Hehe... I was about 9 when I played the original SimCity. A friend of mine and I "shared" a copy for a while. Whenever he needed a code, he'd call me up and I'd look it up for him.
Ah, those were the good ol' days before scanners, color copiers, and the Internet:-)
I love how they think you can make a free codec sound less appealing by putting quote marks around it...
Latest and gentlemen, Vorbis and Theora aren't REALLY free codecs, they're merely "free" codecs.
And don't get me started on how they conflate Ogg (a free container format which can hold data from various codecs) with Vorbis and Theora (free audio and video codecs, respectively).
Thanks, I will definitely look into it! As I said, I'm not so confident that I'm quite there at expert level yet, but it does sound like I could make some good contributions.
One thing I'm hesitant about: have they figured out what will be the license for Citizendium's original content?
I agree there is a difference in quality between the two Biology articles (particularly with the overemphasis on philosophy) but the quality so far may have something to do with how long the article has existed, the amount of edits and number people working on it so far compared to Wikipedia's entry. To use the Biology article as an example, experts and current contributors at Citizendium may know their Biology but not be as good at writing articles but the improvements will come over time as Citizendium continues to exist and gains more authors who are good at both Biology and writing articles (as happened with Wikipedia).
While I personally would not have "approved" the Biology article with its current introductory paragraphs, I think it is a little unfair at the moment comparing current Citizendium to current Wikipedia -- it would make more sense to compare current Citizendium after one year to where Wikipedia was at around the same time. Perhaps the Biology article on Wikipedia back then was not so great either. Fair point, Wikipedia does have a significant head start (although some of Citizendium's content is explicitly a "fork" of Wikipedia content used as a base).
However, I am instinctively a believer in the "more eyeballs" theory of open source development... more people poring over, tweaking, and writing code will improve its quality. If Citizendium is seen as limiting or discouraging by potential contributors, the articles won't be worked on as much, and I expect it will suffer in the long run. Then again, I do wish them great success, and if they attain it, I have no doubt that other projects will learn from their experience of emphasizing expert input!
Also, to correct one misapprehension in your post: anyone can contribute to Citizendium--they just have to sign up for an account; it's not just the experts who write the articles--in fact, the bulk of articles are created and edited by ordinary, non-"expert" authors and the experts (called editors in CZ-speak) mostly just keep a watch over the factual aspects--making sure there are no egregious or subtle errors creeping in--and making the odd contribution here and there (being experts in their field they are most likely also very busy outside of their Citizendium responsibilities). Yeah, that's a good point. Although I believe that there's a widespread perception that non-experts aren't so welcome at Citizendium. This is probably wrong, but maybe they need to do a better job of advertising this fact. Personally I am a PhD student and might be close to expert qualifications in one small field, but many of the topics I edit on wikipedia are peripheral to or outside of this field... and yet I think they're my most valuable contributions.
I guess I basically don't feel that Wikipedia is lacking or shoddy in any of the fields where I rely on it for technical information (condensed matter physics, materials science, signal processing, computer programming). Some articles lack sorely in terms of having a coherent narration of their topic, but in terms of fact-checking and having ability to follow up with more in-depth references, I'm very pleased. So perhaps I just don't clearly see what the separate expert role brings to the table...
I ought to play around with Citizendium more and find out for myself.
Citizendium seems to me to be a solution in search of a problem...
We already have a widely-used free encyclopedia that a lot of people see as reliable, broad, and timely... or at least a good enough combination to be useful: Wikipedia.
When I heard about Citizendium, I expected that the articles it produced would be very thorough, thought-provoking, and clearly written. But I'm pretty disappointed. Take a look at the article "Biology", for example. It meanders through a lot of history and philosophy of biology, without getting into much of the actual material, and without even clearly identifying subfields or specific topics. Way too general. Contrast it with the wikipedia article, which is of a similar length but much more USEFUL in my opinion. The wikipedia article gets straight to the point in outlining the framework of modern biology, identifying subfields and topics of interest, containing useful images, and ABOVE ALL the wikipedia article has extensive links to almost any term that might be unclear or simply merit further interest.
Obviously the experts who are in charge of the citizendium articles really know their stuff, and are good at writing... but are they good at really providing a dense framework of information? To me that's the most useful thing about wikipedia... anyone can write an article on anything they know a bit about, and have it appear quickly. So a user can almost effortlessly investigate the details of a subject, and its interconnections with other topics.
The reason I keep a record player and some records around is because it's fun to use. It's what I used as a kid (growing up in the 80's, I was probably the last generation to have vinyl as a primary format). I like taking out the big records and putting the needle down. It's fun to count the tracks to try and get the right song you want (and then slightly miss).
It's a fun toy. I'll always have a record player. Absolutely! Record players are totally fun. I like the album cover art a lot and speeding up and slowing down the records, and all that other analog silliness.
However, I'm quite sure that records aren't going to return as the format of choice for people who just want high-quality sound. Even if some crazy audiophile's $100,000 analog stereo sounds better than my $20 CD player and $50 speakers;-) The low cost, small size, insensitivity to physical environment, and SNR of digital audio is just too good to beat.
This is nonscientific listening, which is still listening - a lot more than a lot of the people who insist there's no difference have done.
Mostly it's from comparing the redbook cd side of a dualdisc to its DVD-audio counterpart, or comparing the SACD layer to the redbook layer on the same disc, or, most tellingly, downsampling my own recordings. Yeah, that does sound pretty good!
Obviously double blind tests are best, but those take a lot of time and energy and after this many years of comparative listening (including double blind tests) I trust my ears. Have you tried doing an ABX test on yourself in your studio? The wikipedia article on ABX tests has some pointers to music software that has plugins to do an ABX test... I think foobar2000 is one.
The problem with studies like the one you mention above is that they tend to use random people off the street. I'd be more interested in studies with people who have trained ears - I'm fully aware after seeing millions of people buy Bose and use the iPod earphones that your average Joe doesn't care about sound quality; the question isn't "can everyone hear the difference" it's "can anyone." That's true. It may well be that 99% of the population can't hear anything over 20kHz, but a few people fraction can. Or 99% can't hear the quantization noise of 24-bit audio any different from 16-bit... but a few can. And I can understand how, if that's the case, that small group would be driven nuts by the noisiness of conventional CD audio.
However, I have a feeling that most people don't care about sound quality not because they CAN'T hear the difference, but because it's not practical for them to do so. They do most of their listening in noisy environments (cars, gyms, offices) and don't want to carry very bulky headphones with them. I would guess the average person can easily distinguish iPod earbuds from a good pair, or a tube amp from a solid state amp. (Not sure about BOSE radios... personally, I think they sound good, but I'm not much of an ear myself!)
Nice, I hadn't seen this latest back-and-forth! I love watching Randi stick it to charlatans of all kinds... subjectivist-audiophiles are among the most annoying to me because they use science-sounding words a lot, and they generally surround themselves with the imprimatur of technology and attention to detail and "accuracy."
I didn't know that human hearing was that good, actually! Let's see... from this chart a 96 dB dynamic range is like hearing someone breathe over a nearby garbage truck, or a normal conversational voice over a jet taking off 200 feet away.
Yeah, sounds about right... just about the limit I'd say.
Snobbery aside, a 96k or 192k stereo recording sounds so much better than a redbook cd that it's patently obvious to almost any listener who cares to sit down and try it.
To my ear 96k and up recordings are indistinguishable from analog, this as someone with no vested interest in the different formats, who has spent a lot of time in the booth and listening to a wide array of high end stereos. They may well sound... but are you sure this isn't due to better mastering of the super-hi-fi format discs? In particular, super-hi-fi discs aren't hot-mastered (compression dynamic range) like many mass-market CDs are.
I don't know the situations in which you listen to Redbook CD vs. some 96 kHz source. Could it be that there were better speakers connected to the super-hi-fi system than the CD system? Different music when you listened to the super-hi-fi system? Better room acoustics? Different overall loudness levels (*VERY SMALL* loudness differences have a significant effect on perceived sound quality)? Etc...
This is why it's really important to do double-blinding testing. Ideally, you should listen to a 96 kHz sample, and a 48 kHz downsampled version of the same thing on the same equipment, without knowing which is which... in an ABX test. And repeat that several hundred times with different music:-) And then do a statistical test for significance to decide how likely it is that your results are not explained by chance.
A recent study basically did that, it seems. The wikipedia summary, at http://en.wikipedia.org/wiki/Super_Audio_CD#Comparing_SACD_and_CD, says:
An article published in the Journal of the Audio Engineering Society Vol. 55 Number 9, September 2007, entitled "Audibility of a CD-Standard ADA Loop Inserted Into High-Resolution Audio Playback" by E. Brad Meyer and David Moran reported the results of their study, which concluded that listeners could not hear the difference between a high-resolution two-channel recording and a CD-quality downsampling of the same recording except at extremely high sound levels. The article concluded that many high-resolution releases sounded better than their CD counterparts, but attributes this to mastering differences. Unfortunately I can't find a PDF of that article anywhere, would love to read it...
No, the modulations in between samples do not have to include a full period. I'm aware of that. The modulations in-between samples have to include HALF a period in order to differ from a straight-line approximation by an amount that exceeds the quantization noise. I took that into account in my calculation. If you want a more rigorous proof of the Nyquist theorem, look at the wikipedia article which is quite good.
Have you ever seen what a waveform looks like when you combine multiple frequencies. It's called a complex waveform and it is this information that is being slightly dumbed down in a digital recording. "Dumbed down"? The waveform is discretized in both time and amplitude. The effects of both of those processes on the frequency content, dynamic range, and noise floor are easily quantifiable and treated in the introductory chapters of any book on Digital Signal Processing.
I got "Digital signal processing: a practical approach" (by Ifeachor and Jervis) from my college library. Helped me understand a lot of stuff about both analog and digital signal processing. I recommend it.
I'm not arguing that Vinyl should or will make a real comeback. I'm just saying that people who continue to defend the RedBook standard are doing so out of naievity. Why do you think that the DVD-Audio standard allows Full Surround 5.1 to be sampled at 96 kHz and Stereo sound to be sampled at up to 192 kHz? Well, first of all, oversampling has cost advantages in some cases. With a higher sampling rate, less aggressive anti-aliasing filters are needed, which can make the system CHEAPER. Faster DSPs and high-capacity discs are so cheap these days, it's probably a good tradeoff to sample faster and save a bit on the anti-aliasing filters. (Less aggressive AA filters *might* also produce some negligible improvements in phase and amplitude response at close to 20 kHz.)
Another advantage of oversampling is that you can do noise shaping (wikipedia it). This makes quantization noise even smaller. As I said above, a CD system has about 100 dB of SNR at full-scale... no way you can hear the noise. BUT if your audio has a very high dynamic range, some of the audio might be far below full-scale (imagine a movie sound-track that includes everything from cannons shooting to footsteps in the grass at night). These quiet segments may benefit from oversampling and noise shaping to reduce the audible quantization noise.
But mostly, let's not kid ourselves... the main reason for the higher sampling frequencies is to market them as something "NEW AND IMPROVED!" to people who are easily impressed by numbers that they don't understand.
That's a very interesting explanation for why people want to believe they're not getting duped. I've heard similar things about how people who lose lots of money to Nigerian 419 scammers are often insistent that they haven't been cheated and it's going to pay off for them soon.
The priceless part:
Mr. Dunlavy has often gathered audio critics in his Colorado Springs lab for a demonstration.
''What we do is kind of dirty and stinky,'' he said. ''We say we are starting with a 12 WAG zip cord, and we position a technician behind each speaker to change the cables out.''
The technicians hold up fancy-looking cables before they disappear behind the speakers. The critics debate the sound characteristics of each wire.
''They describe huge changes and they say, 'Oh my God, John, tell me you can hear that difference,' '' Mr. Dunlavy said. The trick is the technicians never actually change the cables, he said, adding, ''It's the placebo effect.'' Hilarious...
You are missing the point. Yes Vinyl can capture higher pitch but it also is an analog recording which means it is a capture of the ACTUAL WAVEFORM. A cd is digital and only has sampings of the waveform. Basically think of it as points on a graph. Each sample is a point and then the player draws a line between the points the best it can. The problem comes in when there is a small difference between how the line is drawn and the actual waveform. I'm not talking about a complete cycle of the wave which would be a sound in that area that humans can't even hear (but can sometimes percieve). I'm talking about small modulations along the waveform that add to the richness of the sound. This is what people are talking about when they say vinyl sounds more warm. Now digital samplings can overcome this problem by increasing the sampling frequency to the point that humans can't percieve a difference. 44kHz doesn't cut it though. And what does it mean that you can't hear a sound, but can sometimes perceive it? I believe that audio researchers have done brain-scan studies where they play sounds above the hearing limit to see if they trigger any brain activity... and didn't find much. Google it.
As for the notion that sampling removes meaningful, audible data, by ignoring modulations between the samples... let's do the math. If you're sampling at 44 kHz, then these small uncaptured modulations must occur in a time frame of less than 23 micro-seconds (1/44 kHz). Since these modulations differ significantly from a straight line drawn between the two sample points, their period must be less than twice this, or less than 46 micro-seconds. This corresponds to a frequency of 22 kHz or higher. We're back to the Nyquist theorem again... all of the "small modulations" you describe are at such a high frequency that they cannot be heard.
This is perhaps slightly counter-intuitive, but widely understood in psychoacoustics. Your ear basically does a (crude) Fourier transform of the sound that you hear: rather than hearing individual pressure pulses, sound is converted into a frequency spectrum which is what actually gets processed by your brain. Your ear can't hear a 25 kHz tone in isolation, and can't hear it when superposed on another sound either.
Or maybe what you're worried about is quantization noise, which is due to the digital signal being discrete not only in time, but in amplitude as well. As I explained above, the quantization noise of a 16-bit system is as quiet as a whisper compared to a shotgun blast in the best case. Maybe something like a whisper compared to a vacuum cleaner in music that uses a large dynamic range (like a classical symphony).
People love to obsess over the noise in digital audio because it is PREDICTABLE and QUANTIFIABLE. It is also very small... the quantization error and aliasing noise of a CD player is too small to hear, according to most scientific research. The noise in analog audio is much harder to predict and describe. This seems to me to be a classic case of color of the bikeshed argument... digital noise is easy to measure and understand, so everyone loves to talk about it. Analog noise is much messier and harder to figure out, so people ignore it, despite the fact that it's much larger in many practical systems.
It sounds like someone at Wired has drank the audiophile kool-aid... There are very strong parallels between audiophily and religion. Agreed. As Douglas Self argues, many audiophiles have explicitly rejected rational explanations for the phenomena they describe:
I have been told by a Subjectivist that the operation of the human ear is so complex that its interaction with measurable parameters lies forever beyond human comprehension.
Vinly is simply better because they know it in their hearts, and no science can disprove that fact. And also the fact that they've spent $2000 on a 3-foot-long wire or ten times that for a bunch of vacuum tubes:-) No one wants to admit they've been completely had...
Exactly right. A 22 kHz sampling rate means no frequency above 11 kHz can be reproduced (realistically more like 10 kHz with a good anti-aliasing filter). 44 kHz sampling rate cuts off at 22 kHz (realistically more like 20 kHz).
This is the Nyquist theorem, which says that the highest sound frequency that can be stored/reproduced in a digital signal is HALF the sampling rate.
If you have young and healthy ears, you should be able to hear many sounds above 10 kHz.
Now, what about amplitude? If you only have two samples going up, how do you know you've not sampled asymetrically and that the amplitude you think you have is wrong? What???
Are you describing quantization noise? If so, then I refuted that above by explaining that a 16-bit ADC has sufficient signal-to-noise ratio to make that pretty much negligible.
Or maybe you're describing aliasing? If so, then others have already mentioned the need for anti-aliasing filters. They get the job done.
So what happens to overtones that make the difference between second octave C on a piccolo sound different from that on a flute (or, even, violin)? Lost it, haven't you. The ammount of "mix in" of these higher tones is lost and that is much of the difference between the instruments. No, straight up wrong. (a) If those higher tones are present in the analog signal, they'll be present in the digital signal as well, provided they fall below the Nyquist/cutoff frequency. (b) If those higher tones are ABOVE the Nyquist frequency, they won't be present in the digital signal. But that's why the Nyquist frequency is chosen to be ABOVE THE LIMIT OF HUMAN HEARING for a high-fidelity digital system such as a CD player.
Nyquist is also the minimally reproduceable limit. It doesn't mean that the errors don't fuzz the limits. Indeed. That's why the sampling rate is chosen to be HIGHER than the theoretical minimum. For example: CDs are supposed to reproduce sound up to 20 kHz. In theory you would need only a 40 kHz sampling frequency to do this. But in practice, the anti-aliasing low-pass filter won't be perfectly sharp... so a 44 kHz sampling frequency is used instead. No problemo!
And you stil (even though the flaming summary said it) that CD's can take a lot more audio compression, reducing the fidelity *on purpose* and has therefore nothing to do with what the CD *could* play but with what they put on there. I mean, you can record a Speccy 8KHz bleep on a CD and that won't get you 16-bit 44KHz sound out of it. Right. All I'm saying is that the CD format is perfectly CAPABLE of good dynamic range. If CD recordings are unnecessarily compressed, this indicates a problem with the people who do the mixing, not the format itself.
You just have a bee in your bonnet about being told stuff you don't believe. No. What gets me is audiophiles who claim that certain equipment possesses mystical qualities that allow them to hear better sound, in defiance of what science and years of audio engineering experience has shown. For example, the jokers at Stereophile who blather on about how their $2000 POWER CABLES enrich the "body" and "sonic depth" of the music. They go on and on about such-and-such expensive equipment, but they never do a frickin' ABX double-blind test.
What you are describing is called quantization noise (or quantization error).
You are indeed correct that the noise signal will be correlated with the input signal under certain conditions, and that this will cause predictable, harmonic distortion.
BUT what you're leaving out is the scale of this distortion. With a proper 16-bit ADC (as used for CD mastering), the quantization noise amplitude is less than 1/2-bit. Meaning that the signal-to-noise ratio is ~100 dB for a full-scale signal without any additional processing. This is like the difference between a shotgun blast a few feet away, and a whisper (http://www.4servnow.com/webapp/GetPage?pid=57). The quantization noise is thus too quiet to hear.
Furthermore, there are designs that can suppress quantization noise even further in digital signals. Someone else mentioned oversampling: that is what is used in the Super-Audio CD (which audiophiles seem to love even though it is digital). An SACD contains a digital sample taken with only ONE BIT of resolution, which sounds like it would be very crude, but at a very very high sampling frequency (about 2 MHz I think). And a marvelously simple technique called noise shaping is used to "push" the quantization error out of the audible frequency range. You can read about it in the wikipedia article on Pulse-density modulation, which is the term for that type of oversampled 1-bit digital signal. (Disclosure: I wrote a good chunk of that article.)
It sounds like someone at Wired has drank the audiophile kool-aid...
Another reason for vinyl's sonic superiority is that no matter how high a sampling rate is, it can never contain all of the data present in an analog groove, Nyquist's theorem to the contrary. Are you kidding me? A CD with a sampling frequency of 44 kHz carries sound up to 20 kHz, which is beyond the hearing limit of most humans. An analog groove may in theory carry sound up to very high frequencies, but is badly limited in practice by the difficulty of cutting a precise high-frequency groove, the non-linear response of the cartridge at high frequency, and a host of other factors. Not to mention the fact that NO ONE CAN HEAR THOSE SOUNDS above 20 kHz! And to get top-notch frequency response out of a record player, you have to obsess over the cleanliness and storage of your records and player... and even then you're likely to degrade the frequency response RAPIDLY to well below the level of a CD (http://en.wikipedia.org/wiki/Vinyl_record#Frequency_response_and_noise)
Wired seems to take all the standard audiophile BS hook, line, and sinker... "analog provides a warmer sound" (much more total harmonic distortion than a digital player), etc.
The argument about hot mastered CDs is particularly hilarious (reduced dynamic range). Basically, this is a result of crappy commercial pressure to sound louder, and is common but by no means universal. The fact that vinyl lacks this possibility is touted as an advantage. It's like claiming that a knife is better than a gun, because you can't shoot yourself in the foot with the knife.
This is a really interesting topic called the GZK cutoff. Basically, you expect that a particle with sufficiently high energy could scatter off of the microwave photons in the cosmic microwave background that permeate the entire universe, left over from the Big Bang billions of years ago. The particles have so much energy that when they hit the little photon, they pop off an entire pion (mass about a 15% of the proton). Since the universe is dense with such photons, you'd think these high-energy particles would just continuously pop off pions until their energies are below the GZK cutoff.
Hmmm... interesting. So is there any current theory that could plausibly explain the detection of these high-energy cosmic rays? Are they thought to be just measurement anomalies?
Is the GZK cutoff truly isotropic like the CMBR, or are there magnetic or other interactions that can disrupt it?
Indeed. The wikipedia article on ultra-high-energy cosmic rays has more info. The energy of such a particle is simply insane...
Some of them apparently violate a theoretical limit on the energy of a particle that has traveled a long way across the universe... leading to the question of where exactly they come from.
You're probably right. I suppose the vindictive part of me just wants to see SCO pounded into a pulp without the slightest acknowledgement of any value:-) But that's not good policy. So yeah, if there's any good code in there, let's have it.
Slashdot Mirror
Hehe... I was about 9 when I played the original SimCity. A friend of mine and I "shared" a copy for a while. Whenever he needed a code, he'd call me up and I'd look it up for him.
Ah, those were the good ol' days before scanners, color copiers, and the Internet:-)
Now I won't need that stupid red code sheet to avoid frequent Godzilla attacks. Hooray!!
I love how they think you can make a free codec sound less appealing by putting quote marks around it...
Latest and gentlemen, Vorbis and Theora aren't REALLY free codecs, they're merely "free" codecs.
And don't get me started on how they conflate Ogg (a free container format which can hold data from various codecs) with Vorbis and Theora (free audio and video codecs, respectively).
Who has $4500 to spend on a laptop, and is so obsessed with a computer game that they want it stamped all over their computer???
Whoever you are... when I grow up, I do NOT want to be like you!!!
In 21st century America, grandma emails YOU!!!!
Thanks, I will definitely look into it! As I said, I'm not so confident that I'm quite there at expert level yet, but it does sound like I could make some good contributions.
One thing I'm hesitant about: have they figured out what will be the license for Citizendium's original content?
While I personally would not have "approved" the Biology article with its current introductory paragraphs, I think it is a little unfair at the moment comparing current Citizendium to current Wikipedia -- it would make more sense to compare current Citizendium after one year to where Wikipedia was at around the same time. Perhaps the Biology article on Wikipedia back then was not so great either. Fair point, Wikipedia does have a significant head start (although some of Citizendium's content is explicitly a "fork" of Wikipedia content used as a base).
However, I am instinctively a believer in the "more eyeballs" theory of open source development... more people poring over, tweaking, and writing code will improve its quality. If Citizendium is seen as limiting or discouraging by potential contributors, the articles won't be worked on as much, and I expect it will suffer in the long run. Then again, I do wish them great success, and if they attain it, I have no doubt that other projects will learn from their experience of emphasizing expert input! Also, to correct one misapprehension in your post: anyone can contribute to Citizendium--they just have to sign up for an account; it's not just the experts who write the articles--in fact, the bulk of articles are created and edited by ordinary, non-"expert" authors and the experts (called editors in CZ-speak) mostly just keep a watch over the factual aspects--making sure there are no egregious or subtle errors creeping in--and making the odd contribution here and there (being experts in their field they are most likely also very busy outside of their Citizendium responsibilities). Yeah, that's a good point. Although I believe that there's a widespread perception that non-experts aren't so welcome at Citizendium. This is probably wrong, but maybe they need to do a better job of advertising this fact. Personally I am a PhD student and might be close to expert qualifications in one small field, but many of the topics I edit on wikipedia are peripheral to or outside of this field... and yet I think they're my most valuable contributions.
I guess I basically don't feel that Wikipedia is lacking or shoddy in any of the fields where I rely on it for technical information (condensed matter physics, materials science, signal processing, computer programming). Some articles lack sorely in terms of having a coherent narration of their topic, but in terms of fact-checking and having ability to follow up with more in-depth references, I'm very pleased. So perhaps I just don't clearly see what the separate expert role brings to the table...
I ought to play around with Citizendium more and find out for myself.
Citizendium seems to me to be a solution in search of a problem...
We already have a widely-used free encyclopedia that a lot of people see as reliable, broad, and timely... or at least a good enough combination to be useful: Wikipedia.
When I heard about Citizendium, I expected that the articles it produced would be very thorough, thought-provoking, and clearly written. But I'm pretty disappointed. Take a look at the article "Biology", for example. It meanders through a lot of history and philosophy of biology, without getting into much of the actual material, and without even clearly identifying subfields or specific topics. Way too general. Contrast it with the wikipedia article, which is of a similar length but much more USEFUL in my opinion. The wikipedia article gets straight to the point in outlining the framework of modern biology, identifying subfields and topics of interest, containing useful images, and ABOVE ALL the wikipedia article has extensive links to almost any term that might be unclear or simply merit further interest.
Obviously the experts who are in charge of the citizendium articles really know their stuff, and are good at writing... but are they good at really providing a dense framework of information? To me that's the most useful thing about wikipedia... anyone can write an article on anything they know a bit about, and have it appear quickly. So a user can almost effortlessly investigate the details of a subject, and its interconnections with other topics.
It's a fun toy. I'll always have a record player. Absolutely! Record players are totally fun. I like the album cover art a lot and speeding up and slowing down the records, and all that other analog silliness.
However, I'm quite sure that records aren't going to return as the format of choice for people who just want high-quality sound. Even if some crazy audiophile's $100,000 analog stereo sounds better than my $20 CD player and $50 speakers
Mostly it's from comparing the redbook cd side of a dualdisc to its DVD-audio counterpart, or comparing the SACD layer to the redbook layer on the same disc, or, most tellingly, downsampling my own recordings. Yeah, that does sound pretty good! Obviously double blind tests are best, but those take a lot of time and energy and after this many years of comparative listening (including double blind tests) I trust my ears. Have you tried doing an ABX test on yourself in your studio? The wikipedia article on ABX tests has some pointers to music software that has plugins to do an ABX test... I think foobar2000 is one. The problem with studies like the one you mention above is that they tend to use random people off the street. I'd be more interested in studies with people who have trained ears - I'm fully aware after seeing millions of people buy Bose and use the iPod earphones that your average Joe doesn't care about sound quality; the question isn't "can everyone hear the difference" it's "can anyone." That's true. It may well be that 99% of the population can't hear anything over 20kHz, but a few people fraction can. Or 99% can't hear the quantization noise of 24-bit audio any different from 16-bit... but a few can. And I can understand how, if that's the case, that small group would be driven nuts by the noisiness of conventional CD audio.
However, I have a feeling that most people don't care about sound quality not because they CAN'T hear the difference, but because it's not practical for them to do so. They do most of their listening in noisy environments (cars, gyms, offices) and don't want to carry very bulky headphones with them. I would guess the average person can easily distinguish iPod earbuds from a good pair, or a tube amp from a solid state amp. (Not sure about BOSE radios... personally, I think they sound good, but I'm not much of an ear myself!)
Nice, I hadn't seen this latest back-and-forth! I love watching Randi stick it to charlatans of all kinds... subjectivist-audiophiles are among the most annoying to me because they use science-sounding words a lot, and they generally surround themselves with the imprimatur of technology and attention to detail and "accuracy."
I didn't know that human hearing was that good, actually! Let's see... from this chart a 96 dB dynamic range is like hearing someone breathe over a nearby garbage truck, or a normal conversational voice over a jet taking off 200 feet away.
Yeah, sounds about right... just about the limit I'd say.
To my ear 96k and up recordings are indistinguishable from analog, this as someone with no vested interest in the different formats, who has spent a lot of time in the booth and listening to a wide array of high end stereos. They may well sound
I don't know the situations in which you listen to Redbook CD vs. some 96 kHz source. Could it be that there were better speakers connected to the super-hi-fi system than the CD system? Different music when you listened to the super-hi-fi system? Better room acoustics? Different overall loudness levels (*VERY SMALL* loudness differences have a significant effect on perceived sound quality)? Etc...
This is why it's really important to do double-blinding testing. Ideally, you should listen to a 96 kHz sample, and a 48 kHz downsampled version of the same thing on the same equipment, without knowing which is which... in an ABX test. And repeat that several hundred times with different music
A recent study basically did that, it seems. The wikipedia summary, at http://en.wikipedia.org/wiki/Super_Audio_CD#Comparing_SACD_and_CD, says: An article published in the Journal of the Audio Engineering Society Vol. 55 Number 9, September 2007, entitled "Audibility of a CD-Standard ADA Loop Inserted Into High-Resolution Audio Playback" by E. Brad Meyer and David Moran reported the results of their study, which concluded that listeners could not hear the difference between a high-resolution two-channel recording and a CD-quality downsampling of the same recording except at extremely high sound levels. The article concluded that many high-resolution releases sounded better than their CD counterparts, but attributes this to mastering differences. Unfortunately I can't find a PDF of that article anywhere, would love to read it...
I got "Digital signal processing: a practical approach" (by Ifeachor and Jervis) from my college library. Helped me understand a lot of stuff about both analog and digital signal processing. I recommend it. I'm not arguing that Vinyl should or will make a real comeback. I'm just saying that people who continue to defend the RedBook standard are doing so out of naievity. Why do you think that the DVD-Audio standard allows Full Surround 5.1 to be sampled at 96 kHz and Stereo sound to be sampled at up to 192 kHz? Well, first of all, oversampling has cost advantages in some cases. With a higher sampling rate, less aggressive anti-aliasing filters are needed, which can make the system CHEAPER. Faster DSPs and high-capacity discs are so cheap these days, it's probably a good tradeoff to sample faster and save a bit on the anti-aliasing filters. (Less aggressive AA filters *might* also produce some negligible improvements in phase and amplitude response at close to 20 kHz.)
Another advantage of oversampling is that you can do noise shaping (wikipedia it). This makes quantization noise even smaller. As I said above, a CD system has about 100 dB of SNR at full-scale... no way you can hear the noise. BUT if your audio has a very high dynamic range, some of the audio might be far below full-scale (imagine a movie sound-track that includes everything from cannons shooting to footsteps in the grass at night). These quiet segments may benefit from oversampling and noise shaping to reduce the audible quantization noise.
But mostly, let's not kid ourselves... the main reason for the higher sampling frequencies is to market them as something "NEW AND IMPROVED!" to people who are easily impressed by numbers that they don't understand.
That's a very interesting explanation for why people want to believe they're not getting duped. I've heard similar things about how people who lose lots of money to Nigerian 419 scammers are often insistent that they haven't been cheated and it's going to pay off for them soon.
Here's a great NYT article on high-end cable BS: http://query.nytimes.com/gst/fullpage.html?res=9D06E1D61739F930A15751C1A96F958260&sec=&spon=&pagewanted=all
The priceless part: Mr. Dunlavy has often gathered audio critics in his Colorado Springs lab for a demonstration.
''What we do is kind of dirty and stinky,'' he said. ''We say we are starting with a 12 WAG zip cord, and we position a technician behind each speaker to change the cables out.''
The technicians hold up fancy-looking cables before they disappear behind the speakers. The critics debate the sound characteristics of each wire.
''They describe huge changes and they say, 'Oh my God, John, tell me you can hear that difference,' '' Mr. Dunlavy said. The trick is the technicians never actually change the cables, he said, adding, ''It's the placebo effect.'' Hilarious...
As for the notion that sampling removes meaningful, audible data, by ignoring modulations between the samples... let's do the math. If you're sampling at 44 kHz, then these small uncaptured modulations must occur in a time frame of less than 23 micro-seconds (1/44 kHz). Since these modulations differ significantly from a straight line drawn between the two sample points, their period must be less than twice this, or less than 46 micro-seconds. This corresponds to a frequency of 22 kHz or higher. We're back to the Nyquist theorem again... all of the "small modulations" you describe are at such a high frequency that they cannot be heard.
This is perhaps slightly counter-intuitive, but widely understood in psychoacoustics. Your ear basically does a (crude) Fourier transform of the sound that you hear: rather than hearing individual pressure pulses, sound is converted into a frequency spectrum which is what actually gets processed by your brain. Your ear can't hear a 25 kHz tone in isolation, and can't hear it when superposed on another sound either.
Or maybe what you're worried about is quantization noise, which is due to the digital signal being discrete not only in time, but in amplitude as well. As I explained above, the quantization noise of a 16-bit system is as quiet as a whisper compared to a shotgun blast in the best case. Maybe something like a whisper compared to a vacuum cleaner in music that uses a large dynamic range (like a classical symphony).
People love to obsess over the noise in digital audio because it is PREDICTABLE and QUANTIFIABLE. It is also very small... the quantization error and aliasing noise of a CD player is too small to hear, according to most scientific research. The noise in analog audio is much harder to predict and describe. This seems to me to be a classic case of color of the bikeshed argument... digital noise is easy to measure and understand, so everyone loves to talk about it. Analog noise is much messier and harder to figure out, so people ignore it, despite the fact that it's much larger in many practical systems.
I like it! "Cosmic rays... nature's FPU errors" :-)
Exactly right. A 22 kHz sampling rate means no frequency above 11 kHz can be reproduced (realistically more like 10 kHz with a good anti-aliasing filter). 44 kHz sampling rate cuts off at 22 kHz (realistically more like 20 kHz).
This is the Nyquist theorem, which says that the highest sound frequency that can be stored/reproduced in a digital signal is HALF the sampling rate.
If you have young and healthy ears, you should be able to hear many sounds above 10 kHz.
Are you describing quantization noise? If so, then I refuted that above by explaining that a 16-bit ADC has sufficient signal-to-noise ratio to make that pretty much negligible.
Or maybe you're describing aliasing? If so, then others have already mentioned the need for anti-aliasing filters. They get the job done. So what happens to overtones that make the difference between second octave C on a piccolo sound different from that on a flute (or, even, violin)? Lost it, haven't you. The ammount of "mix in" of these higher tones is lost and that is much of the difference between the instruments. No, straight up wrong. (a) If those higher tones are present in the analog signal, they'll be present in the digital signal as well, provided they fall below the Nyquist/cutoff frequency. (b) If those higher tones are ABOVE the Nyquist frequency, they won't be present in the digital signal. But that's why the Nyquist frequency is chosen to be ABOVE THE LIMIT OF HUMAN HEARING for a high-fidelity digital system such as a CD player. Nyquist is also the minimally reproduceable limit. It doesn't mean that the errors don't fuzz the limits. Indeed. That's why the sampling rate is chosen to be HIGHER than the theoretical minimum. For example: CDs are supposed to reproduce sound up to 20 kHz. In theory you would need only a 40 kHz sampling frequency to do this. But in practice, the anti-aliasing low-pass filter won't be perfectly sharp... so a 44 kHz sampling frequency is used instead. No problemo! And you stil (even though the flaming summary said it) that CD's can take a lot more audio compression, reducing the fidelity *on purpose* and has therefore nothing to do with what the CD *could* play but with what they put on there. I mean, you can record a Speccy 8KHz bleep on a CD and that won't get you 16-bit 44KHz sound out of it. Right. All I'm saying is that the CD format is perfectly CAPABLE of good dynamic range. If CD recordings are unnecessarily compressed, this indicates a problem with the people who do the mixing, not the format itself. You just have a bee in your bonnet about being told stuff you don't believe. No. What gets me is audiophiles who claim that certain equipment possesses mystical qualities that allow them to hear better sound, in defiance of what science and years of audio engineering experience has shown. For example, the jokers at Stereophile who blather on about how their $2000 POWER CABLES enrich the "body" and "sonic depth" of the music. They go on and on about such-and-such expensive equipment, but they never do a frickin' ABX double-blind test.
What you are describing is called quantization noise (or quantization error).
You are indeed correct that the noise signal will be correlated with the input signal under certain conditions, and that this will cause predictable, harmonic distortion.
BUT what you're leaving out is the scale of this distortion. With a proper 16-bit ADC (as used for CD mastering), the quantization noise amplitude is less than 1/2-bit. Meaning that the signal-to-noise ratio is ~100 dB for a full-scale signal without any additional processing. This is like the difference between a shotgun blast a few feet away, and a whisper (http://www.4servnow.com/webapp/GetPage?pid=57). The quantization noise is thus too quiet to hear.
Furthermore, there are designs that can suppress quantization noise even further in digital signals. Someone else mentioned oversampling: that is what is used in the Super-Audio CD (which audiophiles seem to love even though it is digital). An SACD contains a digital sample taken with only ONE BIT of resolution, which sounds like it would be very crude, but at a very very high sampling frequency (about 2 MHz I think). And a marvelously simple technique called noise shaping is used to "push" the quantization error out of the audible frequency range. You can read about it in the wikipedia article on Pulse-density modulation, which is the term for that type of oversampled 1-bit digital signal. (Disclosure: I wrote a good chunk of that article.)
Wired seems to take all the standard audiophile BS hook, line, and sinker... "analog provides a warmer sound" (much more total harmonic distortion than a digital player), etc.
The argument about hot mastered CDs is particularly hilarious (reduced dynamic range). Basically, this is a result of crappy commercial pressure to sound louder, and is common but by no means universal. The fact that vinyl lacks this possibility is touted as an advantage. It's like claiming that a knife is better than a gun, because you can't shoot yourself in the foot with the knife.
For a devastating rebuttal of audiophile BS from a very experienced engineer, read Douglas Self's site: http://www.dself.dsl.pipex.com/ampins/pseudo/subjectv.htm
This is a really interesting topic called the GZK cutoff. Basically, you expect that a particle with sufficiently high energy could scatter off of the microwave photons in the cosmic microwave background that permeate the entire universe, left over from the Big Bang billions of years ago. The particles have so much energy that when they hit the little photon, they pop off an entire pion (mass about a 15% of the proton). Since the universe is dense with such photons, you'd think these high-energy particles would just continuously pop off pions until their energies are below the GZK cutoff.
And recent high-statistics experiments verify the existence of the cutoff, refuting earlier suggestions that there were anomalously high-energy cosmic rays: Observation of the GZK Cutoff by the HiRes Experiment
Hmmm... interesting. So is there any current theory that could plausibly explain the detection of these high-energy cosmic rays? Are they thought to be just measurement anomalies?Is the GZK cutoff truly isotropic like the CMBR, or are there magnetic or other interactions that can disrupt it?
Indeed. The wikipedia article on ultra-high-energy cosmic rays has more info. The energy of such a particle is simply insane...
Some of them apparently violate a theoretical limit on the energy of a particle that has traveled a long way across the universe... leading to the question of where exactly they come from.
You're probably right. I suppose the vindictive part of me just wants to see SCO pounded into a pulp without the slightest acknowledgement of any value :-) But that's not good policy. So yeah, if there's any good code in there, let's have it.