24/96 tracks

scott s

n00b
Joined
May 27, 2011
Messages
11
well I finaly downloaded a sampler from hdtracks and must admit that the sound to me is better than standard cd 16/44. This is thru a usb dac however, not my stock internal sound card. Wish more rock and modern proggressive jazz/rock titles were available though. Still enjoying my ripped to wav cd's though!
 
Yeah, those high sample rates are amazing if you're listening through the right gear.
 
I have loads of 24/96 vinyl rips and digital downloads. Couldn't agree more, however you do need the right equipment to hear it.
 
Last edited:
I find some of them are good, some of them sound like ass, as in not as good as the MFSL CD track. Mixed bag, obviously sample rate doesn't automatically mean better quality. I think there's more to gain from 24 bits than 96kHz, but it's really gonna require some accurate gear to tell. Personally I think 24 bit 48 kHz is the point of diminishing returns. Most important is not bunking up previously accurate recordings with lossy compression. We should stop distributing music compressed (lossy). Storage is cheap and lossless encoding works well now.
 
well I finaly downloaded a sampler from hdtracks and must admit that the sound to me is better than standard cd 16/44. This is thru a usb dac however, not my stock internal sound card. Wish more rock and modern proggressive jazz/rock titles were available though. Still enjoying my ripped to wav cd's though!

Try re-sampling those to 16/44.1 FLACs or even LAME V0 or 320 kbps mp3s, and then ABXing them. The vast, vast audible difference is in the recording and mastering, not the super-high bit/sampling rate,
 
I've found a lot of these DVD-Audio albums are complete crap. Most are just the same clipped loudness-war crap we get on CD and remasters, so I end up looking for MFSL type stuff, and original-release CDs. Basically, for the most part, if the music is in some kind of HD format, and actually sounds good, then it isn't anything I would ever listen to without wanting to kill myself (Jazz, Classical, off-the-wall Indie garbage, Folk) Anything more than 16/44.1 is pointless as an end for vinyl.
 
I have also started collecting HD FLAC rips recently, mostly 24/96 but I have a few that are 24/192 and they blow away the quality of mp3 and standard FLAC. There is simply vastly more detail and range, when I play them back in foobar I love watching the 20Khz+ spectrum graphs dance around - this is all information that CD simply is not capable of capturing that is missed.

But there also is a lot to be said for the quality of the master, a lot of modern music sounds terrible because pretty much everyone uses excessive compression these days. It gives the impression of sounding "good" on shitty equipment (like boomboxes and portable media players). I have actually read articles by people who make their own CDs using vinyl as a source, and they sound much better than their modern mastered commercial counterparts, the difference is plain to see in a spectrum graph.
 
...the sound to me is better than standard cd 16/44.
Yeah, those high sample rates are amazing if you're listening through the right gear.
...they blow away the quality of mp3 and standard FLAC. There is simply vastly more detail and range
Can you guys share with us the results of your double-blind tests?

I love watching the 20Khz+ spectrum graphs dance around - this is all information that CD simply is not capable of capturing that is missed.
All that glorious completely inaudible information.
 
Vinyl is far inferior to CD. I started collecting vinyl, and mostly new releases of pop stuff, and found it was the same, horrible, clipped, loud crap that is on CD. Very disappointing and a waste of money. Now I'm stuck with a bunch of newer vinyl albums that I don't want and that I just ended up downloading the CD version anyway. Maybe the record companies will deserve my money when they can produce something that doesn't blow out my eardrums and doesn't have all it's frequency content in bass. And for old stuff that is mastered properly, it's even more worthwhile to find the first CD release of pre-CD music. Mostly like the vinyl, but without all the noise from being played on archaic cheap equipment set up wrong, and collecting dust for 3 decades.

20 khz+ stuff may be inaudible as a single sine tone, but it doesn't mean it's not relevant information to the sound under 20 khz. Keep your Hydrogen Audio phooey out of this thread. If you don't like this HD stuff, then ignore it, be an adult.
 
All that glorious completely inaudible information.

Precisely. Allow me to enlighten you as to why infra/ultrasonic information is critical to our perception of sound. Simply put - it sounds more real, because it is more real. While you are entirely correct that from a technical perspective humans cannot "hear" frequencies below and above a certain range - we can PERCEIVE them through various means, such as feeling them in our bones when it comes to infrasonic frequencies, as well as perceiving the effect of resonance and interaction with objects around us. All of this adds up to a more true to life representation of a performance.

For a lot of people listening to music is an emotional experience, and while I am certainly no quack who would spew pseudoscience (this extended detail is quantifiable, as it can be plainly seen in a spectrum graph), there is a technical basis for our "unexplainable" preference for full-range sound reproduction.
 
Last edited:
A lot of sound perception is just that, perception...the brain's doing. Our "ears" are terribly dumbed down from how they were created to be. Our brain has to filter out so much noise we've manufactured into our environment that most people don't need quality music. Most people are dumbed down, anymore, and their senses show it. The ears can do amazing things, you just need to train your brain. Take some LSD and listen to some 64 kbps music. Your brain won't focus on the horrible compression, and it will actually sound good, and your mind will also be elsewhere. It also has the power to make good sounding music be even better.
 
20 khz+ stuff may be inaudible as a single sine tone, but it doesn't mean it's not relevant information to the sound under 20 khz.
In many cases the audible content above 20 kHz is predominantly the result of psychoacoustic dither — random noise designed to reduce quantization distortion, shaped (often aggressively) into the inaudible range. Typically (almost universally) this noise is non-harmonically-related, which means it has no relationship to musical content and is thus pretty deserving of the term "noise". The rest is generally comprised of very low-amplitude harmonic information, limited in many respects due to microphone limitations, and additional broadband noise introduced during the production phase. There will almost always be some content beyond 20 kHz that is relevant to the content below it, but it's only useful to us if it's perceptible. That's the key.

If you don't like this HD stuff, then ignore it
I have nothing against it.

While you are entirely correct that from a technical perspective humans cannot "hear" frequencies below and above a certain range - we can PERCEIVE them through various means, such as feeling them in our bones when it comes to subsonic frequencies, as well as perceiving the effect of resonance and interaction with objects around us. All of this adds up to a more true to life representation of a performance.
This is contingent on a transducer's ability to reproduce this content and at sufficient pressure levels to trigger resonance in objects in the room. While you are correct that so-called subsonic information can be perceptively relevant, it is again contingent on the ability to drive sufficient sound pressure to produce any kind of perceptible response (and a source with a higher sampling rate affords no meaningful advantage in this regard anyway). See above, also, regarding the type of information present in the frequency range beyond ~22 kHz.

Realistically, considering the average person is unlikely to notice when you tape a "kick me" sign to his back, said person is also pretty unlikely to perceive any kind of difference between audio reproduced with or without frequencies beyond ~22 kHz. If you have some research to show otherwise, though, I'd very much like to see it.
 
Just because your ears don't relay to your brain an audible tone doesn't mean it's imperceptible or unimportant. Just because you don't feel literal shaking doesn't mean some sense isn't relaying that information to your brain. There is a lot more to sound that hasn't been scientifically proven or put into a stupid theorem yet.
 
Part of the reason for higher sampling rates includes pushing any DSP-induced distortion into the higher (inaudible) frequencies.

If you do an audacity sweep with some of the higher sampling rate music files, you do see this, spikes in FR in the inaudible regions (50k+) which is likely where DSP distortion was pushed out to (this is not from the original recording as studio mics don't usually go much up above 40k)
 
Just because your ears don't relay to your brain an audible tone doesn't mean it's imperceptible or unimportant. Just because you don't feel literal shaking doesn't mean some sense isn't relaying that information to your brain. There is a lot more to sound that hasn't been scientifically proven or put into a stupid theorem yet.

I see. "Stupid theorems."

Just because some hypothesis hasn't been disproven doesn't make it reasonable to believe when there's both a more plausible explanation with evidence to support it and substantial evidence against said hypothesis. That doesn't necessarily mean that the hypothesis is false - only that it is not only very improbable that it's true, and also unreasonable to believe is the most likely explanation.

Read up on Open-mindedness.

Show me blind tests demonstrating a statistically significant link for high sample rates to audibility in music playback and it will peak my interest - and peer reviewed, preferably.

So far, all the evidence points the other way - the only passed blind tests between standard 16 bit, 44.1 kHz audio and higher resolution audio are a result of extremely high listening levels that reveal the slightly higher characteristic noise floor of 16 bit quantization versus 24 bit quantization. In few, if any, real life listening situations is that even applicable - almost every recording has a significantly higher noise floor to begin with, and these differences have only shown to be audible at very high listening levels.
 
Everyone wrongly dictates that the other 16 million-some points you gain from going from 16 bit to 24 resolution is only going to what is below the 16 bit noise floor just by using some logarithm math. I wish people would stop defending CD. It's not like they just came up with 44.1 khz based off of actual studies of hearing, and it's not like they would have still chosen 16/44.1 if they invented it these days with today's storage spaces. It was a tradeoff between quality and space. Do you really believe if they were given a Blu-ray disc and told to make an audio format for it that they would just choose something even as low as 24/48? You can use all your fancy jargon and theorems, but it's fact that you cannot perfectly reproduce an analog signal digitally. There will be loss. Just because a CD can reproduce tones up to 22.05 khz doesn't mean it can do so with more complex waveforms like music. And again I say, just because you can't pick out a tone at over the supposed human hearing threshold doesn't mean it isn't valuable information that your brain indeed perceives when it's part of other information. There are forces at work in this universe that we don't see or perceive, but when they affect other things, we perceive that. The brain also doesn't need our 5 bodily senses to get information. The CD format is outlasting any other format, and we are keeping around old technology just for the stubbornness of a few.
 
Everyone wrongly dictates that the other 16 million-some points you gain from going from 16 bit to 24 resolution is only going to what is below the 16 bit noise floor just by using some logarithm math. I wish people would stop defending CD. It's not like they just came up with 44.1 khz based off of actual studies of hearing, and it's not like they would have still chosen 16/44.1 if they invented it these days with today's storage spaces. It was a tradeoff between quality and space. Do you really believe if they were given a Blu-ray disc and told to make an audio format for it that they would just choose something even as low as 24/48? You can use all your fancy jargon and theorems, but it's fact that you cannot perfectly reproduce an analog signal digitally. There will be loss. Just because a CD can reproduce tones up to 22.05 khz doesn't mean it can do so with more complex waveforms like music. And again I say, just because you can't pick out a tone at over the supposed human hearing threshold doesn't mean it isn't valuable information that your brain indeed perceives when it's part of other information. There are forces at work in this universe that we don't see or perceive, but when they affect other things, we perceive that. The brain also doesn't need our 5 bodily senses to get information. The CD format is outlasting any other format, and we are keeping around old technology just for the stubbornness of a few.
We're keeping CD around, since most mastering equipment isn't much better, nor will most people care.
 
Keeping the CD around is pointless. Most people get their music as a lossy source, either legally, or illegally. The CD-ROM doesn't really even have much if any use anymore. How would it hurt the audio world if they went with a better format? It would make you cheapos who think hard drive space is too expensive cry.
 
Just because your ears don't relay to your brain an audible tone doesn't mean it's imperceptible or unimportant. Just because you don't feel literal shaking doesn't mean some sense isn't relaying that information to your brain.
If this is perceptible information — in any capacity — a properly-administered A/B test will indicate that. These tests can be difficult, relative only to the test taker and will never be 100% conclusive, but what it does is clears out any potential bias. If someone is going to promote the apparent perceptible advantages of higher-resolution audio, it stands to reason that apparent differences should be substantiated in some way so as to not potentially mislead others into a certain way of thinking based on what are merely bias-prone subjective impressions.

The listening experience doesn't have to be based on faith or on whimsical postulations of the finer workings of the human brain. This kind of thinking only abstracts away the reality of our perception and tries to elevate audio into some sort of magic or voodoo — which it isn't. Your emotional response is a separate issue entirely. If you feel more comfortable on an emotional level with HD audio, I have no quarrel with that. When someone extols the audible virtues of it, though, that's when I think substantiation is warranted.
 
24/96?
http://theaudiocritic.com/plog/index.php?op=ViewArticle&articleId=4&blogId=1

I have also started collecting HD FLAC rips recently, mostly 24/96 but I have a few that are 24/192 and they blow away the quality of mp3 and standard FLAC. There is simply vastly more detail and range, when I play them back in foobar I love watching the 20Khz+ spectrum graphs dance around - this is all information that CD simply is not capable of capturing that is missed.

But there also is a lot to be said for the quality of the master, a lot of modern music sounds terrible because pretty much everyone uses excessive compression these days. It gives the impression of sounding "good" on shitty equipment (like boomboxes and portable media players). I have actually read articles by people who make their own CDs using vinyl as a source, and they sound much better than their modern mastered commercial counterparts, the difference is plain to see in a spectrum graph.

Cds got a locked snr of 93db
 
Well when they use "psychoacoustic" means to compress audio for the dumb-eared/brained people to listen to, how is that not emotional or playing a function of brain? Again, until they make a test that favors both camps, I won't take this stupid ABX test. It, too, plays on emotions and other human factors.

1. No one's skull is the same shape, down to the atom.
2. No one's ears, inner, outer, middle, are the same shape, down to the atom.
3. No one's neural path from the ear to the brain is the same.
4. No one's brain is wired the same.

Thus, not everyone hears the same things.
 
Again, until they make a test that favors both camps, I won't take this stupid ABX test.
The value of an ABX test is that it favors no particular result. It's merely designed to test your ability to discern one sample from another. It strips potential bias and human emotion by obfuscating each sample such that you aren't aware of which is which, thus any biased preference for one over the other is effectively nullified. Beyond that it tests nothing, which is precisely where its value lies.
 
Everyone wrongly dictates that the other 16 million-some points you gain from going from 16 bit to 24 resolution is only going to what is below the 16 bit noise floor just by using some logarithm math.

Could you provide mathematical or empirical evidence supporting your claim? You're flying in the face of established science by claiming this. Higher bit depth reduces quantization noise, and thus lowers the noise floor. This is the same as increasing dynamic range - which doesn't matter since our recordings aren't limited by the dynamic range of 16 bit quantization in the first place.

I wish people would stop defending CD. It's not like they just came up with 44.1 khz based off of actual studies of hearing,

Yes, they did.

and it's not like they would have still chosen 16/44.1 if they invented it these days with today's storage spaces.

Ahh, but the vast, vast majority of audio is still distributed in 16/44.1 (or 16/48 in the case of computers/DVDs) despite being free from the apparently crippling fetters of the Redbook audio specification.

It was a tradeoff between quality and space. Do you really believe if they were given a Blu-ray disc and told to make an audio format for it that they would just choose something even as low as 24/48?

If it doesn't make an audible difference, as all the evidence so far has shown, you'd be wasting resources by making audio BluRays at high bit/sampling rates rather than a much cheaper format like Redbook CDs. Since most music is distributed via the Internet today, you skip the whole issue by making music available in whatever format and bit/sample rates that turn out to be popular.

You can use all your fancy jargon and theorems, but it's fact that you cannot perfectly reproduce an analog signal digitally. There will be loss.

Yes, and nor can an analog storage method perfectly reproduce an analog signal. Actually, it does a far less accurate job than digital storage of music in every case.

With analog storage, you've got wow and flutter that are many orders of magnitude worse than digital jitter. You've also got significant wear on the recording medium every single time the recording is played back. The analog mediums have real, although not easily quantified as with digital, resolution limits that affect the accuracy of the signal as well.

There's also the generational loss every time you copy the music from one storage format to another - from the initial reel to reel tape recording to the mixed master, all the way through all the intermediate steps to the final vinyl or tape final product. With digital recording, you skip ALL of that, and you can have a bit-perfect copy of the original master at home - with FAR more fidelity to the original analog input to the pre-amps in the studio than an analog recording. Oh, and if you want to simulate all of that distortion - you can do so, digitally. That includes any coveted even-order harmonic distortion.

Anyway, the loss you have regarding the original analog signal... Increasing the bit depth of course reduces the quantization noise, of course. Anyway, if you had infinite bit depth, and ignoring the limitations of DACs (since we're talking hypothetical and impossible situations here), you would reproduce the signal below the Nyquist frequency (after accounting for the necessary filters, of course) perfectly. Zero error, mathematically (again, since we're talking a hypothetical situation - there is no reality). You don't need a higher sampling rate to reproduce the lower frequencies, even relatively high 10-20 kHz ones.

The point is that at what point - what bit depth - does reducing the quantization noise no longer make a difference to humans? Except for extremely loud playback of recordings with very, very quiet sections, all testing done so far has indicated that by 16 bits we've reached that point. The quantization noise is very, very quiet indeed - to the point that it doesn't matter. Note that this is very different to any issues with the sampling rate - where of course we're talking about the reproduction of higher frequencies, not the reduction of quantization noise.

Just because a CD can reproduce tones up to 22.05 khz doesn't mean it can do so with more complex waveforms like music.

Yes, it does. The level of quantization noise is the same for any signal at the same level, no matter how complex it is. It doesn't matter if we're talking about a single 1 kHz sine wave or a symphony, if they're at the same level they're reproduced equally well in the digital (and back to analog via a DAC) domain with the same level of quantization noise. It's not like lossy compressed audio where data is thrown out or approximated when it's deemed inaudible or at least less audible - it's all there.

And again I say, just because you can't pick out a tone at over the supposed human hearing threshold doesn't mean it isn't valuable information that your brain indeed perceives when it's part of other information.

And that's why we do blind testing - to see if we actually perceive those things - higher frequencies and lower quantization noise (or other artifacts in the case of lossy compressed audio, as well as dithering or constructive/destructive interference of higher frequencies that result in measurable but not necessarily audible differences in lower frequencies) as a result - or if the differences we "hear" are a result of another, non-signal related cause such as expectation bias, visual stimulation, etc.

There are forces at work in this universe that we don't see or perceive, but when they affect other things, we perceive that. The brain also doesn't need our 5 bodily senses to get information. The CD format is outlasting any other format, and we are keeping around old technology just for the stubbornness of a few.

We're keeping it around for the massive backwards compatibility, and the excellent quality of the format. If Redbook CDs weren't good enough, they would have gone the way of the cassette tape by now, or replaced by more than a tiny degree by higher quality formats. But they haven't. HDCD, SACD, and DVD-Audio have not been a resounding success, even among audio enthusiasts - although that doesn't have so much to do with the measurable superiority of high-res formats nor the general inaudibility of that - there's a whole lot of other things at play.

That's not to say that high res music is useless to us consumers. It may be a waste of space from an audibility perspective, yes, but we still benefit from it (beyond multichannel sound, a definite advantage in capability - although multichannel 16/44.1 & 48 PCM is certainly available as well). Why? Because these high resolution formats have given record labels a reason to better master recordings! Given the choice between a horrible mastering of some recording on a CD and a good mastering on some high-res format, I'll buy the high-res format if I have the equipment I need to play it back! Why? Not because it's high-res. Because it's better mastered. That has nothing to do with "higher quality files" and everything to do with how the tracks are prepared. They could be output to Redbook just fine, but since they're not there's still an advantage to getting the high-res files despite that advantage not being inherent in the higher resolution itself.
 
Last edited:
You don't need fancy math to show you can cram more resolution into something of fixed size. You can have a 50" TV that's 1080p, but does that mean whatever captured that image is 50"? NO! You say it's wasting resources to expand, well I say it's wasting resources to keep around old technology. Yesterday's music didn't have the good equipment to benefit from any good audio format. Now, today...today's music isn't good enough (between the content itself, and the ass hats (Rick Rubin type idiots) who do the audio work) to even take advantage of YESTERDAY's technology.
 
You don't need fancy math to show you can cram more resolution into something of fixed size. You can have a 50" TV that's 1080p, but does that mean whatever captured that image is 50"? NO!

What are you trying to say? You're rambling on incoherently - ignoring math and empirical evidence will get you nowhere. Judicious application of math and science are the reason we are able to have the society we have today - you can't just throw them out when you don't like the results.

You say it's wasting resources to expand, well I say it's wasting resources to keep around old technology.

How is it wasting resources? I have nothing against moving to, say, an all-online distribution method of music (although I do like having something to hold and look at as I listen to music - album covers and inserts, especially for LPs, are cool - but interactive extras with digital albums are nice too, and with increasing portable smart device use there's a huge untapped potential) - but you're suggesting, say, a new audio distribution standard based on BluRay discs or something similar, despite having no audible advantage. Eventually BluRays are going to be replaced too, you know... By keeping Redbook CDs - since no one is going to remove DVD (and thus CD) backwards-compatibility from BluRay players - we are able to continue using the multitude of perfectly fine CD players in existence today that are NOT compatible with new media - the ones in our vehicles, in our boom boxes, in our tertiary home stereos, in our computers - even at Grandma's house, where she finally bought a DVD player - and so on. Obviously, this becomes less important over time as Internet distribution of music becomes more and more popular. With a simple stereo line-level input in our vehicles or any other playback system, we can plug in any device and use any source that we want - from vinyl (after a pre-amp) to an iPod to whatever future line-level source you want to use. Obviously, that's an ideal that's being approached today - most new cars, receivers, etc. come equipped as such.

Yesterday's music didn't have the good equipment to benefit from any good audio format. Now, today...today's music isn't good enough (between the content itself, and the ass hats (Rick Rubin type idiots) who do the audio work) to even take advantage of YESTERDAY's technology.

Is that so? Give a listen to Blue Train. You'd be surprised what was (and is) do-able with "primitive" 1950s-era recording equipment when you have recording engineers and mastering engineers that know what they're doing. I'm not suggesting that modern recording equipment or playback equipment and formats aren't better (measurably they certainly are, the level of audibility is a different issue) - only that skillful and judiciuous use of ANY half-decent medium is far more important than the characteristic qualities of the medium itself. Hence, as you've pointed out, the shit audio quality of much music produced today despite being done on far superior equipment.
 
Today's society disgusts me. Take me back to the 1950s and the Southeastern US and I'd be happy and fit in.
 
All about speed and quantity. There are some who get it right though. A handful of recordings each year are pretty damn good. I get a catalogue of a smaller production company that is "old school". Can't recall the name although I can "see" the whole damn catalogue in my minds eye lol. Just no words.
 
It would be nice if one could get music through other channels before it gets butchered by the bad producers.
 
It would be nice if one could get music through other channels before it gets butchered by the bad producers.

dynamic range of 16bit is 16*6=96
Dynamic range of 24bit is 24*6=144

A dac with a snr value of 120db will add 6db noise to the signal starting from 20bit (6*20) till 24bit.

A dac with a snr value of 108db will do it at 18bit (18*6)

The highest a dac ever couldve produced was 24/96. Thats the $5000 dacs you get.
 
That's expressed by the container, not individual samples. X-bit PCM utilizes X number of bits for the representation of each sample.
 
For unsigned PCM, yes, but for signed PCM 1 bit is reserved for positive for negative.
 
Anyway, the loss you have regarding the original analog signal... Increasing the bit depth of course reduces the quantization noise, of course. Anyway, if you had infinite bit depth, and ignoring the limitations of DACs (since we're talking hypothetical and impossible situations here), you would reproduce the signal below the Nyquist frequency (after accounting for the necessary filters, of course) perfectly. Zero error, mathematically (again, since we're talking a hypothetical situation - there is no reality). You don't need a higher sampling rate to reproduce the lower frequencies, even relatively high 10-20 kHz ones.
Just to be clear to get a perfect reproduction of the frequencies below the nyquist point you would need

1: a perfect anti-aliasing filter
2: a perfect sample and hold circuit
3: an infinite resoloution ADC
4: an infinite resoloution DAC
5: a perfect recontstruction filter

1 and 5 are the biggest problem. Perfect filters simply don't exist so you will either get distortion to signals below the nyquist point or you will get signals above the nyquist point converted into signals below the nyquist point.

Haviing the nyquist point only 10% higher than top of your desired frequency range means that the anti-aliasing and reconstruction filters need to be very agressive and that means a lot of potential distortion to the signal, especially if using analog filters (if you want to get the best possible performance out of a given sample rate it's better to oversample, filter digitally and then downconvert since digital filters can come much closer to perfection than analog ones).
 
For unsigned PCM, yes, but for signed PCM 1 bit is reserved for positive for negative.
It isn't "reserved". In a two's compliment integer, the 'signage bit' is simply the most significant bit in a two's complement representation. The actual resolution doesn't change: a signed integer is able to represent the same number of unique values as an unsigned integer. No difference.
 
Just to be clear to get a perfect reproduction of the frequencies below the nyquist point you would need

1: a perfect anti-aliasing filter
2: a perfect sample and hold circuit
3: an infinite resoloution ADC
4: an infinite resoloution DAC
5: a perfect recontstruction filter

1 and 5 are the biggest problem. Perfect filters simply don't exist so you will either get distortion to signals below the nyquist point or you will get signals above the nyquist point converted into signals below the nyquist point.

Haviing the nyquist point only 10% higher than top of your desired frequency range means that the anti-aliasing and reconstruction filters need to be very agressive and that means a lot of potential distortion to the signal, especially if using analog filters (if you want to get the best possible performance out of a given sample rate it's better to oversample, filter digitally and then downconvert since digital filters can come much closer to perfection than analog ones).

Thanks for the additional info - as an ME rather than an EE or CE/CSE or similar, I don't spend too much time learning all of the details necessary for an unabridged understanding of DAC systems, but obviously a rudimentary understanding is helpful.

I find it interesting that you say 1 and 5 are the biggest problem when all of them are necessary for perfect reproduction (not to mention a method of storing infinite quantities of digital data, and other hoops to go through as well), but I understand that your point is that those are the largest real-world obstacles to fidelity.

Still, it remains that the distortions induced by any competent DAC system are tiny (even up to 20 kHz) in comparison to both analog music storage alternatives and the downstream components - amplifiers and transducers. The differences are certainly measurable - and in at least some cases [although extreme, like the Sony CDP-707 ESD vs. Phillips CD100 (14-bit) - the Phillips was the first commercially available CD player], audible. Modern DACs have of course improved leaps and bounds over that, and objective testing (both measurements and blind testing) shows how well they do so.
 
There are still leaps and bounds to go before we have realistic sound reproduction true to the source. It can't be accomplished staying where we are in formats.
 
Nyquist-Shannon sampling theorem tells us the sampling rate figure only needs to be twice that of the highest frequency of the waveform being captured. Why would any faster be better then? Surely slower is better then? When you push an ADC faster and faster to sample at something insane like 192,000 times a second @ 32 bits it is going to be less accurate than doing so @ 24 bit / 96 KHz (still overkill).

Why waste data on things you cannot hear? Is it for the benefit of your pet cats ultrasonic hearing capabilities?

I am in my 20s and I can't hear 18 KHz, maybe red book audio is too good for my ears then? Perhaps I'd be just fine with a sampling rate of 'only' 36 KHz? Higher sampling rate from my understanding then has no perceivable benefit and may be even detrimental when pushed too far. As for bit depth, well I don't know, but higher dynamic range isn't a bad thing. I do remember though being made jump out of my seat by the enormous sound of the canon in Tchaikovskys '1812' from a DVD audio disc with a particularly insane mastering job, I guess this effect could also be achieved with 16 bits right? Going from relaxing listening level to eardrums being nearly blown out (had my amp turned to 4 o'clock to even hear the rest of the track properly, its usually at 9)

All this said maybe some people in this thread should listen to some properly dithered 8 bit audio samples padded out with 0s to 16 bit, it still sounds good. Loosely associated to this post another practice I feel should be sharply criticized is the wastefullness of ripping vinyls at 24/96, it is such a waste of HDD space, memory is a bit rusty on this but I think you only need 11 bits for equivilent digital resolution. The specifications of vinyl are quite laughable when quantified and compared to redbook CD.

PS: did anyone else ever read about the time Mr. Ivor Tiefenbrun of audiophile company LINN audio (who once stated, "we will never produce digital equipment!") couldn't detect the presence of a primitive analogue -> digital converter in the signal chain of a vinyl setup. The irony.
 
Last edited:
There are still leaps and bounds to go before we have realistic sound reproduction true to the source. It can't be accomplished staying where we are in formats.
Don't blame formats first and foremost, blame the fact that the best and most expensive loudspeaker systems on the planet have more distortion than cheap consumer level electronics.
 
Back
Top