http://www.audiostream.com/content/draft (http://www.audiostream.com/content/draft) (part 1)
http://www.audiostream.com/content/theres-...eve-silberman-p (http://www.audiostream.com/content/theres-no-such-thing-digital-conversation-charles-hansen-gordon-rankin-and-steve-silberman-p) (part 2)
After I read all of thew pages, I actually found the explanation rather plausible. But somehow my heart says something is terribly wrong with the description, but I have no idea what it is.
Nuke this if old.
I only read three sentences, but what seems to be wrong is that they have no idea/are in denial about the meaning of "digital".
Well, all transmissions are by their nature analog, but the encoding of that signal and data that is transmitted, along with error correction codes are what matters. You can transmit digital information by river stream with bits or chunks of wood (yeah, it would be slow); and it's not the river (or electrical current) that matters, what matters is the signal itself.
Yeah, I stopped reading after a short time too.
First of all, they seem to think that digital and analog are opposites. They're not. Digital simply means representing something as a number and analog simply means representing something with a similar phenomenon. So, digital in audio means representing something as a stream of numbers while analog usually means representing a sound pressure with an electrical signal or the depth of a groove (vinyl), or the strength of a magnetic field (tape).
Anyway, they are referring to a few things like slew rate (it takes time for the signal to change levels), absolute voltage (there is a "grey" zone between the "black" of the "0" and the "white" of the "1") and timing (If the bit changes to the correct state but at the wrong time, this is equivalent to changing to the wrong level at the correct time) but the thing is, those engineers that designed it aren't dumb. Most digital signals need some kind of clocking signal. When only one signal stream is available, like with CD and S/PDIF, there are tricks to embed the clocking signal and the data into one stream.
AS: Since there's no such thing as 1s and 0s in digital transmission, what is being sent over our USB/Firewire/Ethernet cables when we play back music files?
CH: An ANALOG signal!
In fact they probably mean an electric signal, because the transmission is in no way 'analog' to any of the data. They confuse the term analog signal for 'physical' signals, like representation as a voltage, mechanical pressure, light intensity etc. The only way you can strictly represent a digital signal with an analog is by counting something, like logs as hlloyge proposed, or maybe by integration (measuring the amount of electrons and dividing that by some number) as digital is purely representing by numbers.
They accuse 'engineers' of not knowing what they're doing, but they themselves are talking nonsense.
"By Michael Lavorgna" is all I needed to read—he's as misguided as Fremer.
How much energy is wasted delivering the data seems to have an effect on sound. As with increased energy usage the amount of EMI/RFI radiation also increases. This might be a reason why applications sound different. If we look at the "top" command in the Terminal application on OS X we see a programs usage and percent time and all the processes associated with that program. In practice the applications with the least required processing time also sounds the best. This may have an indication of why file types sound different. If you unpack a lossless file on the fly the processing time increases measurably and that tends to decrease the sound quality.
It's just random words and phrases. I think this article was generated by a top-notch Markov algorithm.
No wonder people get confused with these kind of statements that appear superficually true but are in fact bunk:
I think this is where things get misconstrued. The signals we think of abstractly as “digital” are in fact high-speed analog square waves, susceptible to all of the same damage and distortions as any other analog signal.
They're high-speed but that's irrelevant. They're not the analog of sound (they're the analog of their own digital abstraction!). They're not susceptible to all the same damage and distortions an ("other") analog signal.
I used to frequent that site when it seemed to be mostly computer audio gear reviews without so much boutique audiophile sauce. Or maybe it was always like that... Anyway that's a common type of article there. They are always tweaking or "upgrading" and telling you it's for the quest for the best sound, like, EVER. The "experts" are product designers of hi-end gear. Weird right?!!
Well, all transmissions are by their nature analog, but the encoding of that signal and data that is transmitted, along with error correction codes are what matters. You can transmit digital information by river stream with bits or chunks of wood (yeah, it would be slow); and it's not the river (or electrical current) that matters, what matters is the signal itself.
Agreed. They've missed the point that analog signals are subject to the same kinds of media errors (that's what they are obsessing over - errors due to the transmission media and its environment) except that with analog there is no way to accurately remove the errors once they are added.
Incredibly messy analog signals carrying digital codes can be cleaned up easily and with stunning accuracy. A good example is the signal that comes out of the phototransistor array in a optical disc player. It looks like $#!! on a scope. Yet quantize it into zeros and ones with some simple comparators and reclock it with a buffer and and a PLL and it is good as new, with parts per million, billion or trillion or better accuracy, only delayed in an moot way.
The basic logic error in their discussion is stigmatizing a data format that they don't like while ignoring the same problems or worse in a data format that they do like.
What do you expect if you have a know-nothing golden eared journalist interviewing a couple of magic cable guys and a magic DAC guy? ;-)
There's no such thing as analog. Every "analog" signal represents a discrete number of electrons being transferred. Therefore, all signals are digital.
This is a better argument than the article's.
"There's no such thing as digital"
Eggs are digital (you buy them by the dozen). Milk is analog (you buy it by the gallon or liter).
Eggs are digital (you buy them by the dozen). Milk is analog (you buy it by the gallon or liter).
Then would eggnog be a hybrid?
You can transmit digital information by river stream with bits or chunks of wood (yeah, it would be slow)
Much cooler example: http://en.wikipedia.org/wiki/IP_over_Avian_Carriers (http://en.wikipedia.org/wiki/IP_over_Avian_Carriers)
Planck discovered that physical action could not take on any indiscriminate value. Instead, the action must be some multiple of a very small quantity (later to be named the "quantum of action" and now called Planck's constant). This inherent granularity is counterintuitive in the everyday world, where it is possible to "make things a little bit hotter" or "move things a little bit faster". This is because the quanta of action are very, very small in comparison to everyday macroscopic human experience. Hence, the granularity of nature appears smooth to us.
These people will never give up. Surrender to the age where science means nothing except a source of impressive terms, where conclusions are drawn based on your pre-conceived faith, and where commerce rules. It will hurt a lot less than trying to "educate the unwilling/disinterested masses" about the ideals of testing methology, peer-review, curiosity-driven fact-finding etc.
Now, where is my homeopathy potion against electromagnetic hypersensitivity?
-k
The problem with this article is that nothing they say is *technically* 100% wrong. It is entirely possible to derive potential, if monstrously tenuous and completely unproven and profoundly absurdly unlikely chains of cause and effect, the systems involved being complicated, that justify almost anything.
I mean, we know that *some* amount of jitter is audible, and we know that regulation cannot *completely* clean up the power supply to the relevant clocking circuitry, and we know that there's *some* change in the demands made on the computer's power supply depending on computer activity...if an infinitesimal, nigh nonexistent effect can be maintained at each stage, you can do this over and over for every branch of physics/EE/whatever under the sun until you conclude, as these articles invariably do, that "DIGITAL AUDIO IS SO COMPLEX (Implied: To all intents and purposes, basically magic. Anything goes. There's no truths in audio. Oh what close-minded dogmatists that may concede that there's such a thing as a pile of crap...)
It's sorta a version of the "Gish Gallop" much derided over at RationalWiki. You see a similar technique with speaker cables: technically, skin effect does have measurable effects at audio frequencies (small increase in resistance at 20kHz), so one can declare that it 'makes a difference' without flat-out making things up, despite the fact that such a claim is manifestly obviously bollocks. Digital audio, and jitter especially, is a rather more complex issue than speaker cables, so the potential to go on and on, hurling potential effect after effect at the audience with no attention to magnitude or the final result, is entirely possible.
And indeed, the engineers involved appear to be in deep enough and sufficiently ignorant of psychoacoustics that if they can lose themselves in these webs of dubious effects, that's good enough for their beliefs.
It amuses me to see all of this hand wringing over picoseconds of jitter in digital audio while at the same time dismissing milliseconds of jitter in analog media like vinyl and tape.
It amuses me to see all of this hand wringing over picoseconds of jitter in digital audio while at the same time dismissing milliseconds of jitter in analog media like vinyl and tape.
+1
There are some fundamentally sound and useful observations in the articles quoted.
Most pertinently, only those living in that transitional late 20th century period would mistake 'digital' as an antonym of 'analog.' That conceptual relationship does need untangling. Digital v Analog makes as much sense as Apples v Food.
Exclusively, precisely: 'digital' refers to the processing of discrete values. Nothing more or less. 'Analog' is a slippery and ubiquitous term: at best, we might contrast digital with an 'analog signal', ie one containing non-quantised information.
Canar's point that 'everything is digital' actually makes sense - unless the distinctly 'analog' string theory is right.
It would have been less controversial if they had maintained focus on the impact of the instrumental basis of the equipment generating and decoding digital data - and used 'digital v analog' rather more carefully. Only programmers with their head in the ether fail to acknowledge that computers are fundamentally mechanical, or that time-domain-sensitive audio playback is not the same as sending a file to a printer.
The straw men arguments above re: information transmission fail to account entirely for time-domain effects: sure you can encode a Miley Cyrus MP3 with wooden blocks, but in real-time decoding the medium and the millieu matter.
Obviously the authors leave open the question of your/my/everyman's 'audibility' of noise and jitter generated by/in digital sources - which isn't a crowd pleaser in certain quarters. But in general you couldn't fairly say they've fumbled the ball.
It would have been less controversial if they had maintained focus on the impact of the instrumental basis of the equipment generating and decoding digital data - and used 'digital v analog' rather more carefully. Only programmers with their head in the ether fail to acknowledge that computers are fundamentally mechanical, or that time-domain-sensitive audio playback is not the same as sending a file to a printer.
I'm not sure what this is intended to mean?
'analog signal', ie one containing non-quantised information.
I don't think it's called analogue because it's continuous. I thought it was called analogue because the current in a wire is analogous to the atmospheric pressure at a particular place and time. Digital could just as easily have been called double analogue because the numbers (ratios) are analogous to the current in a wire.
So there is no such thing as digital. Only double analogue.
'analog signal', ie one containing non-quantised information.
I don't think it's called analogue because it's continuous. I thought it was called analogue because the current in a wire is analogous to the atmospheric pressure at a particular place and time. Digital could just as easily have been called double analogue because the numbers (ratios) are analogous to the current in a wire.
So there is no such thing as digital. Only double analogue.
You should probably read these:
http://en.wikipedia.org/wiki/Digital_signal (http://en.wikipedia.org/wiki/Digital_signal)
http://en.wikipedia.org/wiki/Analog_signal (http://en.wikipedia.org/wiki/Analog_signal)
It would have been less controversial if they had maintained focus on the impact of the instrumental basis of the equipment generating and decoding digital data - and used 'digital v analog' rather more carefully. Only programmers with their head in the ether fail to acknowledge that computers are fundamentally mechanical, or that time-domain-sensitive audio playback is not the same as sending a file to a printer.
I'm not sure what this is intended to mean?
There is a danger of considering 'digital' in purely numerical terms. Focusing like a programmer solely on logical pathways and interpreted values, it's easy to overlook the awkward fact that output is generated by physical machinery, not pumped from some Platonic flowchart. Algorithms are hardware independent, digital audio processors aren't.
It's axiomatic that measurement (and audio) systems respond differently when different mechanical apparatus is integrated into them.
But to be fair to the article, it was Steve Silberman who grabbed the 'no such thing as digital' headline. Charles Hansen put it better, I think, when he said: “All the problems with digital are analog problems'.
There is a danger of considering 'digital' in purely numerical terms. Focusing like a programmer solely on logical pathways and interpreted values, it's easy to overlook the awkward fact that output is generated by physical machinery, not pumped from some Platonic flowchart. Algorithms are hardware independent, digital audio processors aren't.
Fortunately, every single person here knows this to the point where it need not be mentioned.
to be fair to the article, it was Steve Silberman who grabbed the 'no such thing as digital' headline. Charles Hansen put it better, I think, when he said: “All the problems with digital are analog problems'.
That is a very good point indeed. You get rid of many problems, and - once you have chosen a
good enough digital format and
adequate conversion (whenever applicable), the remaining problems are not related to the "digitalness". Of course, if you have a ground loop issue when you conect by metal one device to another, you will also get that when you connect a (copper) cable supposed to carry a digital signal.
There's no such thing as digital problems becoming "There's no such thing as digital". Nifty.
I have only scanned quickly over the article, and I'm trying to point my finger at the very essence of where they're wrong.
AS: Since there's no such thing as 1s and 0s in digital transmission, what is being sent over our USB/Firewire/Ethernet cables when we play back music files?
CH: An ANALOG signal!
Steve Silberman: I think this is where things get misconstrued. The signals we think of abstractly as “digital” are in fact high-speed analog square waves, susceptible to all of the same damage and distortions as any other analog signal.
This made me cringe. It seems someone doesn't know the difference between "data" and "signal"...
When saying there is a "gray" area between a digital "0" and a digital "1", and the errors that might creep into data corruption, it makes me believe they never heard of differential data transmission (USB and Ethernet is a prime example).
They haven't addressed any technicalities of things like S/PDIF, just stated it's "flawed".
The kind of noise (that is apparently audible to them) would interfere with the general operation of any digital device, but they still seem to have no idea about "data" and things like differential transmission over longer distances.
I don't see a single reference to a research paper or even Wikipedia article, they have a list of books at the end, but what they should've read is an introductory book for a signal processing course at University...
To me, most of the stuff is what I call
Audiophoolery, it's kinda blending into esoteric beliefs.
'analog signal', ie one containing non-quantised information.
I don't think it's called analogue because it's continuous. I thought it was called analogue because the current in a wire is analogous to the atmospheric pressure at a particular place and time. Digital could just as easily have been called double analogue because the numbers (ratios) are analogous to the current in a wire.
So there is no such thing as digital. Only double analogue.
Behold the conceptual slipperiness of 'analog'; we can't even decide how it's spelt! There is a relationship between analog and analagous, but I can't think of an analogy to explain it . . .
Calling both optical and coaxial SPDIF cables 'digital' is peculiarly reprehensible, too: fibre cable involves properly quantised transmission. Henceforth, I move that we call optical SPDIF and digital signal over coax 'SPIF' to avoid confusion.
to be fair to the article, it was Steve Silberman who grabbed the 'no such thing as digital' headline. Charles Hansen put it better, I think, when he said: “All the problems with digital are analog problems'.
That is a very good point indeed. You get rid of many problems, and - once you have chosen a good enough digital format and adequate conversion (whenever applicable), the remaining problems are not related to the "digitalness". Of course, if you have a ground loop issue when you conect by metal one device to another, you will also get that when you connect a (copper) cable supposed to carry a digital signal.
There's no such thing as digital problems becoming "There's no such thing as digital". Nifty.
I'm not sure what an 'analog problem' is. Seems indivisible from the problem of creating an analog of a recording in a listening room.
And I'm definitely not sure whose definition of 'good enough' we should go with.
'analog signal', ie one containing non-quantised information.
I don't think it's called analogue because it's continuous. I thought it was called analogue because the current in a wire is analogous to the atmospheric pressure at a particular place and time. Digital could just as easily have been called double analogue because the numbers (ratios) are analogous to the current in a wire.
So there is no such thing as digital. Only double analogue.
Behold the conceptual slipperiness of 'analog'; we can't even decide how it's spelt! There is a relationship between analog and analagous, but I can't think of an analogy to explain it . . .
Calling both optical and coaxial SPDIF cables 'digital' is peculiarly reprehensible, too: fibre cable involves properly quantised transmission. Henceforth, I move that we call optical SPDIF and digital signal over coax 'SPIF' to avoid confusion.
An analogue signal has nothing to do with being continuous.
An analogue signal is representing one set of physical information, with
another physical information. For instance: the gas paddle in your car, uses a wire that is being pulled or pushed: that's an analogue signal (one of those mechanical signals goes to an automatic transmission, for instance). If you have an ambient pressure sensor, going from 0.8bar to 1.2bar and it represents this data with an analogue signal using electricity, it might be over the range of 2.0V to 4.5V. Now, if that data is send over a line as digital signal, this adds the digital data layer to it: a "0" is >3.5V and a "1" is <2.5V, for instance. Now, you might wanna add things like self-clocking, etc. to it, whatever. Now, this digital data.
OK, let's get the "continuous" problem out of the way: A digital or analogue signal, both can be intermittent (as in: noncontinuous).
I'm not gonna go into the technicalities how and why a digital signal is easier to strip from errors, and how to make a digital line more robust against noise, without adding finger-thick shielding. I invite anyone who's interested, to go their local university, and sneak into the signal processing course...
It would have been less controversial if they had maintained focus on the impact of the instrumental basis of the equipment generating and decoding digital data - and used 'digital v analog' rather more carefully.
Don't you think he
was being deliberately controversial?
After all, with a speech that nonsensical, I guess controversy is his best tool of the trade.
An analogue signal has nothing to do with being continuous.
An analogue signal is representing one set of physical information, with another physical information. For instance: the gas paddle in your car, uses a wire that is being pulled or pushed: that's an analogue signal (one of those mechanical signals goes to an automatic transmission, for instance). If you have an ambient pressure sensor, going from 0.8bar to 1.2bar and it represents this data with an analogue signal using electricity, it might be over the range of 2.0V to 4.5V. Now, if that data is send over a line as digital signal, this adds the digital data layer to it: a "0" is >3.5V and a "1" is <2.5V, for instance. Now, you might wanna add things like self-clocking, etc. to it, whatever. Now, this digital data.
OK, let's get the "continuous" problem out of the way: A digital or analogue signal, both can be intermittent (as in: noncontinuous).
I'm not gonna go into the technicalities how and why a digital signal is easier to strip from errors, and how to make a digital line more robust against noise, without adding finger-thick shielding. I invite anyone who's interested, to go their local university, and sneak into the signal processing course...
You're on a wild herring chase, I think: the contention is not 'analog = continuous’. The differentiator with regard to signalling is that analog is
non-quantised. It could be any medium translating an input to an output.
With a small correction the first sentence in your second paragraph is spot on . . . “An analogue [] is respresenting one set of physical information with another physical information”.
It would have been less controversial if they had maintained focus on the impact of the instrumental basis of the equipment generating and decoding digital data - and used 'digital v analog' rather more carefully.
Don't you think he was being deliberately controversial?
After all, with a speech that nonsensical, I guess controversy is his best tool of the trade.
Sure - it's a headline: its job is to grab attention. It worked. So shoot him. Nonsense or nay depends on how you define 'thing'!
As the OP feared, apart from that, it's all sensible stuff. The only objection you might fairly make is that they don't address audibility thresholds. I would guess they know enough about that to avoid getting into it (again, see Dunning-Kruger).
So shoot him. Nonsense or nay depends on how you define 'thing'!
Way OTT.
When saying there is a "gray" area between a digital "0" and a digital "1", and the errors that might creep into data corruption, it makes me believe they never heard of differential data transmission (USB and Ethernet is a prime example).
Which is not used in S/PDIF, eh? USB and ethernet communicate two-way, and data failing the CRC-check can be re-sent. S/PDIF is one-way communication and cannot. RAM has ECC, why is that? Because it has to cope with bit-errors by itself. Hard drives' firmware implement ECCs for the same reason, but re-reads are sometimes indeed necessary (have a look at this: http://www.youtube.com/watch?v=tDacjrSCeq4 (http://www.youtube.com/watch?v=tDacjrSCeq4) ). And even that is not satisfactory for all applications, which is why you hear buzzwords as
silent data corruption and
end-to-end data protection. "Digital" data transfer does not imply that these measures are unnecessary, rather it means that they are possible (or at least way more feasible) to implement. S/PDIF uses very little of it, but fortunately the data stream and real-time threshold are not awfully difficult to cope with, and contrary to audiophile belief, a single-sample error a day - or what the failure rate is in practice - won't crash a plane nor your bank account.
Calling both optical and coaxial SPDIF cables 'digital' is peculiarly reprehensible, too: fibre cable involves properly quantised transmission. Henceforth, I move that we call optical SPDIF and digital signal over coax 'SPIF' to avoid confusion.
Optical or coax, they are both transmission channels for digital communication. On this subject, you appear to be miserably clueless.
There's no such thing as analog. Every "analog" signal represents a discrete number of electrons being transferred. Therefore, all signals are digital.
This is a better argument than the article's.
+1
I looked it up and I believe digital means making music with your fingers and toes, and analog means reproducing music only with devices that work the same as the original instruments (string like speakers for guitars for example.) They make dictionaries for a porpoise, people!
It would have been less controversial if they had maintained focus on the impact of the instrumental basis of the equipment generating and decoding digital data - and used 'digital v analog' rather more carefully.
Don't you think he was being deliberately controversial?
I read a few of his pieces and decided that he is an entertaining writer but that he either doesn't read his references or is a deliberate
liar practitioner of poetic liberties.
Case in point:
http://www.wired.com/medtech/drugs/magazin...t?currentPage=2 (http://www.wired.com/medtech/drugs/magazine/17-09/ff_placebo_effect?currentPage=2)
"Placebos Are Getting More Effective. Drugmakers Are Desperate to Know Why."
"Beecher's prescription helped cure the medical establishment of outright quackery, but it had an insidious side effect. By casting placebo as the villain in RCTs, he ended up stigmatizing one of his most important discoveries."
He cites:
"The Powerful Placebo"
http://www.jgh.ca/uploads/psychiatry/links/beecher.pdf (http://www.jgh.ca/uploads/psychiatry/links/beecher.pdf)
Beecher's article's conclusion starts out "When subjective responses, symptoms, are under
study, it is apparent that the high order of effectiveness
of placebos must be recognized."
..and goes on in the same spirit: positive.
No way was the placebo cast as a villain.
I call that taking liberties with one's main reference. Read the articles for yourself and reach your own conclusions. ;-)
It would have been less controversial if they had maintained focus on the impact of the instrumental basis of the equipment generating and decoding digital data - and used 'digital v analog' rather more carefully. Only programmers with their head in the ether fail to acknowledge that computers are fundamentally mechanical, or that time-domain-sensitive audio playback is not the same as sending a file to a printer.
I'm not sure what this is intended to mean?
There is a danger of considering 'digital' in purely numerical terms. Focusing like a programmer solely on logical pathways and interpreted values, it's easy to overlook the awkward fact that output is generated by physical machinery, not pumped from some Platonic flowchart. Algorithms are hardware independent, digital audio processors aren't.
It's axiomatic that measurement (and audio) systems respond differently when different mechanical apparatus is integrated into them.
I think you're forgetting how audio (and more generally modern) systems are designed. We purposefully build them to be linear time invariant so that we don't have to consider the "awkward fact" that they're are many components of varying nature in a system. Since they are LTI, the final output is simply the superposition of the underlying components. Hence, we can look at each component in isolation without loss of accuracy.
So I would say that it only looks like people are being myopic until you more carefully think about the problem, and then you realize that actually the people behind it have been 1000x more clever than you initially thought.
Calling both optical and coaxial SPDIF cables 'digital' is peculiarly reprehensible, too: fibre cable involves properly quantised transmission.
No this is absolutely false. Both coax and fiber are waveguides. At the level of the medium there is no difference between them besides frequency, with coax topping out at a few GHz, and fiber topping out at a few hundred THz. There is no requirement of "properly quantised transmission" in either system. Both can carry quantized or unquantized information.
FWIW, if you think about this more carefully, you will realize that it is impossible for any transmission medium to enforce one kind of transmission over another. Such things are the properties of receivers, not of the medium.
I'm not sure what an 'analog problem' is.
Problems that occur in the analog domain. The example of a ground loop seems fitting.
Case in point:
http://www.wired.com/medtech/drugs/magazin...t?currentPage=2 (http://www.wired.com/medtech/drugs/magazine/17-09/ff_placebo_effect?currentPage=2)
"Today, to win FDA approval, a new medication must beat placebo in at least two authenticated trials."
It's nothing more than a pipe dream, I know, but if only similar standards were also required from audio equipment suppliers, maybe we'd have less of these magazines and websites (along with their plethora of self-proclaimed experts) where pseudo science and blurb seem to thrive.
Calling both optical and coaxial SPDIF cables 'digital' is peculiarly reprehensible, too: fibre cable involves properly quantised transmission. Henceforth, I move that we call optical SPDIF and digital signal over coax 'SPIF' to avoid confusion.
Optical or coax, they are both transmission channels for digital communication. On this subject, you appear to be miserably clueless.
My jokey post in response to a jokey post (SPIF? really?!) contained a kernel of truth, though: in transit, an optical cable conforms in the purest sense to the strictest definition of digital: quantised. And only 'not pulses of light' when interpreted by a digital transceiver. Whereas the physical properties of a coaxial cable inevitably bring 'analog domain' effects into play. My objection was to to the term 'digital cable', which is ubiquitous but nonsensical.
It would have been less controversial if they had maintained focus on the impact of the instrumental basis of the equipment generating and decoding digital data - and used 'digital v analog' rather more carefully. Only programmers with their head in the ether fail to acknowledge that computers are fundamentally mechanical, or that time-domain-sensitive audio playback is not the same as sending a file to a printer.
I'm not sure what this is intended to mean?
There is a danger of considering 'digital' in purely numerical terms. Focusing like a programmer solely on logical pathways and interpreted values, it's easy to overlook the awkward fact that output is generated by physical machinery, not pumped from some Platonic flowchart. Algorithms are hardware independent, digital audio processors aren't.
It's axiomatic that measurement (and audio) systems respond differently when different mechanical apparatus is integrated into them.
I think you're forgetting how audio (and more generally modern) systems are designed. We purposefully build them to be linear time invariant so that we don't have to consider the "awkward fact" that they're are many components of varying nature in a system. Since they are LTI, the final output is simply the superposition of the underlying components. Hence, we can look at each component in isolation without loss of accuracy.
So I would say that it only looks like people are being myopic until you more carefully think about the problem, and then you realize that actually the people behind it have been 1000x more clever than you initially thought.
Calling both optical and coaxial SPDIF cables 'digital' is peculiarly reprehensible, too: fibre cable involves properly quantised transmission.
No this is absolutely false. Both coax and fiber are waveguides. At the level of the medium there is no difference between them besides frequency, with coax topping out at a few GHz, and fiber topping out at a few hundred THz. There is no requirement of "properly quantised transmission" in either system. Both can carry quantized or unquantized information.
FWIW, if you think about this more carefully, you will realize that it is impossible for any transmission medium to enforce one kind of transmission over another. Such things are the properties of receivers, not of the medium.
I'm not sure what an 'analog problem' is.
Problems that occur in the analog domain. The example of a ground loop seems fitting.
You've further split my already split hairs. And we could keep on splitting down to the bottom of distinction between wave energy and 'digital' matter. Any way you slice it, optical transmission is a step closer to purely quantised digital than propagation through copper. Your ground loop illustrates the point.
With regard to the real-world translation of LTI goals, look no further than the influence of proximal power regulation to clock at either end of that optical cable. in the field, time invariance doesn't prevail.
But I would accept that it's more helpful to concentrate on the receiver than the medium, and in fact you could make a sound practical argument for 'digital's’ threshold of definition occurring right there.
Any way you slice it, optical transmission is a step closer to purely quantised digital than propagation through copper.
This is absolutely, completely false. I'm not sure how you think copper and/or fiber work, but your internal model is very wrong.
With regard to the real-world translation of LTI goals, look no further than the influence of proximal power regulation to clock at either end of that optical cable. in the field, time invariance doesn't prevail.
Can you explain what you mean? I don't know how to parse this statement as you have written it.
Calling both optical and coaxial SPDIF cables 'digital' is peculiarly reprehensible, too: fibre cable involves properly quantised transmission. Henceforth, I move that we call optical SPDIF and digital signal over coax 'SPIF' to avoid confusion.
Optical or coax, they are both transmission channels for digital communication. On this subject, you appear to be miserably clueless.
My jokey post in response to a jokey post (SPIF? really?!) contained a kernel of truth, though: in transit, an optical cable conforms in the purest sense to the strictest definition of digital: quantised. And only 'not pulses of light' when interpreted by a digital transceiver. Whereas the physical properties of a coaxial cable inevitably bring 'analog domain' effects into play. My objection was to to the term 'digital cable', which is ubiquitous but nonsensical.
Sorry just saw this, but no, you're totally wrong here. Any electromagnetic field capable of carrying information is quantized in power. No matter the frequency.
EM waves in a fiber are encoded by an integer number of photons. EM waves in coax are quantized into an integer number of photons. The 60Hz noise on your power supply comes in an integer number of 60Hz photons.
You're probably not aware of this though because it doesn't matter. You're trying to make the jump from "field is quantized" to "must be digital", but thats not correct. The quantization level spacing for all 3 of those examples is far smaller than thermal noise at room temperature, so the statistics of any of the media are capable of being both analog or digital depending on the encoding chosen by the users of the medium.
"Today, to win FDA approval, a new medication must beat placebo in at least two authenticated trials."
It's nothing more than a pipe dream, I know, but if only similar standards were also required from audio equipment suppliers, maybe we'd have less of these magazines and websites (along with their plethora of self-proclaimed experts) where pseudo science and blurb seem to thrive.
Repeat sentence with "audio equipment" replaced by "cosmetics".
An analogue signal has nothing to do with being continuous.
Actually, being continuous in time and amplitude is sufficient to make something an analog system.
An analogue signal is representing one set of physical information, with another physical information. For instance: the gas paddle in your car, uses a wire that is being pulled or pushed: that's an analogue signal (one of those mechanical signals goes to an automatic transmission, for instance). If you have an ambient pressure sensor, going from 0.8bar to 1.2bar and it represents this data with an analogue signal using electricity, it might be over the range of 2.0V to 4.5V. Now, if that data is send over a line as digital signal, this adds the digital data layer to it: a "0" is >3.5V and a "1" is <2.5V, for instance. Now, you might wanna add things like self-clocking, etc. to it, whatever. Now, this digital data.
Thats actually not necessarily sufficient to make it digital. By convention, all discrete time, discrete amplitude systems are digital. Discrete amplitude, continuous time systems may or may not be considered digital depending on the context. In digital communications, that might be considered digital, but in most other contexts simply using a discriminator is not sufficient (and hence an discriminator is not considered an ADC).
FWIW, my favorite example of an analog system that uses binary encoding is laser disk.
OK, let's get the "continuous" problem out of the way: A digital or analogue signal, both can be intermittent (as in: noncontinuous).
This depends on your context and may or may not be true depending on what you are doing with it.
I'm not gonna go into the technicalities how and why a digital signal is easier to strip from errors, and how to make a digital line more robust against noise, without adding finger-thick shielding. I invite anyone who's interested, to go their local university, and sneak into the signal processing course...
In a signal processing course, digital is generally used exclusively to refer to discrete time and amplitude, as DSP techniques do not generally apply to continuous time systems.
I don't understand the concern about transfer of digital information over analog channels. Say 0.8V and higher voltage is a zero and 0.3V or less is a one (similarly for transmissions based on radio waves/light or whatever else). Anything in between is undefined, i.e. it may be interpreted as either a zero or a one "randomly". In case your chosen transmission channel then produces too many undefined values on the receiving end, you add enough redundant error correction coding to your data (or decrease the data rate) to make up for it.
While the above is grossly over-simplified (e.g ignores analog signal modulation techniques), it's no rocket science.
Entirely analog transmission just does not (cannot?) precisely remove/isolate channel noise from information. I don't think there's anything beneficial about that.
Entirely analog transmission just does not (cannot?) precisely remove/isolate channel noise from information. I don't think there's anything beneficial about that.
Actually it can, but those methods are so 'simple' that most might not call it error correction at first look. Differential signaling (balanced audio cables) is one trick where part of the noise coupled on a line can be easily filtered. It works really well for most noise sources.
Entirely analog transmission just does not (cannot?) precisely remove/isolate channel noise from information. I don't think there's anything beneficial about that.
Actually it can, but those methods are so 'simple' that most might not call it error correction at first look. Differential signaling (balanced audio cables) is one trick where part of the noise coupled on a line can be easily filtered. It works really well for most noise sources.
The example makes the point that in the analog domain, there's a strong tendency to need a separate fix for every different kind of problem, which gets complicated and expensive.
For example, how do you remove FM distortion from an analog link?
With digital, the making the medium work at all is the fix for just about everything, and you can "fix everything" relatively simply. CD players are a good example of this.
When I said "precisely remove/isolate channel noise from information", I meant "precisely" as in "perfectly". Twisted cables, differential signalling and whatnot are far from perfect and they only solve (mitigate) specific sources/mechanisms of transmission channel noise.
Ok, so physics indicates that the world is "discrete" (at least the kind of physics that most of us know). So analog signals are really digital.
And digital communications happens over links that may be the same as the old "analog" links. So digital is really analog.
This rapidly leads into a pointless debate over words. What matters is the end-result. By treating the channel in an abstracted manner ("digital"), we can do logical operations that relies on things like memory and mathematical blocks to combat channel issues that are very hard to cure using a limited set of "analog" means.
Back in the day, I had poor image quality in my tv, but it worked. Now enter the digital age, everything happens over digital, encrypted links that "negotiate" the format. And my Sony tv and Onkyo amplifier cannot seem to agree on how lipsync info is to be transmitted over HDMI 1.4, as a result people move their lips 150ms after I hear them talking. Or some source cannot be shown because of HDMI DHCP issues. Or my tv channels might not have analog echo and noise (continous problems), but rather there are sudden loud transients (lost packets) or pixelated video. It seems that every technological advance made in the last 30 years have been used to give us more crap, more compromise, less reliability. Talk about progress :-)
-k
It seems that every technological advance made in the last 30 years have been used to give us more crap, more compromise, less reliability. Talk about progress :-)
This reminds me of a tweet. From memory:
"It's the future! Where are my flying cars?" he spoke instantly to an audience of thousands from a device in the palm of his hand
I read a few of his pieces and decided that he is an entertaining writer but that he either doesn't read his references or is a deliberate liar practitioner of poetic liberties.
Case in point:
http://www.wired.com/medtech/drugs/magazin...t?currentPage=2 (http://www.wired.com/medtech/drugs/magazine/17-09/ff_placebo_effect?currentPage=2)
"Placebos Are Getting More Effective. Drugmakers Are Desperate to Know Why."
"Beecher's prescription helped cure the medical establishment of outright quackery, but it had an insidious side effect. By casting placebo as the villain in RCTs, he ended up stigmatizing one of his most important discoveries."
He cites:
"The Powerful Placebo"
http://www.jgh.ca/uploads/psychiatry/links/beecher.pdf (http://www.jgh.ca/uploads/psychiatry/links/beecher.pdf)
Beecher's article's conclusion starts out "When subjective responses, symptoms, are under
study, it is apparent that the high order of effectiveness
of placebos must be recognized."
..and goes on in the same spirit: positive.
No way was the placebo cast as a villain.
I call that taking liberties with one's main reference. Read the articles for yourself and reach your own conclusions. ;-)
I thought, it being Wired, something must have been written about this in the science blogs, and sure enough: http://www.sciencebasedmedicine.org/placeb...ou-think-it-is/ (http://www.sciencebasedmedicine.org/placebo-is-not-what-you-think-it-is/)
Silberman apparently shows up (http://www.sciencebasedmedicine.org/placebo-is-not-what-you-think-it-is/comment-page-1/#comment-30414) in the comments, if you're interested what he has to say in his defense.
Or my tv channels might not have analog echo and noise (continous problems), but rather there are sudden loud transients (lost packets) or pixelated video. It seems that every technological advance made in the last 30 years have been used to give us more crap, more compromise, less reliability. Talk about progress :-)
DVB-T has many channel configuration options, see the table at the end of the following chapter:
http://en.wikipedia.org/wiki/DVB-T#Technic...B-T_transmitter (http://en.wikipedia.org/wiki/DVB-T#Technical_description_of_a_DVB-T_transmitter)
It could be configured for better error tolerance at the cost of usable channel bandwidth.
In addition, the digital terrestrial television technologies deployed today are quite dated. DVB-T, for example, does not allow to carry a base high-error-tolerance low-fidelity stream together with a higher definition additional stream (with lower error tolerance). This would allow people with poor signal to still watch some low-fidelity TV instead of a blank screen.
Ultimately, it's a matter of choice how the digital technologies are designed and utilized. It's not the fault of "digital" per se.
Digital error correction is actually much more robust than older analog schemes. However, as we have built more digital devices, we have used more and more spectrum, requiring that transmissions be squeezed into as little bandwidth and power as physically possible.
If you actually just wanted to transmit 480i video on the old USA VHF channel 2 at 1960s power levels, you could have dependable H.264 video at distances where analog wouldn't even show up. However no one does this because that would waste a large amount of very precious RF spectrum.
Entirely analog transmission just does not (cannot?) precisely remove/isolate channel noise from information. I don't think there's anything beneficial about that.
Actually it can, but those methods are so 'simple' that most might not call it error correction at first look. Differential signaling (balanced audio cables) is one trick where part of the noise coupled on a line can be easily filtered. It works really well for most noise sources.
You're massively overestimating how powerful those techniques are. Balanced detection has a limited common mode rejection ratio (CMRR) in any practical device. For example, in analog optical systems, obtaining even 30dB CMRR is considered outstanding. Do that on a noisy channel and you have a noise floor of 30dB. In such circumstances some form of digital error correction is essentially mandatory.
Communication Systems Engineering by Proakis and Salehi is a good book on the subject. Some of you should study it or something similar. It would cut down on a lot of the nonsense being said.
Proakis's text is my go-to reference on discrete time systems, but I think that book is probably well beyond most of the people posting in this thread, and does not go into any detail on the electromagnetic physics needed to understand the performance of analog systems.
The core problem here is that everyone has opinions about analog and digital systems, and for digital they may even be fairly sensible-ish, but to really understand analog you need a graduate level education in EE. Without that it is almost impossible to really understand what is happening and you're just going to get lost in a mess of meaningless arguments about what is and isn't quantized without understanding what it really means.
...to really understand analog you need a graduate level education in EE. Without that it is almost impossible to really understand what is happening and you're just going to get lost in a mess of meaningless arguments about what is and isn't quantized without understanding what it really means.
Yeah, with my CS degree I only had to take a couple of EE/CE courses, and I couldn't tell if the article was spouting pseudoscience, or just EE knowledge that was over my head.
That is, until I got to this part:
...unless one is planning to use a wireless link, which has not only potential long-term health risks...
... Okay then, I know I can ignore this article entirely.