|Style over substance|
Audio woo consists of unsupportable claims for devices or methods for getting better sound quality from systems that reproduce recorded music. Such claims are made by manufacturers, hobbyists, and writers in the field.
The quest for perfect sound
An audiophile is someone dedicated to producing the highest possible fidelity in the playback of music. While this can certainly be a reasonable goal, much of the industry that caters to them sells extremely high priced equipment that claims to improve sound through highly dubious mechanisms and/or badly misunderstood real science. Most double-blind studies have shown that there is no difference in detectable sound quality for most of the equipment sold, and some of it is actually inferior in quality to less expensive products. Many of the products use appeals to magical thinking and pseudoscience to explain their mechanisms. While some of them do indeed provide a small, subtle improvement in sound quality, the unscientific claims by the manufacturers as to why they might sound better lead to a subculture of people who are utterly deluded about how to get the best sound from an audio system.
These particular audiophiles are widely known in the electronics trade as people who put subjective sound experience above objective measurement as the primary way to judge sound reproduction; as a result, many refuse to admit that their beloved "warmer" sound (often lamented as having disappeared with the introduction of transistors, and then the compact disc) is actually a form of distortion. Double-blind testing, as in many other pseudoscientific pursuits, is shunned since it removes the power of suggestion (by brand name, equipment appearance, or price) from the testing room and has a frustrating inability to confirm their prior prejudices. At the extreme end of this particular brand of audiophiles are the "tweakers" whose obsession leads to belief in some of the more bizarre claims (read scams) of the subjective audiophile industry.
There exists an inverted class of audiophiles who claim that the sound of vinyl, i.e. records played with a contact stylus, is superior to the sound of compact discs or other digital music - a interesting school of thought given the analogue/digital audio debate.
An anonymous wag, in the tradition of The Devil's Dictionary by Ambrose Bierce, once defined such an audiophile as someone who listens to the stereo, not the music. Indeed, there are even people who collect and assemble such systems with no goal of even listening to music!
There is some validity to claims of increased sound quality under certain circumstances. The electrical principles at work here are well understood, and there are real engineers and scientists who work in this field. Noise can be reduced by improved filtering; frequency response through more flexible, higher quality diodes and amplifiers; and so on. However, the major stumbling blocks are the recording quality and later compression and manipulation of the original signal itself, the cost of better parts, and just how much any change actually increases quality. Once a certain point is reached, any further increases become harder to detect. The point of diminishing returns is rapidly reached.
Explanations for why manufacturers, audiophiles and the "tweakers" perpetuate the myths of the subjective audio crowd are varied but one simple explanation is cognitive dissonance: having just spent thousands of dollars on something that does very little or nothing, it is easier to adjust one's thinking to "Wow! spending hundreds of dollars on little ceramic insulators that keep my speaker wires off the floor has opened up the sound stage and improved the distinction between the violin parts!", than it is to admit the fundamental realities that say "I can't believe how dumb I have been" and "my wife will kill me!". The brilliant thing about this is that it works in inverse proportionality to how dumb the purchase was - so the more ridiculous the tweak, the more believing the purchaser will be. This brings into question the review policy of subjective audio reviewers such as Positive Feedback, where reviewers often own the equipment they are reviewing.
Explanations can also include the concepts of buyer's remorse, which can actually fuel the effect in a negative feedback loop.
The tube thing
Subjectivist audiophiles often express a preference for tube-based amplifiers due to their perception of a cleaner sound, to the point where scam artists have sold products claiming to make transistors function like tubes.
As a general rule, to mainstream audio folks tubes are not inherently bad, though perhaps a bit high-maintenance; however, while tubes are very popular in guitar and other instrument amplifiers due to their unique sound (even-harmonic distortion resulting from asymmetric soft clipping), they are unnecessary in playback amplifiers. Skeptics of tube amplifiers argue that audiophiles prefer the inherent distortion provided in the antiquated technology while paradoxically claiming the sound is purer and more perfect. 
The tube thing can get truly stupid. A vacuum tube computer sound card, anyone?
The vinyl thing
A great many audiophiles hold up vinyl records as being the be-all and end-all of sound reproduction. Vinyl, though largely obsolete since the CD improved portability and instant gratification, does have a few uses, particularly archival (it's substantially more robust than optical or magnetic media). In addition, the limitations of vinyl records spared them from the "Loudness Wars" — the constant race among record producers to use audio compression to make songs as loud as possible, often distorting the sound in the process.
However, vinyl records inevitably suffer from wear during playback, especially with use of a needle rather than a laser or a scanner -- so unless you're willing to shell out multiple thousands of dollars for a laser turntable, your records won't last for long. Dynamic range is also limited -- depending on bandwidth, sample rate, and storage capacity, a digital format can equal the performance of an analog medium of equivalent bandwidth (a CD, which can sample just outside the range of human hearing, can produce a much more faithful representation than a standard vinyl LP).
The bi-wiring thing Unfortunately, few hi-fi speaker manufacturers dare to release speakers without bi-wiring ports as the all-prevailing nature of the myth would mean they simply would not be able to sell them. To quote an honest manufacturer:
"Will it sound any different if you biwire? Some users think it does, but I've never heard any differences, nor have any of our laboratory measurements or scientifically controlled double blind listening tests ever demonstrated there are audible differences. Axiom includes the extra terminals as a nod to those enthusiasts who believe that biwiring results in audible benefits and for the bi-ampers."
Note that the bi-wiring inputs provided are essential for those who intend to bi-amp their speakers. However, as many of these speakers contain passive crossovers, it would seem that they defeat the purpose of bi-amping, the ultimate benefits of which are derived precisely from avoiding these crossovers.
The digital thing
Three words: bits is bits.
Okay, here's more: digital data streams don't work much like analog streams, at least from the user's point of view (the actual situation makes up the greater part of a textbook on signal processing, or at least a fair handful of Wikipedia pages). At least from the point of view of content, digital signals don't degrade gracefully; they drop hard and lose big whacks of information as the decoding hardware on the client end gets confused. When an analog signal is weakened, you get signal attenuation and snow; when a digital signal degrades, you get dropouts, stutters, and (on video hardware) blocky and smearing images. (There's a couple of different ways of sending digital signals. For our purposes we'll focus on electrical rather than optical transmission; on the consumer level, optical connections like TOSlink are subject to the same sort of woo but it makes even less sense.)
The first thing you need to know is the basics of how analog signals are converted to digital, starting with the Nyquist-Shannon sampling theorem and how it relates to signal encoding. To understand what a digital signal looks like, imagine a continuous wave signal of N hertz; in order to represent this frequency at all, the sample rate has to be at least 2N (in simple terms, at least one point for the peak of the wave and one point for the trough). In digital signal processing, each sample is now reduced to a digital number, usually a scalar value, and read out as a stream of bits to the destination device.  This approach is more or less the basis of the usual representation of uncompressed audio data on a CD or in a .wav file, and the data can be compressed, transformed, or transmitted like any other form of digital data.  As a result, though any digital signal is going to be an approximation of the analog signal it represents, its precision can be as narrow as you need it to be, within the limits of the sampling device. On a reliable link, an uncompressed signal has zero generation loss between source and destination.
The important thing to know is that when discussing ways to "improve" a digital signal, the analog carrier is a low-level  issue that is largely transparent to the parts of the system that actually handle the data. If an analog carrier signal is corrupted on an analog device, the result may be scrambled, attenuated, or otherwise damaged, but the signal will still get through; however, if a digital signal is disrupted, the receiving hardware won't be able to make sense of the analog signal and will therefore decode either a wrong value or no value at all; barring some seriously incompetent circuit board design, the input of the analog stage has no influence on the input to the digital stage. Subtle signal changes claimed by people who spend extra cash on pricey HDMI, TOSlink, and USB cables simply can't be caused by changing cables, since it implies that the cable is actually physically changing the data it passes through.  Fundamentally, then, what's going on with digital audiophoolery is a failure to understand the difference between analog and digital communication. You may as well expect your software to run faster or your cheezy fanfics to read like Stephen King or Toni Morrison.
There are cases where cable quality makes a difference; for example, cheap straight-wire Cat3 Ethernet cable is no longer widely available, since it results in excessive crosstalk at high speeds resulting in massive packet loss; the twisted-pair wires favored for most telecom and data applications work in a way distantly related to shielded coax to prevent that. But it is still the case that from the digital standpoint, the result of a low-quality cable is not data corruption but outright data loss. And fiber optic isn't subject to this to begin with, as light transmission physics are rather different from the radio band signals used by electrical connections. This also applies to storage media -- a CD could have a scratch that affects tracking, optical imperfections that screw with the laser, or physical damage to the recording layer, but in either case you won't hear a degraded signal because you won't hear anything at all. (DVDs and Blu-Rays even more so, since they don't deal in raw data streams at all but computer files -- if your DVD isn't being read properly, the signal chain is the least of your problems.)
Note that when dealing with compressed AV streams, the rules do change; most audio compression formats (MP3, AAC, Ogg Vorbis, etc) and virtually all currently widely used video formats (MPEG-x, DV, AVCHD, etc) throw out large amounts of data during the compression process in order to bring the file size down to something manageable. This usually won't be a problem for the majority of listeners/viewers, though sharp eyes and ears can detect artifacts of the compression process even at high bitrates, and transcoding between compressed formats will cause generational loss by amplifying existing artifacts and creating new ones. However, this is not really relevant to the discussion of audio woo as it is actual double-blind demonstrable fact. If for some reason you do need to tweak a digital signal, it's possible to patch a computer or effects processor into the analog stream like musicians do, but you'd likely have to write your own effects filters in a language like Nyquist or build them yourself in a manner similar to an analog stomp box for a guitar.
Audio cables are inherently very cheap to make, store and transport. The vast majority of cable sellers do not manufacture anything: they choose generic products from industrial cable makers, add some cosmetic changes for brand differentiation and then apply huge markups to the retail price. This is abetted by the fact that audio cables have practically zero maintenance, spare-part, and after-sales service costs.
Audio dealers also favor expensive cables, as they can add their own excessive markup on products that cannot malfunction and will never require servicing, spare parts, or any kind of dealer support. Many dealers utilize a "fake choice" trick on unsuspecting consumers: they are quick to loan a variety of cables for a home trial, giving the consumer the "choice" of keeping and paying for the cable they prefer and returning the rest. The returned cables have no sign of having been used before and retain their "brand new" status and perceived value. The consumer is thus fooled into making an entirely free choice (between similarly overpriced products) and the dealer clinches a profitable sale regardless.
Over-engineered and insanely expensive power cables are perhaps the easiest product to dismiss. The claim that 6 feet of $1,800 power cabling improves sound quality is ludicrous for the same reason that using such a cable won't make a DVD player display a sharper picture, although similar claims are made to that effect. The claim also ignores the many hundreds of miles of boring cable being used to deliver the electricity to the house of the audiophile.
The effectiveness of the magical power cable would rely on the following assertion: The characteristics of the electricity are being altered by their passage through the magical cable. In more scientific terms, certain cables can, in fact, decrease resistance, heat and long distance power losses, however when it comes to a home audio system, none of these are relevant. Resistance of the cable is tiny compared to internal resistance, heat is not an issue, and you're not going 3000 miles with your audio signal. The cheap copper cable your equipment came with is more than adequate.
Luckily this can be very easily tested, although this is not something to do at home! The following steps are dangerous to you and your equipment, and provided for illustrative purposes only. Do this stuff at your own risk.
You could hook-up an oscilloscope or a spectrum analyzer directly to the mains current. This provides a baseline measurement. Next you can take the same measurements with both a regular power cord, and then with the expensive cable.
The AC mains current will run through a transformer in order to adjust the power before it's routed to the various components of the equipment. The same measurements can be applied to the outputs of the transformer to see if swapping the power cords makes any difference. This is ideally done as a blind test, so that the person taking the measurements doesn't know which cable is currently being used. At this stage, it's unlikely that there will be any difference in the readings.
If there are no obvious difference found in these tests, then how on earth could the power cord be making a difference to sound quality? Fairies cleaning the electrons in some way invisible to physics?
All data transmission standards specify a set of minimum performance parameters a cable must meet in order to be certified as standards-compliant; thus, even if it's true that a super-engineered HDMI cable will have a far better physical signal quality than a philistine HDMI cable, in the end it just won't matter for the simple reason that both cables fulfill the minimum specifications required to correctly transmit an HDMI signal.
Bits is bits. But you can still get $500+ Ethernet, USB or digital interconnect cables if you are willing to ignore all that rubbish about how digital signals aren't affected by cable quality at the short distances found in the average home.
Abuse of science
As more and more of the audio chain becomes integrated and non-user-serviceable, some hobbyists have begun to focus on small bits of solid-state physics in the attempt to tweak ever smaller performance gains out of their equipment. As a result, they invoke much pseudoscience in order to justify massive cost outlays for things like marble turntables and $100+ power cables.
- The green marker thing: It was rumored at one time that coloring in the edge of a CD could make the disc sound better; in response to skeptical comments that "bits is bits" (i.e. digital data sounds the same no matter what source it comes from), some audiophiles have offered the ad hoc idea that the green marker prevents stray laser reflections... or something. Though no longer as widely believed, the markers are still available and sold explicitly for that use.
- Treating CD surfaces: While it is true that a clean CD is a good thing, a cult developed around "treating" the playing surface of CDs with Armor-All (specifically) to improve the sound.
- Skin effect: In radio engineering, signals in the VHF, UHF, and microwave bands sometimes tend to congregate on the outside of a conductor, increasing the effective resistance of the conductor and causing signal leakages at sharp bends in the conductor, requiring careful laying of antenna and patch cables to avoid signal loss. While the effect is negligible to nonexistent at audio frequencies, some audiophiles nevertheless place great emphasis on avoiding skin effect. (It is of some concern in high-bandwidth video applications, but for the most part people don't obsess nearly as much over S-Video cables.)
- Hyperseparation of channels: Numerous audiophile systems avoid stereo power amplifiers, using so-called "monoblock" (i.e. single-channel) amplifiers to completely eliminate interchannel crosstalk. While fairly harmless in terms of function (though not necessarily to the audiophile's wallet), it is generally considered overkill by sound engineers.
- Obsessively shortening the signal chain: Given the opportunity, many audiophiles would prefer to avoid as many amplification stages as possible, to remove even the tiniest loss of signal definition. As a result, some audiophile systems eschew even a preamplifier, piping the output from the playback device directly to the power amp. Sound engineers find this to be especially hilarious in light of the vast number of op-amps in any given mixing board.
- Esoteric cabling: Manufacturers gladly pretty much make up pretend scientific-sounding reasons for why their incredibly expensive interconnects and speaker wires are supposed to sound better. There is also a sub-cult of people who think their speaker wires sound better if they are supported on magical insulators. At least one test found that audiophiles couldn't tell the difference between coat hangers and fancy Monster brand cables used as speaker wires. Super-technobabbleific woo is particularly hilarious when applied to digital cables (S/PDIF, HDMI, etc); again, bits are bits and bit loss is often more than readily apparent.
- The magical clock: A particularly ingenious scam artist once sold a "specially modified" digital alarm clock for $400 each. Plugging it into the mains outlet was supposed to effect amazing improvements in the quality of sound. He also sold a number of other strange things, like little bits of foil to stick to things. The subjectivist audio press were duly impressed.
- Another outfit, that also makes a magic clock, also offers to teleport better audio and video reproduction thousands of miles - over your phone. Genius. 
- Brand power: The first-generation PlayStation is supposedly actually a hifi-quality CD player, because, hell, it's a Sony. When skeptical people opened one up and found fairly cheap parts not particularly similar to ones found in Sony's top-of-the-line equipment, the audiophiles retorted, "it's the connectors, duh". People remain somewhat unconvinced.
While many of the claims made by audiophiles are rightly derided as pseudoscience or woo, there is a track record of so-called "golden ears" being right about some things that "engineers" failed to get right. A few examples:
- Early solid state amplifiers (and even some later ones), while measuring far better than their tube (valve) predecessors, sounded worse. It turned out that incorrect implementations of negative feedback to reduce harmonic distortion (THD) was causing intermodulation distortion (IM) artifacts. Of course, once the engineers realized this, they solved the problem. Early transistor amps were also producing large amounts of "crossover distortion", a kind of distortion that is caused by transistors switching on and off (i.e., each of the pair of transistors only conduct one half of the waveform and then have their outputs combined). This has been greatly ameliorated by "idling" the transistors during the off period of their duty cycle.
- Early CD players were accused, in spite of their being marketed as "perfect sound forever", of sounding nasty compared to analog equipment. The "brick wall" filters at around 20-22 kHz used to keep digital artifacts out of the reproduction chain were creating unpleasant phase effects in the audible region. Also, some of the early players only had 14-bit D/A converters. Better ways of eliminating these artifacts (oversampling converters) eliminated this problem. Many early CDs were also mastered poorly, since engineers were not used to the nature and requirements of the format, and this led to them sounding harsh. With decades of experience in digital recording and mastering, this is no longer an issue.
- Most contemporary CDs are mastered very aggressively, with the purpose of their being listened to in very noisy environments (car radio, personal audio devices used in urban environments etc). The trend has been named "The Loudness Wars" and makes for CDs that sound too loud and too compressed in a normal home audio situation. This kind of mastering could not and (and still cannot) be done on vinyl records, as they would then present excessive tracking demands to the pickup cartridge and cause distortion or record skipping. Consequently, an LP version of the same music can sound much more naturally mixed and mastered than its CD equivalent, which has been artificially "optimized" in the studio for use in noisy environments. This is not an inherent equipment fault, but the result of deliberate marketing decisions made by music sellers.
- Some high end headphones have a very high power draw (low impedance), meaning that when hooked to a standard PC or CD/MP3 player jack, they cannot draw enough power, which decreases sound quality. A small, fairly simple amplifier can greatly increase quality under these circumstances. These units are small enough to fit inside a mint tin, very cheap, and so simple that anyone who can read schematics and solder can make one themselves.
- Over very long lengths, cables can affect audio quality if the gauge of the wire is insufficient. Thin speaker cable at lengths of 50 feet (15 meters for those living in the 21st century) or more will show a noticeable drop in quality. Similarly digital cable over very long lengths will show noticeable signal degradation. However these are issues that could only really affect very large venues such as stadiums and concert halls where industrial grade (and relatively cheap) cable is used. This is the grain of truth that is heavily manipulated to sell very expensive cable to the monetarily overburdened.
- Classical audiophiles now consider 24-bit 96kHz (usually as a FLAC) the acceptable minimum, even though no DBTs have proven that 24/96 can be differentiated from CD (16/44.1). Linn typically offer 24-bit 192kHz.
- S/PDIF transmits clock signals through a special format called "biphase mark code" in order to support a wide variety of signal rates. If insufficient care is taken in designing or implementing the clock reconstruction process, this can result in "jitter" as the bits are sliced up wrong on the receiving end.
There are other examples, but, sadly, these truths tend to encourage careless audiophiles to set greater store than appropriate to every wacky claim put forward, both by manufacturers and "golden ear" reviewers.
Shit or get off the pot
Audio quackery is now in the sights of James Randi, who is putting his $1 million prize where his mouth is — "simply because if some of their claims were true, they would be paranormal." No takers so far, a few pikers.
There are people who do work with recorded music for a living, and if they don't get it right they lose their jobs. Studio recording relies on all-transistor analog stages and digitizing the signal in the highest quality possible (typically 32-bit float/96kHz; a lot of tools now do 48-bit fixed point, and 192kHz ADCs are pricey but readily available pro kit). Depending on the budget and style of the engineer it is uncommon to still record or mix to tape. In higher end studios most processing is done in the digital domain or "in the box." Studio monitor systems use solid state amplifiers and speakers with predictable characteristics so they can hear precisely what they are recording or mixing. Although low quality speakers known as "truth boxes" are used to make sure the mix will sound good in a low-fidelity playback environment, high quality speakers and systems are also used to make sure the recording is being mixed and mastered to the artist's or producer's liking. The only tubes in evidence are used in some microphone preamplifiers, vintage compressors, and of course in instrument amplifiers where they are prized for either their sweet sound or for their pleasing characteristics when overloaded.
A problem, which has been around for a long time, is the over-compression of recordings so they will always be "loud" on the radio. See loudness war.
In spite of this compromise, often provided for or demanded by the marketplace, one thing is notable in the professional sound recording and reproduction environment: professional sound engineers ignore virtually every bit of audio woo promoted by charlatans and worshiped by sufferers of nervosa audiophilia, since it is wasteful to spend the customers' money on hardware that does nothing.
Useful things you can do that will actually work
Spend the bucks on decent speakers. Everything else in the chain has pretty much reached the point of diminishing returns, so speakers are the lowest-fidelity component left, and spending more money can still result in noticeable gains. Note that, however, many expensive speakers are also rubbish, just to keep the decision interesting.
In particular for analog playback devices, avoid sharing a circuit with a large inductor like a fan, air conditioner or a particle accelerator. Use of a surge suppressor can clamp down on transient voltages and a 1:1 transformer can cut down on harmonics. Most amplifiers, though, will already use a transformer which naturally even out circuit harmonics. A dedicated ground might help in some situations as well. However, all these are meant to correct problems associated with distributed power and there's no "performance gain" by using them when no disruption is present.
- An example of what this article strives to debunk.
- Audiophoolery - Ethan Winer, eSkeptic
- Ethan Winer, Ethan Winer, AES Audio Myths Workshop video
- Roger Russell takes on myths about speaker wire.
- Athens Audio Club - a revealing video of uber audiophiles.
- Just incredible: a $3549 3ft. USB cable.img
- ↑ Write up of double-blind test for audiophile equipment
- ↑ Article on audiophiles
- ↑ Audiophiles and the Limitations of human Hearing
- ↑ Positive Feedback's Review Philosophy
- ↑ The Audio Critic issue 26 The 10 Biggest Lies in Audio
- ↑ http://www.cs.huji.ac.il/~springer/
- ↑ The Audio Critic issue 26 The 10 Biggest Lies in Audio
- ↑ Axiom Audio
- ↑ The basics of bi-amplification.
- ↑ wp:Nyquist-Shannon sampling theorem
- ↑ For example, a very rough approximation of a sawtooth wave with 8-bit sampling might look in a hex editor something like: 00 20 40 60 80 A0 C0 F0 00 20 40 60...
- ↑ This is somewhat unnecessary to the point of this discussion, but for examples, check the Wikipedia articles on wp:Frequency shift keying, wp:Phase shift keying, and wp:Quadrature amplitude modulation for three examples of how it's done and wp:Digital Radio Mondiale for a widely-known example of a real-world application.
- ↑ Or losslessly compressed like FLAC or Apple Lossless; lossy compression such as MP3 is a different story and really isn't a signal chain issue.
- ↑ If you want to be really technical about it, OSI levels 1 and 2.
- ↑ A quick case study: Audio journalist Malcolm Steward wrote an article in August, 2010 describing the alleged effect of swapping out an SATA cable in his NAS box on the sound quality of a very expensive digital media player. After getting crucified by legions of mainstream computer geeks who knew what he was claiming was beyond impossible, Steward threw a special-pleading temper tantrum for the ages and shut down comments on the relevant blog posts.
- ↑ Nyquist is a loosely LISP-based programming language designed specifically for processing digital sound files; the original web page can be found here, and the most commonly deployed version is part of the Audacity multitrack audio editor.
- ↑ AC power filter: I can't see a difference, but I know it's there!
- ↑ $1,800 for a power cord? I'll take 2!
- ↑ Do coat hangers sound as good as Monster cables? Yes. Yes they do.
- ↑ Machina Dynamica's Clever Little Clock
- ↑ Machina Dynamica's Teleportation Tweak
- ↑ PlayStation 1: The audiophile's dream?
- ↑ The ear can tolerate fairly large amounts of harmonic distortion because the distortion products are natural overtones (harmonics) of the signal. In contrast even small amounts of intermodulation distortion are objectionable, because the distortion products have no harmonic relation to the signal.
- ↑ The CD Story (Kees A. Schouhamer Immink, one of the engineers who created the format); Meridian 208 CD player/preamplifier (Stereophile, December 1990) - Phillips had initially specified CD as 14-bit and duly developed a 14-bit DAC, the TDA1540 - then Sony came in and said "no, we want 16-bit." Phillips went "shit" and their initial CD players had to make do with the 14-bit DAC until they released the 16-bit TDA1541 in 1985.
- ↑ Misleading Wire Demonstration
- ↑ http://www.randi.org/jr/2007-09/092807reply.html#i4
- ↑ 27.0 27.1 http://www.randi.org/jr/121004science.html#11
- ↑ This message not placed by the RW contributor who makes and sells speakers for a living.