 sounds better and I am sure that 192 is even better than that. But 24 @ 44.1 is pretty good. I'm happy there. As you point out it's easier on the processor the storage etc. What really matters is what your recording! Audio people have been willing for years now to spend 5 times as much for something that returns a 5 or 10% boost in quality. But that has all become pretty much moot these days because even the most humble equipment is capable of what only the best used to be. Can you imagine what the Beatles, Pink Floyd or Freddie Mercury could do on modern systems? It boggles the mind. IMO it's time to start concentrating on the talent once again and to stop loading our minds down with all this technical crap ..... Fats
---------------------------------------------------------------------------------
To help keep this site alive and growing, please show your support by purchasing one of our caps, studio clocks, T-Shirts etc. Banner ads too!
---------------------------------------------------------------------------------
Tannoy, Dynaudio, Blue Sky, JBL, Earthworks, Westlake, NS 10's :D , Genelec, Hafler, KRK, and PMC
Those are good. …………………….. Pick one.
---------------------------------------------------------------------------------

Originally posted by sdevino:
Dither does not have anything to do with sample rate conversion.

My point was that dithering doesn't just apply to sample rates, bit depth, or even just audio. Dithering is just the process of fixing data to be better represented when certain pieces of data have to be removed.

That's pretty interesting about adding noise to the LSB -- I had no idea. Is the noise just added to the single LSB, or the least few significant bits? I've done a little searching on this topic, and it looks like this process may be unnecessary for higher bit rate conversions like 24 -> 16 because of the relatively accuracy of 16 bit data. Do you know if it's common place to still apply this algorithm when converting with these higher sampling rates?

:c:

hey, let us make a party, play and have some fun.
The audience out there doesn´t give a shit if the LSB is dithered or not.
:)

Alecio,
Guess what. there are people out there who have engineering backgrounds other than recording. And they actually DO care about this stuff.

Some of us like the techy stuff. I for one find many people to be curious about what is true and what is myth regarding many digital concepts. I signed up to moderate because I have a strong technical digital background. I will be happy to help you out with any of the digital myths you are worried about.

Like for instance: what sample rate to use and whther or not it makes any difference.

Hey POrk, the dither gets applied different ways using different algorithms by different vendors. The general idea is to apply just enough dither to randomize the LSB in a way that is agreeable to most listeners.

Usually it is applied just below the LSB/2.

For Alecio, a well dithered digital mix bus is the a very important part of a good sounding digital mixer like my HD2 or your O2R. Knowing how to work with the dither can improve your mixes in certain situations.

Steve

Just curious, has anybody looked at the front end (analog input side) of a typical A/D converter with a scope with no signal applied? I scoped out a converter we use at work for a non audio application, and I couldn't believe the crap that I saw. I am just wondering how clean a good audio converter is at the front?

How did you terminate the input? Most modern audio gear outputs have very very low AC impedences (like almost zero). This makes for low noise pickup on the transmission line.

If the input to the converter is basically open circuit you are not measuring what is there under nominal conditions.

Steve

Many thanks Steve. Man, I am working at 44.1k/24 bits, PT 5.1.1 Mix
. I used to do 48k/16-18 bits ( adat XT) and 48k/24 ( Dm800 - Pro Tools 4.3).

this article explains it (dither) nicely: http://www.digido.com/index/pmodule_id=11/pmdmode=fullscreen/pageadder_page_id=27/

There have been quite a dew studies done. The fact is that the human ear, unless you are a freak of nature, or you have taken REALLY good care of yourself, by the time you are past your 20's is unlikely to hear past 16, let alone 20kHz. My brother is a doctor and an audiologist. I have these facts around. EVEN IF we could here 20kHz now, the ear does not recognize complex tones. So that BS about the instruments harmonics being captured and complex waves being represented is just stupid. Yes you will hear localization and/or spacial effects, but we can only hear pure tones from about 8kHz on up. The ear, at high frequencies, has about the same resolution as a 48kHz sample rate.

Besides all of this. How many of you actually own a microphone that can record flat out to 48kHz (for recording at 96kHz) or one that is flat to 96kHz (for recording at 192kHz)? None. So, you are going to tell me, that you are going to recreate some complex high-frequencies, assuming that it worked, and you are going to capture those frequencies with a mic that is flat out to only 20kHz or so? Come on. What a bunch of crap!

I own a PT HD rig by the way. I do hear some differences. But if you get really honest with yourself, you would have to do some REAL strict testing to make it all fair. Same clocking when recording and playing back, same converters, same material, same monitors, same sweet spot. Unless you are doing un-electrified music, anything above 48kHz is a waste of hard drive space.

Originally posted by lowdbrent:
So that BS about the instruments harmonics being captured and complex waves being represented is just stupid.

I think you are referring to sampling frequencies above 22khz (assuming a sampling rate of 44khz). The reason this happens is because the samples won't be accurate if it sees anything above 22khz. It will cause distortion and various artifacts. To fix this, a high pass filter is put in place to get rid of any possible higher frequencies (i.e. harmonics). There is a lot of info available on this subject -- just google for nyquist.

:c:

Time to put this to rest.

The top wave is a sine wave done at 18KHZ at 16 bit/48KhZ

The bottom wave is the same sine wave at 18KhZ at 32 bit 192K.

Which one resembles a sine wave?

I invite you to perform this test yourself and at 44 years old, I can hear 18KhZ. The 16 bit 48K sounds shreeky, the 32 bit 192K is a pure tone, audible and visual.

No contest folks. I master in the highest resolution I can run. I also repair wave problems with 16 bit <48K.

I rest my case.

Fu*in' right on!! My wife is 40ish and can still hear 24K! We call her "dog ears". Back in the 70's when I worked on mostly 4 tracks, I would get her to listen to my "bounces" just to make sure I wasn't generating internal feedback off the heads. She's my hero! If the limit of your hearing is 16K then you need to find another line of work! Kurt

Now you really want to fuck up the naysayers and flat earthers?

Lets do one at 12K and I am SURE everyone can hear that.

Now folks, do you really want your amps and speakers to work with a signal that looks like this?

12000.

Top is 16/44.1K

Bottom 32 bit 192K

I hear it, I see it, therefore, it is.

Again, open a wave editor and a sine wave generator as I did and get busy looking at the mess 16bit 44.1K or 48K does.

After all this sitting back and laughing senselessly at all the established people and their beliefs, yes ones here.. why...why did not one of you actually take the 2 mins it takes to perform the test?

Because you love listening to others and believing unfortunant untruths.

PS, I have worked in digital since 1978. This is not something new, I just waited this long, hopeing someone would come to their senses and do the test.

Again, I rest my case.

I was talking about the average person having average hearing loss by a specific age. I can hear 18kHz in one ear myself and I am 35.

There are freaks of nature out there. But consider this. The cochlea's size dictates what the hearing range can be for each person. The hairs inside the ear can only be so long and so short. (The smaller receive the highs.) So, for one to have hearing out to 24kHz is nearly impossible, and is truely a freak thing.

As far as the sine wave test, that is crap. Nobody is disputing that 16-bit sucks. The issue is sample rate, not word length. and more to the point, the issue is that the ear cannot hear complex tones that the higher smapling rates are supposed to reveal. All that is needed to accurately reproduce a waveform for humans is the content in the waveform that exists within that range. You can't even tell a difference between a sine wave and a square wave at 15kHz, or so. If you perceive a difference it will only be the 1.4 x increase in amplitude of the fundamental in that waveform over the amplitude sine wave.
You couldn't hear a difference at 8kHz. Again you might perceive a difference for the reasons above, AND more importantly you would hear 16kHz component, which is there in 48kHz recordings

If you record at 48kHz, isn't that all we get above 8k, a sine wave (roughly) because only the first overtone is converted, all the rest are lost? Yup, just as the ear hears it. Your ear hears a fundamental and a first overtone, just like the sampling system does. Anything above 48kHz is a waste.

I Strongly dissagree.

What you said "I couldn't hear" I actually can.

You do not have to believe.

I said mine.

No argument here, you can choose your own beliefs.

No need for me to go further, You have a lot to digest.

I welcome you to continue with them actually.

It gets worse...

Alecio says..

hey, let us make a party, play and have some fun

LOL!! It is not only a party it is a fucking riot!!!!!!

LOL!!!!!