[OddE]

  One hell of a convenient way to shut out the test to most people. Why don't you record it for us, if you don't mind?

 
-Oh, yes, please. I'd be more than happy to sox the heck out of the original, convert, downsample, drop bits &c to my heart's content, then turn everything back to 24/192 and let people try to listen for differences.
 
Downside to doing this over the Internet, of course, is that anybody can have a look at the files in, say, a suitably plugined Foobar and tell which sample is which.

 

[KeithEmo]

  @bookman you're still mixing mp3 bitrate with pcm bitrate, so I rest my case. you don't understand anything of what is said to you however how many times it's done. you're obviously a bot and an admin should remove your account to prevent more spamming of the exact same mistaken message over and over again.
 
@keith and of course you buy whatever you want, and there is nothing wrong with getting the perfect reproduction.
but I really don't see the rational behind paying for something I can't hear. if using only my human senses I fail to hear a difference and have no way to know unless I look at the numbers. how is it different from suggestion and placebo? at least when I buy a screen with a better gamut, it's something within the threshold of my senses that will be upgraded. I may find the upgrade meaningless and not worth the money, but it will be noticeable. with highres the changes are clearly outside of my hearing threshold. just like radio frequencies are outside of my hearing threshold and I wouldn't care to pay for an album that includes them just like I wouldn't pay more to have infra red on my screen. if it's outside of what my senses can get, it's outside of my subjective experience and audio is only that to me. I don't care about the life of the singer, and I don't care if there are plenty of sound at 24khz. that's not part of my experience of the music.
so while I certainly understand your reasoning, I don't share it.
 
my second problem with your post is the sound system. how many amps can resolve highres files at normal listening level into a real load? and of course the obvious, how many transducers can hope to get close to even cd quality? so what are we talking about here when we say perfect copy? if in the end our sound system is way below the CD resolution, and we fail to hear a difference, it's really just paying more for an idea.

 
Actually, I can think of several "rationales" for buying something that doesn't sound "provably" different to me - some practical, some potentially practical, and some just a matter of "self satisfaction"....
 
1) Even assuming that I can't hear any difference today, I might actually be able to hear one later. While I don't subscribe to the whole "golden ear idea" in general, it is true that our abilities do change over time, and we do sometimes "learn how to listen better". I may attend a live concert and suddenly discover that the speakers I thought sounded "just like live" really don't, or I may simply start paying more attention to certain aspects of the sound that I hadn't noticed before, or I may decide I like binaural recordings of chamber music with lots of natural ambiance as well as multi-tracked pop-rock. (I have one recording that has an odd little sound at one point, which I had always assumed was simply the recording microphone clipping; it turns out that it's a vibration coming from part of the drum; once I realized that, and heard a real drum make that sound, I started noticing whether it sounded natural or not on my speakers and headphones.)
 
2) Even assuming my abilities don't improve, I might buy better recordings someday, or change other equipment which renders the difference audible. (Certain speakers or amplifiers tend to make certain errors more audible - whether because they're more accurate, or simply because they emphasize them. For example, bright speakers make recordings with poorly recorded high-end sound more obvious. And many people here will surely attest to the fact that they notice things when listening on headphones that they don't when using speakers.) 
 
3) I may have a current or future technical justification. When I take pictures with my camera, a high-quality JPG (lossy) version of most pictures will often look just as good as a true lossless RAW frame - out of the camera. However, when I try to adjust it later in Photoshop, the artifacts on the compressed picture will often become obvious due to the processing. Most of us don't "re-master" our music, but many of us do use "processing" like surround-sound decoders, or spatial processors, or even noise removers, some of which may be affected by differences we can't hear, and they may be affected in ways that we can hear. (To use an example from the days of vinyl and SQ surround sound. The decoders used to play SQ encoded material use phase relationships to decide which parts of the audio belongs in which speaker. In many cases, if you have a record that has been mechanically damaged, the distortion from the damage is "pushed into the rear channels" at a boosted level by the decoder, which can make an album that sounds only slightly damaged in stereo virtually unlistenable through the decoder.) In the current context, next year's surround decoder may use high frequency phase cues present in the music to locate various instruments, and so may work with 96k recordings and not 44k ones. (While I agree that we can never know for sure where that would end, making recordings with at least a little bit of safety margin seems prudent.) 
 
4) Not all "limits" are as black-and-white as many people think. To take your example; most people would agree that buying a monitor that would display a gamut up to 850 nm would be silly (that's the "color" used by many IR remote controls). However, to say that "you can't ever see it" is wrong. In fact, light of that color is visible to most people, but only if it's bright enough. (If you look at the dot from "an invisible 850 nm LASER", you will indeed see it as a faint pink visible dot, because it is in fact slightly visible to most people.) In that case, I would agree that being able to see that color on your monitor probably serves no useful purpose, but I'm not so sure that everything that isn't directly audible is "useless".
 
5) Sometimes extra "safety margin" serves other benefits. For example, even though most of us probably don't hear much above 20 kHz, someone with an engineering background would still avoid an amplifier that was only able to amplify "20 Hz to 20 kHz" or that had a distortion plot that rose sharply right above 20 kHz - because good performance up to 50 kHz or so almost always signifies excellent performance inside the audio band, and performance that fails to extend past 20 kHz tends to suggest that problems exist inside the "audio band", even though they may not be visible on standard measurements. And, for another possibility, there was a recent AES paper that suggested, although I wouldn't say that it rose to the level of proof, that some people notice shifts in the sound stage on recordings that are band-limited to 20 kHz. (Their test showed that, even though their test subjects reported that the recordings "sounded the same", the location of instruments in the sound field was sometimes shifted when a recording was band-limited to 20 kHz. They suggested that, even though a 44k sample rate can record all audible sound, it may not be able to record the phase cues that our brains and ears use to determine location accurately enough.
 
6) As for your final point..... I simply disagree with the premise. Many of us get better equipment "as we progress in the hobby", and the technology itself improves. It would be foolish to buy something that is audibly inferior simply because my current system isn't able to let me hear its flaws, when the system I own next year may make them obvious. (I hear lots of details on my electrostatic headphones that I never noticed before on my speakers.) It would even be foolish to buy something that doesn't sound any better on ANY system available today, if that situation is likely to change. Twenty years ago you couldn't buy a TV that would let you see how much better the picture is on a Blu-Ray disc than on a DVD; yet the difference was really there, and now can be seen on most TVs.
 
In that last situation, if I bought my entire collection of movies as DVDs when they were "the current technology", I might end up buying them all over again as Blu-Ray discs. However, if I'd had the opportunity to buy them as a direct copy of the digital theatrical master, which is better than both DVD and Blu-Ray discs, then I would still have "the best copy available". (That option isn't available for video, but it is equivalent to buying a 24/96k or 24/192k copy of the audio master.)
 
I know lots of people who bought a significant amount of music on 128k AAC files then, after upgrading their music system, and realizing that the difference was audible to them after all, ended up having to buy it all over again (or pay the upgrade fee). Buying a version that's "a lot better than our ears" rather than one that's "just barely better than we believe is audible" seems like a good form of insurance against that (and, in most other situations, most people I know would consider a "safety margin" to be a good thing.)
 
I'm going to offer two very different examples - in other contexts - to support my point.
 
1) There are several devices designed to automatically remove ticks and pops during vinyl playback. Many of them use the ultrasonic content of the audio signal to decide the difference between a "legitimate tick" and a record scratch; because scratches have significant ultrasonic content while music recorded on records does not. You could use one of these devices (or equivalent software) to remove ticks and pops from the archive recordings you had made of your favorite albums if you'd made those recordings at 96k. However, it wouldn't work if you'd recorded them at 44k because the ultrasonic content the device relies upon would be missing.
 
2) Since you mentioned monitors and visible gamut.... There is a system quite similar to the click and pop remover that is commonly used to automatically detect and repair scratches on slides. It works by recognizing that certain colors of light that are invisible to the human eye are blocked by the surface coating on slide film. (Basically, by scanning the slide at these frequencies, you can produce an "image" of scratches in that surface, and use that information to control the correction process.) Of course, you could only use this system on a scanned and stored image if it included a gamut much wider than the range of the human eye. (So, if someone was archiving important photographs before this system existed, it would benefit them today if they'd scanned them using IR and UV light as well as visible light, even though, at the time, there was no apparent reason to do so.)
 
To me, all of this simply suggests that "getting the best quality copy you can afford" rather than "one that's just good enough" really does make sense. Now, in that second example, it might not pay to buy a special scanner to record a whole lot of information you might never use. However, in the case of music, where there is already a "starting point" that is limited by the ability of the microphones and mixing equipment we're using, it seems to make sense to hold onto a little extra information - since we already have it anyway - just in case we may want it later.
 
 

 

[Ruben123]

You wanna do a test?  Set up 3-4 mics on a drum set. Botnick style or however you choose.  Or hell, 1 mic hanging over the snare periscope-style will do the trick.

Record at the highest resolution your AD converters can do.  Mine do 24/96 reliably (24/192 is still wonky on my rig).

Just hit the snare hard 4 times. Then do some rolls on the snare. Then 30 seconds of hi hat work. Open and close the high hat as you roll on it. Now for your big finishing move, smash each cymbal and let it decay all the way out. Get to silence. Then do a few more  with some stick work on the top of the cymbal.  

Overall, record about 4 minutes of smashing drums.  No compression, no effects, just raw microphone data with basic stereo pan if using more than 1 mic. If you can play beats with fills go for it, but it's not needed.

OK back to the computer -- mix it to 2-track stereo and output it at native resolution. Then dither and downsample it down to 16/44. Make different copies for as many dithering algorithms as you have on your rig (I have 3 installed).

Now go back to your full resolution version and study it.  Listen to how long the decays are. Listen to the high hat detail (there's tons of it). Listen to the attack and decay of the snare. Pay attention to how the snare hits you, how much snap it has. Listen for stick noises, breathes, room hum, ambient sound off mic, etc. Determine the virtual placement of the drums in the mix, make notes if you want. How wide is the soundstage? Where is that drum/cymbal located at in the mix? 

Once you've made yourself familiar with the drum sound at full resolution, picking out the degradations in the 16/44 files should be easy.

You will hear everything get smaller. All decays will get shorter and harsher, with a cut off at the end that wasn't there.  Room noise and breathes will also be less than before. The soundstage will get slightly narrower and the instrument placement will be slightly fuzzy and appear less focused on one particular spot. The detail of the high-hat will start to be compromised.

If you continue this test and take a 16/44 file into LAME and make MP3s out of it, you will hear your beautiful acoustic drums turn into samples of cardboard. You will hear your full splashy cymbals turn into 808-sounding digital recreations of a cymbal.

This happens for almost every instrument as you degrade the signal. Given that multitrack recording is the art of making between 4 and 200 tracks all work together in a layered and musical way this degradation of every instrument really takes it's toll on the final mix.

You are hearing the effects of ultrasonics on music, and what happens when you remove them.  This is why musicians that spend big money learning their craft and recording it are all for high resolution digital. They want what they worked so hard making heard by the world. Why else bother making it?

With what headphones ? And could you do the same blindfolded?

 

[KeithEmo]

   
Yeah a good inline replying method eludes me :/
 
Hi-res price is tangential to the point of having tests and statistical outcomes from those tests that we deem as good enough evidence for audibility. And audibility is apropos to any conversation on the benefits and pricing of hi-res, because if we can't actually hear a difference with blinders on, then such non-benefits shouldn't be touted as rational for charging us even $1 more.
 
Re old equipment. Whatever attenuation old players actually managed, it's still true that many people simply don't need bone-flat response up to 20kHz, because our hearing drops off before then. I mean really, how bad could a filter that's aiming for 19k and allows aliasing up to 24k going to sound to those of us who are struggling to hear 18k at normal listening levels? But I was -1 for the first batch of CD releases, so people with personal experiences with the earliest equipment should chime in.
 
It's true, I expect a modern 24/192 master to outperform tape in every way. This still doesn't change the fact that quantizing many tracks to 14bit does squat to the signal, and that people who actually hear 20kHz seem a rare breed.

 
I sort of agree that touting the differences as "audible" if they aren't is somewhat misleading (even if they are otherwise "desirable" for other reasons). However, I think it falls well within the bounds of "ordinary advertising claims".....
 
As for the filters, the problem is that creating an analog filter with those characteristics is difficult, and manufacturing one that performs to specification is expensive. Even a "perfectly" designed filter is going to be a compromise between a sharp roll off and a smooth frequency and phase response (any practical filter that is sharp enough will almost certainly have unacceptable frequency response and phase ripple inside the audio band, and any filter with good frequency and phase response will not be sharp enough). And, once you have a decent filter designed, you're going to be stuck with another tradeoff between performance and cost - because the components accurate enough to manufacture an analog filter that precisely are extremely expensive. It's sort of like being tasked with designing a racing car that can go 500 MPH AND get 100 MPG - and then sell it for $35k. I'll bet that the older recordings that sound really good were done with a filter that accepted a slight loss of high frequencies in return for a smoother overall frequency and phase response.
 
Modern oversampling technology totally avoids those tradeoffs (at the cost of a few other minor ones), but oversampling was only developed later, as a way to avoid those compromises, after the standard was set.

 

[KeithEmo]

With what headphones ? And could you do the same blindfolded?

 
As for which headphones..... If you can hear the difference with ANY headphones, then that proves that the difference is real, at which point you're just testing headphones. Personally, whether you happen to like the way they sound or not, I've found electrostatic phones to usually do the best job of making differences audible and exposing tiny details.
 
Sadly, most of the commercially available test discs are in fact Red Book CDs, although a few are starting to appear at higher resolutions.
(A lot of places who sell high-res downloads also offer various sample discs and even free download samples.)

 

[RRod]

   
I sort of agree that touting the differences as "audible" if they aren't is somewhat misleading (even if they are otherwise "desirable" for other reasons). However, I think it falls well within the bounds of "ordinary advertising claims".....
 
I'll bet that the older recordings that sound really good were done with a filter that accepted a slight loss of high frequencies in return for a smoother overall frequency and phase response.
 
Modern oversampling technology totally avoids those tradeoffs (at the cost of a few other minor ones), but oversampling was only developed later, as a way to avoid those compromises, after the standard was set.

 
Agreed that it's hard to point to anything and say it's at the height of advertising shystering. Still, oversold is oversold.
 
As far as older recordings: I've read that pre-emphasis was used to help with older DACs, but the language isn't so clear on how exactly it helped either the digital or analog components. As far as testing out older tracks, one could do a comparison of relative energy in the last 1/2 or 1/3 octave for recordings of the same material, but eh. It would be great if someone had specs for older ADCs or DACs lying around. 
 
Certainly oversampling / Σ-Δ helped bring cheap flatness to consumer level products, but that doesn't mean that pro gear back in the earliest days didn't perform just as well, and I would think that companies like Telarc were using the best they could get their hands on.

 

[KeithEmo]

   
Agreed that it's hard to point to anything and say it's at the height of advertising shystering. Still, oversold is oversold.
 
As far as older recordings: I've read that pre-emphasis was used to help with older DACs, but the language isn't so clear on how exactly it helped either the digital or analog components. As far as testing out older tracks, one could do a comparison of relative energy in the last 1/2 or 1/3 octave for recordings of the same material, but eh. It would be great if someone had specs for older ADCs or DACs lying around. 
 
Certainly oversampling / Σ-Δ helped bring cheap flatness to consumer level products, but that doesn't mean that pro gear back in the earliest days didn't perform just as well, and I would think that companies like Telarc were using the best they could get their hands on.

 
You're sort of mixing apples and oranges now. Pre-emphasis is a process of boosting certain frequencies when you record them, then reducing them by an equal amount at playback, and serves the purpose of "noise shaping". If I know my tape is going to introduce high frequency hiss, I boost the highs when I make the recording. Then, when I play it back, I reduce the highs by an equivalent amount. This lowers the high frequencies in the music to the level they were at originally, but reduces the high frequency noise added by the tape. As long as the process is handled properly, there should be no net alteration in the sound. Pre-emphasis is usually used on tapes, and both Dolby noise reduction and RIAA equalization are really just very specialized types of it). There were some issues with early CDs because not all devices handle the pre-emphasis and de-emphasis properly (so some CDs end up sounding very shrill if they were recorded with pre-emphasis but end up being played without the matching de-emphasis - usually because the player doesn't support it or the flags inside the file aren't set correctly).
 
A simple high frequency roll off is actually easy to compensate for - by simply boosting the highs by the appropriate amount. (Which means that an ADC with a slightly rolled off high end is easy to correct for - by boosting the highs slightly in the mix. However, an irregular high end would be much more problematic.)
 
The usual issue with early filters was what we would call "an irregular response". I can have two filters that both "spec out" at "-1 dB at 20 kHz, but one may be flat to 15 kHz, then slope smoothly down to -1 dB at 20 kHz, while the other is flat to 18 kHz, -1 dB at 18.5 kHz, 0 dB at 19 kHz, -1 dB at 19.5 kHz, 0 dB at 20 kHz, and -3 dB at 21 kHz. Simple filters tend to have smooth response curves (like the first example), while the complex ones you need to get flat in-band response and a sharp drop off afterwards tend to have a lot of ripple (which is what that second example has).
 
Some early equipment did in fact perform very well - but not all of it. (And, yes, some early digital equipment had a tendency to be designed for "good specs" rather than sound quality.) And, in the beginning, that equipment was largely unfamiliar to most of the pros who were using it. So you ended up with equipment that was far from perfect, and which required someone to be knowledgeable of the various tradeoffs involved if you wanted to get the best performance from it, but which was often being operated by people who weren't intimately familiar with those quirks and tradeoffs. Add to that the fact that there was also a tendency to specifically adjust the sound to fit what "customers expected the digital version to sound like" so people would have incentive to replace their albums with CD versions, and you had a sort of formula for inconsistency.

 

[castleofargh]

@Keith I don't hold my breath for a headphone/speaker that out resolve CD. I most certainly hope I'm wrong and it will happen in my lifetime, but I really don't hold my breath.
 
nothing wrong about thinking about the future, and even though bookman wants to make me the champion of promoting mp3, I have never bought music in mp3 format and would always keep 2 archives of my music in flac. it's most likely unnecessary, but it's the kind of habit I got from my past mistakes in photography ^_^.
 
I agree with several things you said on principle as motives to buy higher resolution, it's simply that I don't see my own hearing with whatever listening experience I will gain, changes the fact that I won't be able to pass an abx. not that I'm pessimistic, just that I've seen enough people fail around the world not to think I'm some uberman who will do it. and as I said I don't think my sound system will in my lifetime outperform CD resolution. so just like I don't see the point of getting better than my odac given how much worse is the rest of my gears, I don't believe that highres can benefit me so I don't pay for it.
but of course it's my own judgment call and I'm perfectly fine with people thinking otherwise. if anything the idea of having the best available does appeal to a great many people. I'm fine with that.
I just wish the publishers would stop their game of restricting some very good masters to highres format.
 
 
 
 
 
 
about booky's pono format test, here is the post that got deleted and got me locked out of the pono topic.. because "hearing test should be kept to sound science" and we were warned about it. fear all the evil of my post!!!!!!!!!  I was answering to a guy asking about the very pono listening test and asking if people heard a difference. so obviously it was easy for me to explain how it wasn't a proper test without explaining what is...

I'll end up banned soon because you only get so many strikes before being out. but the idea of being vetoed for posting something both on topic and factual, while half of the posts before me were pure fallacies, false claims and anti science hatred, that really tells a lot about what currawong thinks and cares about. and good for us, that's who we have to do most of the moderation in the sound science section \o/ yay!!!!!! go science!!!!
anyway here it is, don't let your kids read it, it's hardcore stuff, or really not. maybe ask your favorite moderator:

except that the "revealer" fails to provide unbiased observation. the prime point of making a correct comparison is not to know what kind of file you're listening to. just like when tasting 2 cakes, you don't tell people that the first one cost 15$ and the second one cost 55$, because from there, even if the 2 cakes are in fact the same, more people will pick the expensive one as being better. demonstration:

 
it's the same kind of problem we have here, the fact that the user is the one selecting the format and knowing what he's listening to is a very significant bias.
it doesn't matter what kind of difference there is between the files, we all know they are different that much is a fact. what matters is to be testing the sound and not the sound+our preconceptions of it.
 
ps: if one of you would be so kind as to record and upload the different iterations of a little piece of music in the highest resolution you can set with your soundcard, that would be cool(not too long to avoid infringing on copyright so maybe 10 or 15s). so that we could check if things are done correctly like Young said. call me a skeptic(well I am), but after the fake online tidal test, I feel like it's a legitimate thing to check first. and if it's fine you all will know you can rely on it a little better. win/win IMO.

 

[RRod]

   
You're sort of mixing apples and oranges now. Pre-emphasis is a process of boosting certain frequencies when you record them, then reducing them by an equal amount at playback, and serves the purpose of "noise shaping". If I know my tape is going to introduce high frequency hiss, I boost the highs when I make the recording. Then, when I play it back, I reduce the highs by an equivalent amount. This lowers the high frequencies in the music to the level they were at originally, but reduces the high frequency noise added by the tape. As long as the process is handled properly, there should be no net alteration in the sound. Pre-emphasis is usually used on tapes, and both Dolby noise reduction and RIAA equalization are really just very specialized types of it). There were some issues with early CDs because not all devices handle the pre-emphasis and de-emphasis properly (so some CDs end up sounding very shrill if they were recorded with pre-emphasis but end up being played without the matching de-emphasis - usually because the player doesn't support it or the flags inside the file aren't set correctly).
 
A simple high frequency roll off is actually easy to compensate for - by simply boosting the highs by the appropriate amount. (Which means that an ADC with a slightly rolled off high end is easy to correct for - by boosting the highs slightly in the mix. However, an irregular high end would be much more problematic.)
 
The usual issue with early filters was what we would call "an irregular response". I can have two filters that both "spec out" at "-1 dB at 20 kHz, but one may be flat to 15 kHz, then slope smoothly down to -1 dB at 20 kHz, while the other is flat to 18 kHz, -1 dB at 18.5 kHz, 0 dB at 19 kHz, -1 dB at 19.5 kHz, 0 dB at 20 kHz, and -3 dB at 21 kHz. Simple filters tend to have smooth response curves (like the first example), while the complex ones you need to get flat in-band response and a sharp drop off afterwards tend to have a lot of ripple (which is what that second example has).
 
Some early equipment did in fact perform very well - but not all of it. (And, yes, some early digital equipment had a tendency to be designed for "good specs" rather than sound quality.) And, in the beginning, that equipment was largely unfamiliar to most of the pros who were using it. So you ended up with equipment that was far from perfect, and which required someone to be knowledgeable of the various tradeoffs involved if you wanted to get the best performance from it, but which was often being operated by people who weren't intimately familiar with those quirks and tradeoffs. Add to that the fact that there was also a tendency to specifically adjust the sound to fit what "customers expected the digital version to sound like" so people would have incentive to replace their albums with CD versions, and you had a sort of formula for inconsistency.

 
The wiki page just didn't make clear the connection to filtering; from what you say it sounds like it was only used for abating noise.
 
A bit of variation between 0 and -1dB up above 18kHz may be "bad" in the universal sense, but I doubt that is what people are hearing as sounding bad. If I play a mix of a 12kHz and 18kHz tones, I can't tell you when the 18k pops in. That's not justification for having non-flat response, of course, but it does illustrate a point about audibility versus measurement.

 

[Thad-E-Ginathom]

Mr Bookman, if you really would like to make your test files as prescribed ---and perhaps Arny will tell you the technical stuff that you have to comply with--- there are plenty of folk who will take the test.
 
I'll even try myself, but with audiometry showing failing hearing from 8kHz, the normally-audible spectrum is more more than I can cope with.
 
Really. People would accept the ABX challenge willingly.  But... you go first.

 

[RRod]

  Mr Bookman, if you really would like to make your test files as prescribed ---and perhaps Arny will tell you the technical stuff that you have to comply with--- there are plenty of folk who will take the test.
 
I'll even try myself, but with audiometry showing failing hearing from 8kHz, the normally-audible spectrum is more more than I can cope with.
 
Really. People would accept the ABX challenge willingly.  But... you go first.

 
He won't accept ABX as the means of testing it.

 

[Thad-E-Ginathom]

Well, that's that then.
 
My "end of conversation" was only a few posts premature.

 

[Gr8Desire]

...  
You wanna do a test?  Set up 3-4 mics on a drum set. Botnick style or however you choose.  Or hell, 1 mic hanging over the snare periscope-style will do the trick.
 
Record at the highest resolution your AD converters can do.  Mine do 24/96 reliably (24/192 is still wonky on my rig).
 
Just hit the snare hard 4 times. Then do some rolls on the snare. Then 30 seconds of hi hat work. Open and close the high hat as you roll on it. Now for your big finishing move, smash each cymbal and let it decay all the way out. Get to silence. Then do a few more  with some stick work on the top of the cymbal.  
 
Overall, record about 4 minutes of smashing drums.  No compression, no effects, just raw microphone data with basic stereo pan if using more than 1 mic. If you can play beats with fills go for it, but it's not needed.
 
OK back to the computer -- mix it to 2-track stereo and output it at native resolution. Then dither and downsample it down to 16/44. Make different copies for as many dithering algorithms as you have on your rig (I have 3 installed).
 
Now go back to your full resolution version and study it.  Listen to how long the decays are. Listen to the high hat detail (there's tons of it). Listen to the attack and decay of the snare. Pay attention to how the snare hits you, how much snap it has. Listen for stick noises, breathes, room hum, ambient sound off mic, etc. Determine the virtual placement of the drums in the mix, make notes if you want. How wide is the soundstage? Where is that drum/cymbal located at in the mix? 
 
Once you've made yourself familiar with the drum sound at full resolution, picking out the degradations in the 16/44 files should be easy.
 
You will hear everything get smaller. All decays will get shorter and harsher, with a cut off at the end that wasn't there.  Room noise and breathes will also be less than before. The soundstage will get slightly narrower and the instrument placement will be slightly fuzzy and appear less focused on one particular spot. The detail of the high-hat will start to be compromised.
 
If you continue this test and take a 16/44 file into LAME and make MP3s out of it, you will hear your beautiful acoustic drums turn into samples of cardboard. You will hear your full splashy cymbals turn into 808-sounding digital recreations of a cymbal.
 
This happens for almost every instrument as you degrade the signal. Given that multitrack recording is the art of making between 4 and 200 tracks all work together in a layered and musical way this degradation of every instrument really takes it's toll on the final mix.
 
You are hearing the effects of ultrasonics on music, and what happens when you remove them.  This is why musicians that spend big money learning their craft and recording it are all for high resolution digital. They want what they worked so hard making heard by the world. Why else bother making it?

Umm... Err...  People have been recording drums since the beginning of the digital era. 

16 bit PCM** easily covers the dynamic range of every drum set I have ever encountered - including the insane range from the PCM-out port on my Roland V-Drums.  I have done thousands of such recordings.

The problem is NOT the number of bits.  It is the playback devices. If I use the full compliment of 16 PCM bits, the music is either too quiet or it distorts too much on loud passages. There is no solution in the digital realm (i.e., more bits or samples will not help). The analog devices that reproduce the sound cannot adequately cover the dynamic range of 16 bit PCM. Oddly enough, 24 bits helps marginally. 24 bits has better resolution by offering more samples within the 16 bit PCM range. Range beyond 16 bits is not really useful - again because analog devices can't reproduce all of them.

BECAUSE OF ANALOG PLAYBACK DEVICE LIMITATIONS VIRTUALLY EVERY DRUM RECORDING IS COMPRESSED to FIT INTO ABOUT 12 PCM BITS.  

Someday we might create better playback devices (BTW: I'm thinking mainly speakers and HPs).  Until then, recordings and remastering efforts will continue to use little of the dynamic range available.
  
** I refer to 16 BIT PCM (LPCM really) because it defines linear spacing between adjacent quantized values (same distance between quiet and loud values) and therefore uses its 16 bits VERY INEFFICIENTLY. Other encoding standards use variable spacing between quantized values focusing more bits where there is actually sound.  As such, they need even less than 16 bits to cover all recording requirements. 

 

 

[KeithEmo]

   
The wiki page just didn't make clear the connection to filtering; from what you say it sounds like it was only used for abating noise.
 
A bit of variation between 0 and -1dB up above 18kHz may be "bad" in the universal sense, but I doubt that is what people are hearing as sounding bad. If I play a mix of a 12kHz and 18kHz tones, I can't tell you when the 18k pops in. That's not justification for having non-flat response, of course, but it does illustrate a point about audibility versus measurement.

 
The purpose of filtering in A/D and D/A conversion isn't at all intuitive.... so perhaps a proper explanation would be in order.
 
I'm sure you're familiar with the basic fact that a digital audio file can ONLY be used to store information up to just under the Nyquist frequency - which is 1/2 of the sample rate. For a 44.1 kHz file, the Nyquist frequency is 22.5 kHz, which is why a CD can't contain any information above that frequency (and using 20 kHz as the cutoff frequency does give a tiny safety margin there).
 
However, in reality, any source is going to contain some information above 22 kHz; which will include high order harmonics produced by some instruments like cymbals, as well as actual noise from equipment like preamps and other electronics, hiss from analog master tapes, and even noise present in the room where the recording was made. The problem is that this extra "unrecordable" content doesn't simply disappear when you feed the content into your ADC. In fact, the opposite is true; if there is any content present above the maximum frequency that can be encoded "properly", it is converted into a very audible and unpleasant distortion during the encoding process. The actual process involved, and the exact results, are somewhat complicated, but the net result is that a significant portion of it is "folded back around the Nyquist frequency" into the audio INSIDE THE AUDIO RANGE OF THE ENCODED AUDIO.
 
Let's take our CD as an example: the sample rate is 44 kHz, and the Nyquist frequency is 22 kHz. Now let's assume that, mixed in with my "audible source material", there is a 28 kHz tone (it could simply be some ultrasonic harmonic of some instrument, or just some 26 kHz hiss in the microphone preamp). If I were to feed that source into an ADC which lacked the proper filtering, that 28 kHz ultrasonic would be "folded down around the Nyquist frequency". This means that most of the energy contained in that 28 kHz tone, which starts out being 6 kHz ABOVE the Nyquist frequency (22+6 kHz), and which cannot be encoded into our output file, will be converted by the encoding process into an equivalent amount of energy 6 kHz BELOW the Nyquist frequency (22-6 kHz). So the energy from our ultrasonic 28 kHz noise source, which was totally inaudible, will appear in our encoded file as noise at 16 kHz. This process will occur with all signal energy which is present in the source that is above the Nyquist frequency - and will essentially end up as noise/distortion in the final file. The actual process is somewhat "messier" than my simplified explanation, so that 28 kHz tone will actually cause noise spikes at other points inside the audible range. And, of course, I used a single tone as an example, while the reality is somewhat more complicated. (Think of it sort of like a distorted and inverted reflection in a camera lens interfering with the desired image.) The purpose of the filter in the ADC is to remove any content outside the frequency range which the encoder can handle properly so this doesn't happen (which means anything above the Nyquist frequency).
 
Unfortunately, most real-world recordings contain all sorts of energy above 20 kHz, ranging from harmonics of actual instruments, to room noise and such, and even to analog distortion products. ANY of this which the filter fails to remove will end up as distortion and noise inside the audio band in the resulting encoded file, which is why it is critical that the filter remove it - or at least reduce it to a very low level.
 
(Oversampling avoids this issue by shifting the Nyquist frequency upwards, thus making it easier to design a filter that is flat to all audible frequencies, yet still has sufficient attenuation above the Nyquist frequency. In order to make a "clean" conversion at 44 kHz, to the level required for a CD, you would need a filter that is flat to 20 kHz, yet down about 80 dB at 24 kHz, which is very difficult to achieve in practice. To get a similarly noise and distortion free output at a 96k sample rate, the filter would have to be flat to 20 kHz, and down 80 dB at 50 kHz, which is a far gentler slope, and so much easier to design and build.)  

 

[arnyk]

 
 
You wanna do a test? 

Yes, and it seems from the above obvious attempt to distract, that it is very likely that you don't want to do a valid test.
 
All you appear want to do is yet another sighted evaluation.
 
You're not dealing with a novice. I am fully aware with all of the well-documented fatal issues with that your ancient and highly flawed listening comparison technology involves, and have already documented it here many times.
 
I'd be glad to review the matter again for your benefit.
 
All this gratuitous discussion about microphone technique may impress novices, but my professional experience with live recording of music probably vastly exceeds yours, so of course I'm not the least bit distracted by it.