The Digital Fallacy

[meep]

Hi All

I've been following the build an audio PC thread with some fascination and not a small degree of admiration. However, there is a facet of the discussion there that troubles me and I thought I'd open a separate thread here so as not to clutter that up too much with my ramblings.

I'll preface what I say by noting that I'm not an electrical engineer. I have, however, worked as a software engineer in the past and have built more home theater PCs than I care to carry from my garage where the rest in pieces having been technically superseded every five minutes. I will also note that what I say here, I say from a position of genuine interest and would love to be pointed towards the science that proves me wrong.

My main point is this: digital data is digital data and no amount of investment in digital cables will affect the perceived sonic capabilities of said data.

Now, before you write me off as a heathen and nonbeliever, I will say this: I absolutely buy the fact that interconnects in the analog domain can have an impact on the signal passing thorough them. All the factors from materials, thickness, length, shielding and external forces can and do combine to have an affect on the signal.

Furthermore, I will absolutely agree that build quality and external factors can affect any devices involved in digital-analog conversion processes and these as well the perceived quality of sound can be affected in both subtle and obvious ways. Furthermore, the PC environment is inherently noisy from an electrical standpoint and this would of course have an impact on internal DACs and any analogue processes being carried out in its environs.

However, I draw the line at consideration of digital audio cables or, more specifically, data connectivity such as SATA cables, having any bearing whatsoever on audio quality.

The purpose of a SATA cable (for example), is to transfer digital data from a storage device (such as a hard drive) to computer memory where it may be accessed by software applications. That data comprises sequences of binary signals (1s and 0s). The storage subsystems involved (hard drive - cable - motherboard chips & buses & RAM) are content agnostic. The data remains a string of 1s and 0s regardless of whether it's a sound file, a word document, an image file or anything else.

Saying that the storage system has a remarkable affect on the perceived quality of the transferred data is incomprehensible to me.

For that to happen, something in the storage or transmission chain would need to alter the data - swap 1s for 0s. Even if that were to happen, the prospect of it happening in such a way as repeatedly affect something like the perceived quality of a resultant audio file is baffling.

For that to happen, the factor affecting the signal would need to target only the segments of that data that relate to specific frequencies (in the case of sound). since the storage system knows nothing of the underlying data function and simply shuffles bits from one place to another, how can something like a cable affect only selected groups of bits in a seemingly targetted fashion?

Digital data can be corrupted. But it doesn't happen in a subtle way. Think what happens if you put a scratched CD or DVD in a player. You get digital break-up, artefacting, nonsense. Does data on a damaged optical medium change in an almost imperceptible way? No, it breaks up completely to the point where it's no longer audio of visual data but is instead digital junk.

Leaving aside audio for a moment, think of it this way. If a data interconnect had a subtle and targeted affect on the data it transmits, what would happen to image files? Opening the file in an image editing application might reveal changes in hue or contrast of the image? I think not. We don't find professional photographers, graphic designers or video editors agonising over the best cables to use and the affect they might have on the colour saturation or other characteristics of their work product!

And this is easy to prove.

Unlike in the analog domain where perception is mostly subjective, in digital, we can measure.

A data file stored on a hard drive or other medium can be checksummed. In simple terms, this is a way of essentially adding up all the bits in a file and using that total to later check if all the bits in a copy of the file add up the same way. If they do, the copies are identical. If they don't, something has gone wrong and the file is corrupt. Digital is fundamentally a binary system, the copy of the data transferred from the hard drive to memory is either an identical copy or it isn't. It would be trivial to write a program to checksum a stored file and do the same for a copy transferred over a particular cable to another storage location (drive or memory).

Now of course, there are some things that can and will affect the resultant playback of the file if it's audio. Files can become corrupted in storage (bit rot) or interference could perceivably affect them. As noted, however, this will manifest as random corruption and result in pops & clicks (at best). And of course, there's no telling what any software might do to the data once in memory or what might happen to it when passed through the DAC process on the way to amplification.

But that an internal data cable, storage device or type of memory chip could affect a set of data comprising an audio file in such as was as to be specifically and repeatedly targeted in such a way as to consistently affect things like specify frequencies or other more qualitative measurements is, to me, beyond credibility.

Though I am open to having that opinion debated and even changed!

I know it's an ongoing debate and if it's been discussed here before, or proven otherwise elsewhere, do point me in the right direction!

Peter

[Sligolad]

Alright I am not an electrical engineer but i will stick one response here and hopefully i might learn something!!

Meep, do you believe that 1s and 0s have any "qualities" other than being routinely rolled out as simply 1s and 0s because from what I have read they do and there are specifications on those qualities, so much so that when you break down those qualities they are analogue in nature?

Its time this simplistic 1s and 0s argument is finally exploded and I hope in the coming year or two some good research will finally support the growing real world examples of 1s and 0s sounding different through different signal paths.

Pearse.

[Fran]

Here's a good bit of technobabble that I think many ears have accepted as true.. from the jplay lads:

In music, timing is everything. And in digital music reproduction doubly so: while producing bit-perfect output is easy, producing it at exact time required by digital formats (e.g. 32 bits every 22 microseconds for CD) is not. Why? Because while your PC may be really fast, it’s also doing hundreds if not thousands other things at the same time it plays music. With so many things going on, do you trust it will always ‘hit the beat’ at just the right time? Programming optimizations in JPLAY are designed to minimize both software & hardware interruptions in order to make it ‘easier’ for PC to ‘keep the rhythm’.

[meep]

Alright I am not an electrical engineer but i will stick one response here and hopefully i might learn something!!

Meep, do you believe that 1s and 0s have any "qualities" other than being routinely rolled out as simply 1s and 0s because from what I have read they do and there are specifications on those qualities, so much so that when you break down those qualities they are analogue in nature?

Its time this simplistic 1s and 0s argument is finally exploded and I hope in the coming year or two some good research will finally support the growing real world examples of 1s and 0s sounding different through different signal paths.

Pearse.

Hi Pearse

I don't believe the bits have any inherent qualities. It's a binary state. As noted, it could be changed (flipped) but the effect of this is to corrupt the file, not changing any inherent qualities of it.

Think of a simple text file. If the data does not travel from storage to memory exactly as intended, the file will become corrupted at best, letters will be rendered to screen incorrectly etc. does the quality of the data stream get affected in some way such that the letters are spaced differently or some such?

While the underlying electrical signal my vary within tolerances, it still arrives at its destination as a bit. So, if a byte on disc looks like 00110101, it should arrive in memory as 00110101. From there, the software that initiated the read will look at the byte and translate it into whatever it needs to be for output (character on a screen, sample in an audio file, pixel in an image).

The software won't render the dot until it has the full amount. Of course, the rendering device may affect the output, but through no fault in the data. For example, the way a monitor is calibrated will affect the perceived look of an image displayed in it. The underlying data is unchanged.

Peter

[meep]

Here's a good bit of technobabble that I think many ears have accepted as true.. from the jplay lads:

In music, timing is everything. And in digital music reproduction doubly so: while producing bit-perfect output is easy, producing it at exact time required by digital formats (e.g. 32 bits every 22 microseconds for CD) is not. Why? Because while your PC may be really fast, it’s also doing hundreds if not thousands other things at the same time it plays music. With so many things going on, do you trust it will always ‘hit the beat’ at just the right time? Programming optimizations in JPLAY are designed to minimize both software & hardware interruptions in order to make it ‘easier’ for PC to ‘keep the rhythm’.

Hi fran

I think what they are talking about there is jitter. I understood this quite well many moons ago as I suffered from I in my early htpc builds. As I recall, it came down to the timing of signals between different components in the the path and would manifest as microstutter in the worst cases.

And in any case,the above validates what I think I'm thinking; the data is the data, it's what the software engineers do to it (or don't do as the case may be) that will have the qualities affects. I would certainly be focusing my attention on software and hardware rather than cabling in the digital domain.

Peter

[Fran]

Meep - I'd believe you 100% except that I have heard a definite difference with USB cables of different types and the direct adaptors.

I too would like to hear an explanation.

Fran

[Rocker]

Peter,

Some time ago I would have agreed with you 100%. But things changed and I saw that digital signals can be affected by the cable that transmits them from A to B. Firstly, considering digital to be merely ONEs and ZEROs is too simplistic a way to view a digital stream. I am not an electricial engineer but I understand that a digital signal is in essence a square wave with a sharp 'rise' and a 'fall'. The transmission cable can cause a distortion in the rise and fall parts of the signal IMHO. I use a Squeezebox to provide a digital signal to the digital input of my CD player. A good friend gave me one of his unused Stereovox XV11 cables which elevated the sound quality by a considerable margin when I replaced the original digital cable with the XV11. You can imagine my surprise at discovering the next day that the sound quality had changed and was overblown in the bass. This problem was solved by siting the Squeezebox on three Cereballs. The Cereballs restored the sound balance, the XV11 showed that the digital stream from the Squeezebox was considerably better than I had thought it was. So much for ones and zeros!

In my experience, one must treat digital signals with the same care and attention as one would with analogue.

[Sligolad]

Hi Peter,

Its those "underlying electrical signals varying within tolerances" which concern me and again the analogue aspect I am referring to in digital.
I am not concerned with the bits being right or wrong, I am more concerned with the ones that barely make it within tolerance but ultimately get clean bill of health as say a 1 when it is sent as a 1 and we never know the difference.
Has there been any studies carried out in this area where logic response times get affected by those bits which are scraping the edge of tolerance.

I try to approach this problem from a very simple manner looking for likely flaws in the thinking and this only because I have heard too many real world examples of cables in the digital domain making a difference.

Surely as a software engineer you have heard software audio players impact digital audio streams yet no measureable difference can be found in the digital streams?

Pearse.

[meep]

I am now intrigued.

while it didn't make sense to me that digital could be anything other than digital, subjective evidence here and elsewhere seems to indicate that cable does make a difference.

I came across this on wikipedia;

The receiver does not control the data rate, so it must avoid bit slip by synchronising its conversion with the source clock. This means that S/PDIF cannot fully decouple the final signal from influence by the analogue characteristics of the source or the interconnect, even though the digital audio data can normally be transmitted without loss. The source clock may carry inherent jitter or wander, and noise or distortion introduced in the data cable may further influence the process of clock recovery.If the DAC does not have a stable clock reference then noise will be introduced into the resulting analogue signal. However, receivers can implement various strategies which limit this influence.

Seems that cable characteristics can affect signal after all by introducing or exacerbating jitter. Could it be that 'better' cables allow better fidelity to the original signal? Possible.

However, this doesn't convince me that something regarding SATA cables as they ferry not audio-fomat data (SPDIF protocol) but raw data.

More investigations required.....

Peter

[item_audio]

There's a lot to be said for not ignoring the time domain: the time it takes for a mechanically-carried signal to rise from zero to one and fall again; bit-interval timing; interaction between clocks; SRC operation . . . at every level, assessing the dynamic behaviour of a realtime digital system is very much more complex than the literally asynchronous business of storage.

We're magnetically drawn to think about music files as if they're no different to image files - and of course they are! But it's not about the files. It's not about whether the data is pulled through an interface from static point A to static point B and lands in the same shape: it's about how the interface behaves during transit.

And very much about how the 'analog' basis of all those mechanisms (they don't waft through some platonic ether) impacts on the rest of the system: a USB cable ducts a significant chunk of the computer's ground-plane noise originating from massively noisy 1kW switch-mode supplies, directly into the DAC's motherboard. Most DACs are literally part-powered by the computer. The same noise is injected into (frequently) shared mains, so it's potentially all over the amplifier, too.

The distinction between 'digital' and 'digital-processing' is important to make: digital is not an antonym of analog.

We've grown so accustomed to the shorthand 'digital components' that we don't often stop to think that they're all still - highly miniaturised - analog components interpreting voltage signals and processing them digitally. And it's that 'analogness' that a realtime-operational DAC + amplifier is sensitive to, whereas printers aren't. Similarly, though for different reasons, a toaster works fine next to a microwave, but a radio doesn't!