Crazy digital audio theories

Since the debate about memory playback, I’ve had this nagging recollection of a series of articles that really riled up the computer audio guys. Anyone remember/aware of these articles?

The above is only a subset of the absurdities from their original The Absolute Sound 4 part series - discussed here:!

I could not find a link directly to the TAS articles but I recall reading them, and then not renewing my TAS subscription.

Some of their tests were crazy - i.e. copying a media file multiple times through Windows Explorer and then claiming a difference in sound quality from generation to generation of copies. In the first article above, they encode and decode into wav (not real-time, just using DB Poweramp) and claim differences in sound.

They attribute some sound degradation to the inclusion of metadata! They argue that the level of compression in cover art affects sound!

Just thought I’d see if anyone else remembers or is aware of these…


Naim also claims that there’s differences between flac and .wav.

If you want to go down the rabbit hole of stupid claims, there’s a few threads on CA, including a listening test between multiple generations of identical files (yarly) and an explanation that a certain Grammy-nominated recording engineer only sends their files over the internet as uncompressed zips, because using a container prevents the degradation usually associated with such barbarity.


(and lest a newbie lands here and buys the claims: they’re unbelievably dumb, and, if they come from a company, their only use is as a bright, red flag flying over that company’s digital products).


Yeah, that was very funny, because the multi-generation thing required abandoning the integrity of the bits, since the copying happened offline — they could not rely on dubious theories of noise transmission or load on the CPU. Maybe that’s why it went away, it didn’t line up with the paranoia of the day.

I think there is an interesting sociological dimension to this. Used to be, we all thought bits are perfect. Then we discovered jitter, and we all bought reclocking devices. Then we mostly lost interest in jitter, maybe because DACs became better (although facts are not really a big influence here). And then we all focused on noise transmission over the wire, and we all bought network streamers (some still do jitter reduction, with a femto clock, even though the USB output is async). Then we began to lose interest in this, because DACs are better isolated. But now the bugaboo-du-jour is grounding!

If you’re still worried about computer noise (or jitter!), you’d better get on the bandwagon, that will soon be as obsolete as head alignment. Grounding, baby!

If I could predict what comes after grounding, I’d be rich…


The current idiot magnet is audiophile network switches, and for the truly brain-damaged, the closely related audiophile ethernet card.


Kooky Marenco.


Don’t forget that some also claim that rips made by the Melco cd ripper which cots around £600 sound better than accuarip using dbpoweramp or eac. Rips are not rips apparently. I get a bad rip won’t play and can skip etc but really one accurate rip sound better than another accurate rip on different hardware.

1 Like

Perhaps the processor is so slow in that unit that it struggles with real-time FLAC decoding?

Exactly. With their logic, copying over a Word document would introduce typos that would clearly be noticed in multi-generational copies. In effect, if their theories were true, computers would be unusable in their current state. Backups would always be corrupt. Text read from the web would be corrupted by the time it reached a browser.

I remember clearly readying one of the 4 part series and being absolutely shocked with how little they understood and how much it seemed like unfrozen caveman audiophile. That TAS would publish that.

Right, and it’s also the case that bad rips like that tend to be localized to a track or part of a track. There would be noticeable changes in the sound quality during a listening session.

If you can find the original 4 part series, you should check out the criteria they developed for ranking/rating the sound quality. IIRC it was really hilarious, like one grade was based on evocation of emotional response. I have an emotional response for you!

1 Like

I remember these TAS articles well. Never a TAS subscriber, but saw all the chatter on the squeezebox forums. This is when I realized that we were doomed (cue the movie, “Idiocracy”)

LOL. I thought you’d announced your retirement?

I just finished reading that whole article. That is some impressive scientific method.

At least he does unplug the switch in the last part and admits that sounds the best. So I’m going to buy two of them and unplug them both for double the sound improvement!


A cynic might suspect that the software/firmware on the devices being tested is so buggy that a change in metadata tickles bugs (like memory corruption) in the main digital sound path :wink: Knowing what I know about embedded software, that might might be as outlandish as it sounds. After all, some of you might remember the recent XMOS USB receiver kerfuffle…

No. That would cause stutter. For those who believe WAV and FLAC sound identical, it’s because after decompression they are bit identical. For those who believe WAV sounds better than FLAC, it’s because FLAC decompression process takes additional CPU power (hence increased noise / EMI / RFI), and additional power draw from power supply which also means increased power supply noise. There’s another industry pioneer who proposes memory access causes more noise than CPU does.

Note that we have (at least one) certified Roon server partner who underclocks the CPU for what I believe to be similar underlying reasoning.

We also have members here reporting GUI-less Roon server to sound better than Roon with GUI. For an official and rational explanation to “crazy” (no disrespect here, just to use the same wording as the thread subject) observations, check these out:

1 Like

Kidding. I know sometimes sarcasm doesn’t survive the written word…

Wouldn’t use of a separate endpoint via network eliminate all these concerns?

Anyway, this is far afield from the goofy conclusions reached in these articles. That a media file sounds different after it has been copied a few times from HDD to HDD. If that were the case, computers would not be usable, the web would not be usable, backups would not be usable.

1 Like

The way to peddle a crazy theory (or a device based on a crazy theory) is to sprinkle it with scientific-sounding talk, some that is correct but irrelevant, some that sounds plausible but is unproven, some that sounds plausible but is wrong, and the final conclusion that sounds crazy but gets its credibility from the others.

Like astrology: people used to just make claims, but when people objected on scientific grounds, they would say that it is not scientifically unreasonable that the gravitational influence from the big planets like Jupiter, and from the stars in the galaxy which are arranged in a flat plane (the Milky Way), that this influence at the moment of birth would affect the developing nerves in the brain and thus influence your evolving personality. Except, if we do the math we realize that the gravitational influence from the midwife or doctor is much greater than that of Jupiter or the Milky Way. But “gravitational influence on the nervous system” sounds better than “just trust the cards and Madame Carlotta”.


I have nothing to add to this thread but I just wanted to say that I am enjoying it immensely. Brilliant posts chaps/chappettes.

Warning: adult language and 1990s sensibilities.

I agree, there is a LOT of craziness in the audiophile community, but just because one doesn’t understand why something may occur doesn’t mean it isn’t occurring. On the surface audio electronics is fairly simple but if you dive in you will discover it is far too complex for most to wrap their head around. Simple example… most will talk about about electricity in terms of current flow which they incorrectly believe to be the flow of electrons through wires like water through a pipe.

If old enough, you will recall in the early 1980s when the crazy theory that cables could make a difference was dismissed by most. “It is just wire. how can it possibly matter?” Concerning switches, it makes sense to me that a cheap switch injecting RF into a system might affect the sound.

I’m not saying everything mentioned is valid, but some of the things dismissed above can easily be heard if you have a system capable of resolving the differences. I would venture that many do not have such a system and so gleefully poke fun at things they can’t hear and don’t understand. I’ll take that a step further. I bet most who are gleefully poking fun have not tried any of the things they are calling crazy. On the bright side, they do retain a lot of their money if they don’t buy in.

1 Like

I can agree that well engineered cable and the like make a difference but that’s as far as I will go, in fear of my wallet.
My rational is living a life listening to music and I have never had it so good. Also the way ordinary people’s jaws hit the floor as I play them music on my Meridian DSP SEs.
This is with basic switches and decent Cat5e and 6 cables.
Another thing is feedback from people who do know how to listen to music, especially working musicians. They do have a good ear and I value their input a lot.

To me, it’s not worth the chase at this time as I have a kitchen to pay for :joy:

1 Like

One has to remember that digital transmissions are engineered to reject noise (RF or other) that may have been injected into the cable.

It is a certainty that no perfect square waves are received (or even generated) at either end. Then there may be additional noise coming from the environment that will mangle the perfect square wave some more.

This is why there are operational margins in terms of amplitude and time (and shape). As long as the received signal is more or less at the high (or low) value for about the right amount of time, it gives a 1 (or zero).

Overshoots, undershoots, ringing and delays… Still works!
Source: EE|Times - Jitter, Noise, and Signal Integrity at High-Speed: A Tutorial–Part I - By Dr. Mike Peng Li, 12.10.07 -

Once the signal has been evaluated as a 1 or a 0, that’s the end of noise on that segment of the transmission.

Noise is rejected at every digital hop, external or internal, and is not carried over. This is valid not only for USB and Ethernet, but also for every transmissions inside digital equipment, inside and between chips.

If noise gets bad enough to make evaluating a 0 or a 1, because a piece of equipment is broken or very badly built, then error detection and correction (usually with a retransmit) kicks in. You do not have to worry about timing, because those transmissions are really quicker than anything that could worry digital music data decoding, and because buffers are in use in most cases anyway.

If noise becomes so bad that there are too many errors detected and too many retransmissions that end in error too, then all breaks down. You go from no errors, to some detected but corrected errors to total failures, without any degraded in-between state.

To illustrate, remember how analogue TV would go from fine to snowy degraded pictures, and how digital TV goes from fine to nothing, no snow (transitional visible artefacts are created temporarily as buffers empty with partial data). It’s the same for digital music, it will go from playing to not playing, without any state of degradation in-between.

Also, we are talking about digital transmissions of data, not music. Those transmissions happen at fixed frequencies. So invoking the fact that HF audio is transmitted better inside USB cable so and so at $1500 because it handles HF aufio better is pretty hard to read.

To make it short, noise in digital transmissions is irrelevant. Totally. Not by chance, but by design. It’s one of the great advantages of having moved from analogue to digital systems. The only place where noise is an issue starts in the analogue half of a DAC and beyond. Never, ever, before.

That means that esoteric digital cables can have nothing on properly built cables sold at affordable prices (which can be bought at any local supermarkets for low two digits values), playing from memory or underclocked or gui-less or optimised systems makes no sense, linear power supplies are a waste of energy by lack of efficiency. Cheap or expensive, there is no “audiophile” digital equipment that will work better, be it a network switch, a computer, a hard drive, SSD or a cable. If it works already, it cannot be made to work better.

What we have here is a lack of understanding of the engineered nature of digital transmissions, applying concepts relevant to the analogue world where they does not apply.

Also remember that transmitting, decoding and playing digital audio is a low-level effort for any computing equipment. If a cheap Chinese smart watch can do it, do not worry about the capacities of any equipment capable of running Roon, or being used around Roon.

Save your money, do not buy into the audiophile digital stuff, buy more music, or book a week of vacation on one of those beautiful islands if that’s more your thing!

EDIT: if you want to put more that $10 into a USB or ethernet cable because it looks better built or because it looks better next to your nice equipment, then, by all means, go right ahead. We are all entitled to non-practical features. I like nice looking expensive watches, when a $2 piece from the supermarket would give me the exact same time with the same accuracy.


But, heh, if there was any ground to USB cables making music better, I would not use them for music, but for the connection to the USB hard drive storing the Excel file wher I balance my accounts. The bottom line would sound much better, I’m sure!

(I have a similar analogy about my porn collection and how things could be improved there too, but let’s not go into that…)