There's one born every minute; the sound quality of different storage medium

Subject: General Tech | January 9, 2015 - 01:22 PM |
Tagged: monster, idiots, audiophile

Believe it or not there is a review out on the interwebs claiming that "'bit-identical' computer audio may well be just as inexplicably inconsistent as analogue."  In other words some hard drives and SSDs will produce better quality audio than others using the exact same audio file.  Two different QNAP NAS devices apparently produced differing audio signals which the writer claims to be able to discern.  Not only that but apparently different HDDs or SSDs inside the NAS also has an effect on the audio flavinoids and topology.  If that is not enough for you then keep reading the link from The Register as they also propose the theory that different types of RAID will change the cromulence of the audio signal as well and while they stop short of describing the audio cables which were used they did stoop so low as to use Belkin CAT6 instead of a product from Monster.  If you believe this and own a mains conditioner for your audio you should definitely let The Register know you are interested in their proposed AudioNAS kickstarter.

View Full Size

"Is it April already? I really cannot tell from this post, which poses the question: "Is it really possible that the sound quality of bit-identical audio files is influenced by their storage medium before being delivered to the hi-fi system's DAC?"

Here is some more Tech News from around the web:

Tech Talk

Source: The Register

January 9, 2015 | 01:48 PM - Posted by Allyn Malventano

One of our readers pointed me to this article while we were out at CES, and the mere thought that a particular HDD or SSD can have 'more engaging bass performance' just boggles my mind. I could go on and on but would just be repeating what comes out every other time someone has argued the audio quality differed over a digital medium, but in this case the writer / reviewer is trying to infer that audio quality is differing over a digital medium with heavy error detection and correction, and where a single bit flip constitutes a complete failure of the devices primary function.

Oh, this hard drive can boot a multi-gigabyte OS repeatedly and without a single bit error (as that would crash the system), but when it plays that .wav audio file, this one model "gave a far from subtle shift in tone and soundstaging". Yeah, right.

From the article: "Maybe we can solicit logical explanations from engineers who understand the low-level mechanics and operation of computer file and storage technologies, and can suggest specific avenues to explore.". I review these things for a living, and I'll be the first to say that they have absolutely zero understanding of how these things really work at the digital level, and they are giving other storage reviewers a bad name by even undertaking such a test and write up. DAC's have had buffering and their own reference clock for ages, so there really is nothing on the digital streaming side that will impact the audio itself. Those guys need to stick to talking about the analog side of things - that's the only place you get an actual difference.

Side question - if this guy claims to hear these supposed (impossible) differences, does he not undermine every other listening test he has done, even of things that *could* sound different? If I had previously followed this guy's work on the audio side, I would immediately disavow anything he said after reading the above linked piece. This drivel curses his audiophile writer peers, as such work makes their entire community look like snake oil selling quacks.

January 9, 2015 | 05:36 PM - Posted by Anonymous (not verified)

The tech industry is the new snake oil, expect there to be plenty of salesmen/saleswomen pitching any, and every, coal oil and menthol derived elixirs to fool the average bumpkin. There are also loads of tablet devices that are really no more useful that a standard paperweight, among the countless devices that come out a CES, and other tradeshows. With the watchdog agencies not doing their jobs, and properly regulating the industry, or enforcing the minimum amount of proper documentation, and data sheets, for many of the devices that are for sale, expect thing like this to continue to happen. Who is there with the pot of boiling TAR, and sacks of feathers, to give these technology grifters their just deserts!

January 10, 2015 | 05:33 AM - Posted by lantian (not verified)

yes he completely undermined his own work, made himself look real stupid and at least for me say never take anything that he has ever said is untrustworthy now

January 10, 2015 | 05:37 AM - Posted by lantian (not verified)

also forgot to mention, but most people i know including my friend aren't in to tech as i am and have no idea of difference between digital and analog, some of them even fallen for the expensive 45euro hdmi is better than a 5euro one, people like this are doing a huge disservice for those that have no knowledge about these thing since they are the ones that will believe this bs

January 9, 2015 | 01:56 PM - Posted by Shambles (not verified)

I'm not gonna lie, this article forced a chuckle out of me.

January 9, 2015 | 02:09 PM - Posted by Anonymous (not verified)

when did we start believing what we see in the web?

January 9, 2015 | 02:10 PM - Posted by Anonymous (not verified)

im a german model if you know any hot woman.

January 9, 2015 | 02:18 PM - Posted by Anonymous (not verified)

I would be surprised if an audiophile quality NAS didn't exist. If not this article is merely giving a blessing to such a concept. I for one eagerly await the silver SATA connectors with marked arrows and articles describing how to properly bake harddrives or SSDs to get the best sound quality from them.

January 9, 2015 | 02:24 PM - Posted by Master Chen (not verified)

That was some very nice marijuana in there silicon valley, indeed.

January 9, 2015 | 02:36 PM - Posted by Anonymous (not verified)

I am lost for words. TheRegister makes a mockery of itself once again.

January 9, 2015 | 02:41 PM - Posted by Allyn Malventano

Actually, TheRegister piece appears to be (appropriately) mocking the original article.

January 9, 2015 | 02:46 PM - Posted by edwinjamesmiller36

"This initial trial was not intended to be an exhaustive study into all the factors that can affect the sound quality of network and computer audio, only to confirm or deny the suspicion that digital bitstream coming from hard disks are not all equal. Which has to be somewhat surprising, to say the least."

If you wished to compare the digital bitstreams between different hard disks, why not actually compare the bitsteams bit-by-bit? Why introduce a human ear listening to a sound further downstream instead of comparing the actual 0's and 1's of the digital information? It makes one wonder if they even know what a digital bitsteam actually is.

Maybe because comparing actual bits to other actual bits would devalue the reviewer's subjective interpretation? Perhaps even show him to be a fraud?

January 9, 2015 | 03:01 PM - Posted by fade2blac

"gave a far from subtle shift in tone and soundstaging"

This hints at a fundamental misunderstanding of digital vs analog encoding of signals by the author. Digital encoding has very specific limitations on accuracy when converting from the analog world in which we live to a digital representation. Some information is invariably lost in translation be it an analog or digital recording. With digital, one can devise an encoding scheme that reduces the losses to levels that are expected to be imperceptible to a human observer. The tradeoff is that once in it's digital form, there is no variability...the digital representation is such that it can be digitally copied and reproduced without any further loss of information/accuracy.

In my experience, when digital audio streams suffer errors the result is most definitely FAR from in obvious cracks, pops, clicks, and garbling of sounds similar to quantization noise. This would be the case if a storage drive is simply failing to retrieve data in a timely manner or is somehow passing along uncorrectable/undetectable errors. This would be considered either a misapplication or fundamentally broken device.

As Allyn stated, whatever the source of the digital bits, they are the same bits once they reach the DAC. At this point any variability in DAC output would not be due to the storage medium. The only value I could see in a audiophile NAS would be one that enforces some sort of file integrity checksums to ensure your data doesn't suffer from corruption or bit rot (a term for the gradual decay of a storage medium over time). This is something that can just as easily be added via software on top of regular off the shelf hardware.

January 9, 2015 | 03:15 PM - Posted by lordbinky (not verified)

What this guy could possibly have dodgy mains power (and no a line conditioner is not a fix-all) so it is possible the difference in power draw between an SSD and HDD is affecting the audio out output due to varying supply voltage, in which case there are bigger issues going on here.

There is also the more likely possibility that he simply has ground loops in the connections between the various components of his setup (which a line conditioner does nothing for). This will let him pick up all kinds of noisy critters where the operation of any nearby machinery and everything off the source of the grounded Ethernet cable (including all the electronics connected on that wiring which I doubt are on line conditioners as well) are inducing some strays on this poorly grounded HI-FI audio setup. End result is not that a NAS is changing it, but that any load difference on his electrical system is generating a difference. If he plugs a hair dryer onto the same outlet as his Ethernet router he will likely get the same phantom audio changes.

January 9, 2015 | 03:34 PM - Posted by Allyn Malventano

I see where you're coming from there, but the author had these NAS's in another room (i.e. likely on another circuit), and were connected to the DAC via WiFi. They unintentionally added even more isolation and therefore further eliminated any possible way for these different devices to cause issues.

Further, from the photo in the piece, it appears they were all connected and powered up simultaneously, which makes sense, as then the guy could easily play from various sources that were all on the network simultaneously.

January 9, 2015 | 04:12 PM - Posted by lordbinky (not verified)

Damn. I was just trying to make a possible scenario where the final output was somehow still different. Their theory is hogwash, but I was trying to give benefit of the doubt that there was something they were hearing even if they misidentified the cause.

Oh well, if their Hi-Fi audio reputations tank I suppose they could apply their hearing powers to ghost hunting.

January 9, 2015 | 04:23 PM - Posted by Allyn Malventano

Yup. You're totally right, and I was doing the same thing in my mind while reading it. Unless these guys' ears can percieve 2.4 / 5 GHz WiFi signals, there's really no way those storage mediums could be altering the music data itself.

January 23, 2015 | 07:36 AM - Posted by Anonymous (not verified)

Sorry to say this but you are completely clueless and your mind is very limited :)

January 9, 2015 | 04:23 PM - Posted by Anonymous (not verified)

This is what I first thought of (I am not going to bother to read the article). I have had machines where hard drive access caused noticable audio interference. This should be increasingly rare on more recent systems due to better isolation of the analog circuitry. This also should not be a factor when playing over the network since it should cause equal network traffic. If there is interference from other hardware (network, local storage, etc) then it could hit at different times in the audio, but I doubt this would be noticable. It obviously has no relation to the integrity of the actual audio bitstream.

January 9, 2015 | 05:36 PM - Posted by Jeremy Hellstrom

but you will miss out on gems like this ... "the Hitachi sounded very ethereal, almost out of phase, and rated it lowest; the Seagate was sharper with a more thumpy bass, slightly brighter with a slight tendency to sibilance."

January 13, 2015 | 08:43 AM - Posted by collie

Best quote of the whole thing "QNAP2 had the blacker silences and deeper spaces between notes"

January 10, 2015 | 12:27 AM - Posted by edwinjamesmiller36

Comes down to how many angels can dance on the read head of a HDD.

January 9, 2015 | 08:07 PM - Posted by D S (not verified)

I have not read the referred article, but I assume I can have a say or two

1. I remember some companies promoting their HDDs as "suitable" for audio - I will come to this later

2. Regarding bit critical corrections, it should be clear that a miss on a bit does not necessarily mean a computer crash. You can have also malfunctioning, especially if the bit loss happens in a point where you have e.g. a number; having 65536 is not that different than having 65535 and it would not result in a crash.

Combining 1 and 2, I could say that *maybe* manufacturers employ different operations when arhiving or reading different file types. For example, they could propably disable some short of error correction when archiving music (I have not any idea of what happens inside a hdd/ssd, I am hust saying :) ) which would result in some loss of precision. But my main concern is WHY would someone implement a function like that in the first place (like probable reduction of energy consumption?).

best regards

January 9, 2015 | 09:39 PM - Posted by Anonymous (not verified)

They wouldn't implement something like that. Any portion of data on storage could potentially be data or instructions which are sensitive to single bit errors. Most data is actually very sensitive to single bit errors because it is compressed. A single bit error can cause something similar to a frame shift error, where it changes the way the rest of the data in the stream is interpreted. You probably would not notice a slight change in color of a single sub-pixel in an image, but if the error is in the compressed data it will usually cause the whole rest of the file to be decompressed incorrectly (depending on file format).

I had some bad dram on a machine; single bit error, only triggered when surrounding bits were set. This still caused a large number of images written before I discovered it to ended up corrupted. Usually the bottom portion of the image ended up being garbage because the single bit error caused any following data to be decompressed incorrectly. This is an issue with how compression algorithms work. This is a good reason to run an exhaustive memory test (memtest86) when building a machine. I am actually considering going with server grade hardware to get ECC support. Unstable systems really annoy me. The way things are going, I may even run a gpu memory tester for a new system build. I have a feeling that a lot of graphics card issues are bad memory since we are putting a couple GB of high speed memory on them now.

January 10, 2015 | 10:35 AM - Posted by D S (not verified)

Concerning compressed data, as you pointed out a single error affects a large portion of the file streamed but it is safe to say, that aside from sync problems, minor errors in a uncompressed stream would go otherwise unrecognisable, exposing the quality of the medium it self. I also remember some lessons about different levels of error correction, so in the long run (read: statistically) different sounding digital mediums could be for real.

Also, I see your comment about errors in modern cards. I am on an oldy (but ..goldy :) ) Ati 5850, which is quite stable. Given of how hard is to have quality parts under modern manufacturing processes of ~15nm, I would not be surprised if the errors you are seeing are of bad ram. Maybe it would be wise to differentiate instruction memory from data memory at some point (having instruction memory in better quality memory) :) Ok, on a practical way now, maybe it is better to stay off the latest and greatest, especially concerning shrinking but instead give each process some time to mature.

best regards

February 20, 2015 | 06:57 AM - Posted by Anonymous (not verified)

I think I can guess what the developers had in mind and it does NOT seem to be some people believe they can hear.

People into computers since the 80s will remember that the first audio cards for PCs sometimes would not give very good sound, because the analogue parts would pick up digital "noise" from the computers circuits. Computers are constantly changing voltages to make ones and zeros, and all those rapidly changing voltages would sometimes create instabilities, or even induce interference on the audio card which could be heard. It was rare, but not unknown.

What I think Sony might be doing here is selling a type of memory that has suppressed EMI/EFI emissions and so is less likely to be picked up by the analogue audio circuits?

Having said that, if the gear was well-designed in the first place, it should not be happening. What person willing to pay this much for memory would have a piece of $5 audio gear so poorly made that it could pick up interference from the small voltage changes inside a memory card?

As many have said, digital data is either a zero or a one, so data CAN NOT change the quality of sound other than by causing interference to the analogue circuits, or outputting data so corrupted that the error-correction has significant problems with it.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

This question is for testing whether you are a human visitor and to prevent automated spam submissions.