i hear diffrences in all but 192 k since i havent heard 192 k yet but the real diffrence is warmth.
You can tell the warmth factor of 96 k over 44.1 or 48k real quick, also think of the processing diffrence if its gettin knocked down to 44.1 you can do a lot more processing on 96k and it still sound less processed then 48k in other words retains more of the quality of the mix without compermising signal loss, or if your a pretty pure recordist or do classical recordings or symphonic work of any kind the diffrence is near amazing.

I have never worked in 96 k but I have heard it. In a side by side comparison the difference is pretty obvious. But the cost of working in these higher rates in terms of storage and processing power is not worth it IMO. I work in 24 bit 44.1 and the stuff I put out sounds fine. It's all coming down to 16 bit 44.1 anyhow..

doulos21 I was just replying to you on another thread. The reason I started this thread, is because I'm thinking of this 002r thing to, well seems we are on the same boat. So, you say it is worth it to go 96.

Kurt, you're online. If 96 is 2x48 then you are doubling the workload on the CPU?? (and using more storage)
Is 48 vs 96 comparable to 15ips vs 30ips on tape??

Golli,
Yes and on the memory too. Storage also.. Kurt

I work in 48kHz alot, because most other studios that I interface with work at 48kHz.

For my in-house work, I do 96kHz and 192KHz sessions. I max out the channel count on the DAW too. The problem for me is not storage, it is the fact that there aren't as many cool plug ins available for 96k and even less are available for 192k. There are meat and potato plugs, but the cool ones that I use (some of Wave's, etc.) are not there yet.

Just because we can sample that high, doesn't mean that we should. The differences between 96 and 192 are more subtle than you would imagine. The differences are stereo depth and imaging related. It is my experience that classical, chamber, etc. style music should get the 192 sample rate. Pop/Rock music should not. That music has little dynamic range comparitively, and less bandwidth.

Do it both ways and listen for yourself. Your music may warrant 24-bit/44.1 possibly.

So more and more lately we're all hearing about 96k becoming the next standard sampling rate at the production level, while audio CD will remain the most widespread end product for the masses for the foreseeable future.

I remember going to a demo put on by the big DAW manufacturer out there about a year and a half ago and the presenter emphasizing the importance, in terms of SRC quality, of working in 88.2k or 176.4k when it is known that audio CD is the end product, due to the easy mathematical fold down to 44.1.

It makes perfect sense to someone like me, who doesn't have the capability of working at higher rates than 44.1k (or 48k, but why work at 48k when delivering to 44.1k?). But for those of you working at higher rates, why would one choose 96k (as opposed to 88.2k or 176.4k) when delivering to CD, due to the more complex SRC required to get to your final medium?

One more thing.......

I've raised this point in several other arenas without ever receiving a fully satisfactory answer.

Simply put, how can the benefits of extended frequency response afforded by higher sampling rates be realized in any way when 95% of off the shelf monitors and microphones in use in studios now only offer 20-20k response? As Kurt points out, transducers constitute the ends of the recording chain, but if you increase your capability (frequency response) in the middle of that chain, the end product is still dumbed down by the ends of the chain.

I've long argued that while the capability to CAPTURE those extended frequencies may exist, they still haven't made their way into a DAW via a microphone, and even if they have via a synth or oscillator, we still don't have the monitors (although I realize there are more and more exceptions here as of late) that can reproduce them for us to hear.

This question is only meant to address frequency response specifically, not the benefits derived from higher resolution, which are readily apparent.

Hope I'm not steering this thread off topic........

accidental double post

I also work at 44.1k/24 bits. avoiding SRC is a must. Record at 96k/192k is pretty pricey. While we are still delivering the final products at regular Cds to me there is no big eal moving that high ( storage, obsolescence planned). When DVD-Audio arrives... then, it will be a completle different approach.
:)

Well put Jamie. You are allways one step ahead of me.
Since we are both considering the 002r, besides the 96k issue staying in 44.1/24 bit there is still room for improvement. In the 002 there is better A/D-D/A and better pres, balanced l/O's. two birds with one stone :roll:

Although manufacturers only spec out to 20-20 (due to the nyquest rate of cd audio being 22.05k) mics and analog gear can easily go above that...

I remeber rupert neve talking about some of his gear going well about 100k and that distortion in the 50k range on one of his pieces was problematic.

You may not hear above 20k but the modulation and the transient response of higher sampling rates is apparent even as you drop down.

The problem lies in how much processing power it takes to do such high end stuff.

Maybe we should start tracking things like drums, vocals and acoustic instruments at 88.2 or 17blah.blah blah and keep things like bass and distorted guitar at 44.1

I'm gonna try that on my next project.

Marsh

Keep us posted.

Nice stuff there guys! Kurt :)

Marshall - I'll be curious to hear how that works out for you.

Again, though, Mr. Neve's console/pre sits/sat in the middle of that chain. I'm still not convinced that the mic's and monitors at the ends of that chain are capturing and producing those frequencies at levels substantial enough to make an audible difference to the listener.

It is my understanding that manufacturers (mic and monitor manufacturers specifically) didn't just start limiting their specs to 20-20k when digital hit the market. And if any mic or monitor offered/offers significant extension of response, wouldn't that manufacturer do all they could to tout their product? To impose the limitations of 44.1k on their spec sheet would be like Porsche saying "With our new (insert model here) you can reach speeds of 100 mph," when in actuality the car can hit 160 mph. It just doesn't make sense to me. I remember a former moderator here saying that he had mic's that had response out to 40k, +/-, but he said that out there they were WAY down, in the neighborhood of -20db. I don't see how the benefits of capturing a frequency that far out of our range, that far down, with maybe one or two mic's on a given project, can be all that beneficial in the end. And then we get back to monitors.......

I'm just asking someone to convince me that the capabilities are FULLY IN PLACE to capture and play back these extended frequencies. I JUST DON'T BUY IT YET.

This is fun. Keep 'em coming.

Even 20 down the difference is audible. I have a set of Tannoy DMT12 that respond flat to 35K.. mmm kay?

You know, the Nyqist theorem states that a sample rate can capture frequencies that are half the total samples per second.

What I've never understood was that different phases for the frequency in question could radically alter the values being sampled by the ADC.

Take for example - an ADC running at 48kHz samples a 10V cosine wave (first sampled value will be 10, next -10, third 10, etc). What if this component hit the ADC at a 90' phase? That would make it a sine (or inverted sine), which means the ADC would be sampling a whole row of zeroes.

Or do ADCs actually compensate for this phase problem and automatically shift these high freqs so that the max amplitude is sampled?

A related problem also exists for frequencies that approach the Nyqist limit - A signal with a frequency slightly below half the sampling rate will appear to be that with a frequency of exactly half of the sampling rate, with fluctuating volume. (It's hard to explain this - it's better if one can get ahold of a signal-generation program and try it out).

Now, no matter how good nth-order DA converters are at making that kind of data sound like the original, unadulterated signal, amplitude-dependant algorithms are going to have a fun time butchering the heck out of the signal.

Of course, these problems only become significant as frequencies approach the Nyqist - but I'm going to take the plunge and say that even my unprofessional ears can hear subtle differences between 44.1 and 48 - contradictory to what the textbooks say about 44.1 being able to sample up to 22.05 and human hearing being able to go up to 20 on a good day. As has been brought up many times in the thread - it might not matter when there are cymbals and dive bombs flying all over the place, but the added depth and warmth it gives to acoustic recordings make the extra space and processing power required for at least 48kHz a justified compromise for me.

This is what I was thinking of, please read:

Want a real dose of Blasphemy? I compared recording at 96kHz and Sample Rate converting down to 44.1, to simply recording at 44.1kHz. I couldn't believe my ears! The track originally recorded at 96kHz and Sample Rate converted down to 44.1kHz had much better sounding highs, maintaining much of the character from recording at 96kHz

Taken from:
http://www.prorec.com/prorec/articles.nsf/files/4AE9C107C78E706886256688000FBE08
If this is true then it's worth it?

Oka. I can accept that 96k is much better than 44k or so.

However, I think that one inch at the mic placement is much more important than all this stuff.

a) what about the SRC down to 44.1k/16?
There would be 2 conversions: 96k/24 to 44k/24 then to 44k/16k. so?
I would direct this to PT guys.

b) why don´t they start working with bigger wordlenghts, let us say moving to 32 bits?

It is much better to record at 44k/24 bits than at 48k/16 bits, agree?

c) now will plugins doing fine at high frequencies and working at 48 bits. Still there happens lots of truncation. Imagine truncation at several tracks. so?

How many times have you tried doing a fade out and having it sounding horrible? Sems will be entering the Dithering Wars again...
lol

Much to chew on here.

As I said, I realize there are quite a few exceptions in the monitor world in terms of extending response beyond 20k.

Simply put, I'm putting forward the notion that the VAST MAJORITY of mic's and monitors are limited to flat response of roughly 20-20k.

Agreement on this point or no?

With that being the case, how significant is it that higher sampling rates afford the capability of capturing extended frequencies?

I don't really mean for this be a debate. I speak somewhat from ignorance, in that I haven't had the opportunity to listen to any high sample rate recordings in critical environments. But from the outside, I've always been a little leery of the argument that the extended response is part of what makes the difference so huge.

Originally posted by Jamie Smith:
Simply put, I'm putting forward the notion that the VAST MAJORITY of mic's and monitors are limited to flat response of roughly 20-20k.

Not so. The majority of cheap semi pro or prosumer speakers and dynamic mics have 20Hz. to 20K response but much of the professional high end stuff has extended response capability. Especially monitors. It is not uncommon to find monitors that extend to 40K or beyond.

i want to work in 96 k cause the time for dvd audio is among us. Though i truely doubt the general public with their cheap speakers and even cheaper headphones that settle for listening to 128kb mp3s are gonna give a crap its still gonna have to happen that were gonna need to be ready for 96 k audio. Have you ever heard up sampling of 41 k to 96 it isnt pretty 48 is better but its still upsampling i just want to be ready for the dvd audio stage as soon as the battle of wich dvd encoding is gonna dominate then cds will be on thier way out my guess is about 2 years.

hey golli im gonna upgrade to the 002r but thats last in my upgrades as i dont seem to need it right now. Im first gettin new moniters event tr6s acoustic treatments for my mixing room mini traps a new digital consol tascam dm24 computer up grade ill build it my self but im budgeting 1,200 for it a new tube mic studio projects multi pattern thingy a avalon m5 and then a 002r LAST but thats just me id say if you could find better ways to spend the $1300 say a good pre or compressor id do that first untill buisness denmands 96k my 2 bits

kurt im sure 40 k responce must look pretty on a digital read out and im sure it gives the over all speaker a flatter responce, but how much does that diffrer if we can only hear up to 20 k? its effective for the life of the speaker and better reproduction up to our hearing range im sure but other then that its gonan get chopped off when it hits the mixdown process. This to me is about when i laughed my ass off when roland came out with their first 96 k digital speaker now that was funny "a speaker that can reproduce all of digital signals" and im thinking wow a speaker can understand the diffrence between digital signals and analog :D

Well, I'm eventually going 002r, just not tomorrow. And my budget is smaller than your's (I also have a wife :)

Hi everyone.. I'm new here,
but I was reading the post, so to add on the discussion...

Like someone noted, the reason why working in higher sampling rate is because of Nyquist theorem... which from my understanding and memory involves the converting digital signal to whatever... Dithering filter roll off at half the frequence, which is fine at 44100... half is just about 22050, which is beyond audiable range... but there's other stuff happening in the converting process, such as the slope of the filter, and if the sample rate is higher, you can make that filter slope more gentle... thereby affecting the way your stuff is converted... and heard at cd frequency.
Checking out Nyquist theorem will explain everything..
skender

So do all converters/clocks with the same sample rate have the same slope, or is it the degree of the slope that separates: the good the bad and the ugly??

Though i truely doubt the general public with their cheap speakers and even cheaper headphones that settle for listening to 128kb mp3s are gonna give a crap its still gonna have to happen that were gonna need to be ready for 96 k audio

No shit. Quite true. And it is bad enough that the music listening and enjoying public have been "dumbed down" big time by the "loudness wars" and mega-compression. Compared to vinyl, just what the fuck is happening to the "art" in the music industry? Are dynamics going the way of the top-hat? I am afraid there will not, and can't, be a trend to go back to well produced and mastered audio productions due to the above mentioned and other reasons. I hope I am wrong. Like comparing a New York pizza with a Domino's "cardboard" special. No balls, no guts, less heart, lost inspiration. (Stepping off the soapbox) ---Lee

Originally posted by doulos21:
kurt im sure 40 k responce must look pretty on a digital read out and im sure it gives the over all speaker a flatter responce, but how much does that diffrer if we can only hear up to 20 k? :D

Well, IMO it is a fallacy that we all only hear to 20K. There have been studies that show that even though the higher frequencies are inaudible we can somehow sometimes still sense them. And I know of engineering “Golden Ears” who have extended hearing well beyond 20K. It is true that the average for most of us around 20K, but that doesn't apply to all of us. I’m deaf basically.. I test to about 16K with a bit of a loss at 7K but I can still hear a difference between 48K and 96K. My wife has tested to 24K. Most women do hear a bit higher than men. But it's not just about how high you can hear. Sub harmonics in the audible range may be excited by ultra high frequencies. Now I still am not saying that we should all run out and go 88.2 or 96, but I am saying let's not justify not making the move by supposing the difference isn't audible or that there is no difference. Because there is and we can hear it in many cases. I just don't think it makes enough of a difference to justify throwing everything away and getting new stuff. I was just getting happy that the whole sample rate upgrade game was settling down. I think the manufacturers benefit more from this whole upgrade thing that any one else. I hear stuff that was done on 16 bit systems that still sounds great! Kurt

I think Skender's got a good point. With higher sample rates, the anti-aliasing filters on an a/d converter don't need to be as perfect (and the converters most of us use DEFINITELY don't have perfect filters, although oversampling helps).

I also think higher sample rates help out ALL frequencies, not just the super-sonic. An 80hz wave sampled at 96khz as opposed to 44.1 khz still contains more than twice as much information as the 44.1khz sample, even if 44.1k is technically (more than) enough to capture the signal. I think that's why the bass end of the mix sounds so much clearer at higher sample rates.

One point in defense of recording at higher sample rates - sure you have to dither and sample down for CD today, but when DVD audio comes out and you want to re-release your material, wouldn't it be cool to just go back to the original recording, pull out the 96k 24bit master and use that straight to DVD. Many people won't need to do that of course but it's a thought.

I have also noticed somebody made a mistake regarding their understanding of these high sample rates. The issue is not about catching frequencies above 20khz, we can't hear them anyway, the issue is better capturing frequencies that we can hear. If you have a frequency at say 12khz, at a sample rate of 48khz the sampler is only capturing information on that wave (the 12khz) 4 times per oscillation. It is taking information at different points in the wave, since the wave is oscillating quite fast it gets less overall information about the wave than say a wave oscillating at 48hz which is getting captured I think about 1000 times per oscillation or whatever, so there is a lot more information about that end of the freq spectrum. The point is that the benefit of recording at 192 or even 96khz is that say at 96 khz the same 12khz freq is getting captured 8 times per cycle instead of just 4, and at 192 it would be 16 times, so the information that the computer or playback device has to recreate the wave upon D/A conversion is significantly higher, we have more information about the state of the wave more often and thus the higher frequencies should be better and more fully and accurately reproduced. It is not about trying to record frequencies at 192khz! That has nothing to do with it. Theoretically with equal quality converters 192 should sound a lot better than 44.1, and if you then sample down with the best downsampling algorithms possible, the end result should still be that the 192 downsampled recording should be better than the straight 44.1 recording, because there was more information captured in the first place and the algorithm did a good job of retaining and most important aspects of the information. Kind of like dithering from 24bit to 16bit, a 24bit recording dithered down sounds better than a 16bit recording under otherwise identical circumstances, but it depends on the dithering algorithm at the end, ie. you probably shouldn't just truncate it.

Micky.

How it sounds BEFORE it even gets converted to digital makes the biggest differrence by far.

WOW

You guys keep going please, im learning alot here. And saving money, my friend is going to a recording school. I told him that im saving tons of money and getting my education online. He asked "where at some online school", I laughed and said sure.

But one question? The dvd audio media? does anyone have a date on when it will come out or know of a website I can read about it? I have tryed to search for info about it, but just cant find any good info on it.

Hello Mickey. Great points. But I did limit my side of that statement to extended frequency response, which often comes up in discussions regarding higher sampling rates. I do realize audio benefits from higher resolution offered by higher rates, just as increased frame rates in video and film provide a better image. And I also realize that in the real world, super-sonic (beyond our range) frequencies affect those within our range. I was just asking specifically who's making and/or using microphones and monitors that also capture and reproduce, with relatively flat curves, these extended frequencies. I wish I had access to higher end systems than I do. Like I said, I only question this because I do not.

How it sounds BEFORE it even gets converted to digital makes the biggest differrence by far.

...How true. Garbage at 192khz is still garbage.

...my friend is going to a recording school. I told him that im saving tons of money and getting my education online.

Don't get me started!! What a complete waste of money! (I live in Orlando, if that gives you a clue about what (where) I'm talking about.)

I was also about to tell you that DVD-A is already in full swing, but come to think of it, I can't remember seeing any DVD-A titles at the record shop. I know SACD is pumping out recordings by the truckload. I'm pretty sure DVD-A recordings are readily available.

The quality of the converter makes a much bigger difference than the sampling rate itself. If you have a converter that sounds better at a higher sampling rate, by all means use it at that rate if the extra bandwidth, processing, and storage required is worth it to you. But higher sampling rates in and of themselves don't necessarily give us anything that 44.1 kHz sampling isn't capable of. There's a lot more to the sound of a converter than the sampling rate.

I remeber rupert neve talking about some of his gear going well about 100k and that distortion in the 50k range on one of his pieces was problematic.

Yes...because it affects the frequencies in the audible range. You don't need to sample that signal with a bandwidth of 50 kHz to capture the effects in the >20 kHz range.

Maybe we should start tracking things like drums, vocals and acoustic instruments at 88.2 or 17blah.blah blah and keep things like bass and distorted guitar at 44.1
I'm gonna try that on my next project.

Most software won't let you do that.

It is my understanding that manufacturers (mic and monitor manufacturers specifically) didn't just start limiting their specs to 20-20k when digital hit the market. And if any mic or monitor offered/offers significant extension of response, wouldn't that manufacturer do all they could to tout their product?

Not necessarily...20 Hz-20 kHz has always been the benchmark because that's what we can hear. A whole bunch of equipment can go lower and a whole bunch can go higher (in fact, in some analog designs a linear reponse well past 100 kHz necessary for good sonics in the audible range) but it's the audible range that's most important.

What I've never understood was that different phases for the frequency in question could radically alter the values being sampled by the ADC.

They actually don't...within the limits of the Nyquist frequency the sampled waveform can be accurately captured in the correct phase.

Take for example - an ADC running at 48kHz samples a 10V cosine wave (first sampled value will be 10, next -10, third 10, etc). What if this component hit the ADC at a 90' phase? That would make it a sine (or inverted sine), which means the ADC would be sampling a whole row of zeroes.

What you're talking about is a pure sine wave at the Nyquist frequency. In a practical sense that will never happen, because that frequency will have been filtered out. But in actuality, even if we had a theoretically perfect A/D converter that could sample a 24 kHz signal at 48 kHz but not a 24.001 kHz signal, the only instance where this wouldn't work would be if it was a pure sine wave that was being sampled. If you're sampling a complex waveform that includes 24 kHz information, that information would be reproduced in correct phase and everything because the reconstruction filters in the D/A converters would know exactly how to fit the "curve" (signal) to the "points" (samples) based on the samples that precedeed and followed that 24 kHz information.

Or do ADCs actually compensate for this phase problem and automatically shift these high freqs so that the max amplitude is sampled?

Sort of...the ADC captured the "points" on the curve, and the DAC reproduces that curve because there is one and only one curve within the limits of the Nyquist frequency that can possibly fit those points. it doesn't need to shift anything.

A related problem also exists for frequencies that approach the Nyqist limit - A signal with a frequency slightly below half the sampling rate will appear to be that with a frequency of exactly half of the sampling rate, with fluctuating volume. (It's hard to explain this - it's better if one can get ahold of a signal-generation program and try it out).

No, that does not happen. The only thing that could theoretically go wrong would be if you were to sample a sine wave at the Nyquist frequency, which could be interpreted as no signal at all. But since there would be no reason to ever do that it's not an issue.

Of course, these problems only become significant as frequencies approach the Nyqist - but I'm going to take the plunge and say that even my unprofessional ears can hear subtle differences between 44.1 and 48 - contradictory to what the textbooks say about 44.1 being able to sample up to 22.05 and human hearing being able to go up to 20 on a good day.

If you're hearing any difference between 44.1 and 48 kHz, it's either an old converter or a poorly-designed one. With today's oversampling converters there's no reason for there to be any sonic difference...all you get is an extra tenth of an octave way up in the overtone range. Back in the "old days" with primitive (by today's standards) filter designs things were different, but today there's just no excuse.

Want a real dose of Blasphemy? I compared recording at 96kHz and Sample Rate converting down to 44.1, to simply recording at 44.1kHz. I couldn't believe my ears! The track originally recorded at 96kHz and Sample Rate converted down to 44.1kHz had much better sounding highs, maintaining much of the character from recording at 96kHz

Hey, if your converters sound that much better at 96kHz, and whatever you use for sample rate conversion doesn't butcher the sound in the process (which many do) there's no arguing with that.

Taken from:
http://www.prorec.com/prorec/articles.nsf/files/4AE9C107C78E706886256688000FBE08
If this is true then it's worth it?

For that particular card mentioned, sure it may be worth it. But keep in mind that that article is five years old. A lot has changed in converter design since then.

a) what about the SRC down to 44.1k/16?
There would be 2 conversions: 96k/24 to 44k/24 then to 44k/16k. so?

Well, that's two issues...sample-rate conversion and dithering (or truncating). If done right, both can be fairly transparent...but especially as far as sample-rate conversion is concerned, most of the cheap stuff out there has audible artifacts that can easily negate the advantages you gained recording at a higher resolution, theoretical or otherwise.

b) why don´t they start working with bigger wordlenghts, let us say moving to 32 bits?

Because it's totally unnecessary to capture audio at 32-bit resolution. We can't even take full advantage of what 24-bit resolution has to offer...the best converters only give us approximately 20 bits' worth of dynamic range. Every system I'm aware of these days will process at a higher bit depth, which is necessary...but there's no need to capture the information at 32-bit resolution.

It is much better to record at 44k/24 bits than at 48k/16 bits, agree?

I don't know about "much" better...it many cases it certainly can be, but if you're recording a signal with only 20 or 30 dB of dynamic range, even 24 bits are overkill.

Though i truely doubt the general public with their cheap speakers and even cheaper headphones that settle for listening to 128kb mp3s are gonna give a crap its still gonna have to happen that were gonna need to be ready for 96 k audio.

There's a difference between being able to hear the difference and giving a crap. That same kid that loves the sound of his downloaded MP3's and has his five surround speakers on five different horizontal planes is the same kid that will get upset when he buys a DVD that doesn't say "96K" on it because more must be better, right? The power of marketing...

This to me is about when i laughed my ass off when roland came out with their first 96 k digital speaker now that was funny "a speaker that can reproduce all of digital signals" and im thinking wow a speaker can understand the diffrence between digital signals and analog

That speaker was simply an analog speaker with a D/A converter built into it. Roland weren't the first to do it. But if you have a system with inferior D/A converters, or one of Roland's systems that was designed to take advantage of those speakers, it could be a valid feature. You can't plug a digital cable into most powered speakers...well, you can, but the results won't be pretty.

Dithering filter roll off at half the frequence, which is fine at 44100... half is just about 22050, which is beyond audiable range...

You're actually thinking about anti-aliasing filters...dithering has nothing to do with sampling rate...but yeah, that's right...

but there's other stuff happening in the converting process, such as the slope of the filter,

And that will have much more to do with the sound of a converter than the sampling rate itself.

and if the sample rate is higher, you can make that filter slope more gentle... thereby affecting the way your stuff is converted...

True, except today's oversampling converters sample at such a high rate that a very gentle analog filter can be used.

So do all converters/clocks with the same sample rate have the same slope, or is it the degree of the slope that separates: the good the bad and the ugly??

I don't know if it's the "degree" of the slope as much as the quality of the filters, but yes, that's really what it is. Or a huge part of it at least...there's also the clock, the rest of the analog circuitry, the power supply...

Well, IMO it is a fallacy that we all only hear to 20K. There have been studies that show that even though the higher frequencies are inaudible we can somehow sometimes still sense them.

Well, some people can hear a little above 20 khz, but I don't know of any studies that have proven that we can sense those higher frequencies...there's one oft-cited study where they had monitors hooked up to test subjects and those monitors showed that our bodies did indeed indicate that those higher frequencies were present on a subconscious level, but none of the test subjects could actually tell (neither "hear" nor "feel" nor "perceive") when those frequencies were present.

And I know of engineering “Golden Ears” who have extended hearing well beyond 20K.

Who? Again, there's a big difference between being able to hear the artifacts of distortion at 50 kHz and actually being able to hear up to 50 kHz.

I test to about 16K with a bit of a loss at 7K but I can still hear a difference between 48K and 96K.

Which just goes to show that it's not the presence of higher frequencies that make a difference.

I think the manufacturers benefit more from this whole upgrade thing that any one else.

True, without a doubt...although if you have the chance to go to the AES convention this fall, talk to the engineers for some of the higher-end converter manufacturers...you'll be surprised at how many of them will tell you that their converters sound identical at all sampling rates, and that the whole 96kHz (and up) thing is more marketing than anything else.

I also think higher sample rates help out ALL frequencies, not just the super-sonic.

They absolutely do not.

An 80hz wave sampled at 96khz as opposed to 44.1 khz still contains more than twice as much information as the 44.1khz sample, even if 44.1k is technically (more than) enough to capture the signal.

All of that extra information is redundant. Once you've run that 80 hHz wave through a D/A converter, it will be identical at 44.1 khz, 96 kHz, 192 kHz, 32 kHz...or 161 Hz, for that matter.

If you have a frequency at say 12khz, at a sample rate of 48khz the sampler is only capturing information on that wave (the 12khz) 4 times per oscillation.

Although it's "only" capturing the information four times per oscillation, that's twice as much as is necessary. The Nyquist theorem has been proven. It's not a theory.

The point is that the benefit of recording at 192 or even 96khz is that say at 96 khz the same 12khz freq is getting captured 8 times per cycle instead of just 4, and at 192 it would be 16 times, so the information that the computer or playback device has to recreate the wave upon D/A conversion is significantly higher, we have more information about the state of the wave more often and thus the higher frequencies should be better and more fully and accurately reproduced.

That does seem intuitive, but there's no science behind that statement. And it's not right. 48 khz is more than enough to perfectly capture and reproduce that 12 kHz wave. You can't "more accurately" reproduce something that's been perfectly reproduced.

Theoretically with equal quality converters 192 should sound a lot better than 44.1

Nope. Theoretically it should sound identical, unless you can hear higher than 22 kHz. In reality it may not be that way, but that has more to do with other factors than the sampling rate itself.

and if you then sample down with the best downsampling algorithms possible, the end result should still be that the 192 downsampled recording should be better than the straight 44.1 recording, because there was more information captured in the first place and the algorithm did a good job of retaining and most important aspects of the information.

While it's true that there was more information captured, all of that information was above 22 khz and will be gone when you downsample to 44.1 khz.

The quality of the filters used certainly can and will affect the quality of the recorded audio within the audio range...and what you've just described is pretty much how oversampling converters work, except that they work that way because they overcome the limitations of the analog filters, not because they capture the >20kHz waveforms with any more "resolution".

Kind of like dithering from 24bit to 16bit, a 24bit recording dithered down sounds better than a 16bit recording under otherwise identical circumstances, but it depends on the dithering algorithm at the end, ie. you probably shouldn't just truncate it.

No, it's actually quite different. A 24-bit recording that exploits the full available dynamic range will sound noticeably better than a 16-bit recording that approaches 96 dB in dynamic range, and if dithered properly, you can hear that extra low-level information "below" the noise floor. However, a 96 kHz signal SRC'd down to 44.1 kHz loses all ultrasonic content.

-Duardo

Interesting response... believe it or not, I read that whole thing through. What you said about my points regarding Nyquist make a lot of sense.