How Hi-Fi Magazines Write about Cables, Part 17: Hard Disks

It’s been nearly a year since my last article in this series (which I find surprising; the time has gone by quickly). A reader contacted me last week with a delightful story of a website called EnjoyTheMusic.com that ran an article saying that:

Listening tests reveal significant sound quality differences between various digital music storage technologies.

Oh, my, where does one even begin with this? The fact that 1s and 0s are 1s and 0s? Or the fact that that hard drives in the tests are connected to all sorts of other components, such as data busses, connectors, cables (yep), and more.

Nah, not even worth suggesting such things. Instead, let’s marvel at this comment:

QNAP1 was found to serve up music with a similar level of rhythmic drive and image soundstaging as a good CD transport playing directly into our system’s DAC. If anything, there was perceptibly more ‘drive’, in the sense of bass euphony and articulation, but this came with increased level which made the sound a tad bass heavy.

I mean, if you think about it, the hard drive should provide exactly the same sound as a “CD transport” playing the same music. But here, there was more “drive” and it was louder (which, you know, is physically impossible since 1s and 0s don’t have loudness).

Nevertheless:

Also, QNAP1 did not sound as clean as CD in the higher registers. Some edgy grain exaggerated the sampled horns that sets the scene in the opening of Primal Scream’s Loaded, adding to the color but nudging it off neutrality. Splash cymbals lived up to their name.

Yes, the 1s and 0s were a bit clipped in the high frequencies. Those would be the frequencies higher than 0, I assume.

But then the Keystone audio testers tried using different hard drives in the same NAS. This is an interesting test, because it could should that EVERY HARD DRIVE SOUNDS EXACTLY THE SAME, but, somehow, through the magic of audiophilia, they didn’t.

In our initial listening tests, I couldn’t discern any tangible difference in sound between the two hard drives. Harris thought 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. Both lacked much drive in presenting the Madonna track, and were certainly ‘mushy’ compared with the best sound quality we’d heard from the QNAP stable.

Because some 1s and 0s have sharper and more thumpy bass. Now, one could suggest that the sharper and more thumpy bass comes from storing the music files on the outer edges of the disk’s platters, and our team of reviewers should have known that…

Then they moved to a couple of SSDs.

Drive three (a solid state type) gave a far from subtle shift in tone and soundstaging. I thought that here this Kingston SSD spread the stage wider, could really pull apart the multi-track layers, and certainly led in blackness too, sounding agreeably quieter than it had any right to. Yet there was also a dull flatness to its presentation, even a graying of timbre.

It could “pull apart the multi-track layers,” you know, the 1s and 0s. And it “led in blackness;” as you know, 0s are blacker than 1s. But it was dull, gray. Sigh.

What about the other SSD?

If the Kingston SSD stood apart from the disk drives for its mostly good yet quite alien character, drive four made itself known for entirely the wrong reasons. This Corsair drive (another SSD) conspicuously highlighted vocal sibilants, and had a hard, relentless quality that was impossible to miss. Strangely, it also robbed the music of pace; it was the least engaging on any emotional level thanks to an enveloping tunelessness that appeared to carve up a song like an MP3 rip.

Well, I won’t be buying any Corsair SSDs for my computers. It sounds like it’s rubbish!

To conclude:

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.

Really now? The bits are different? Hmmm, and how might one prove that… I wonder if there are ways to, you know, copy files and compare them to PROVE THAT THEY ARE EXACTLY THE SAME.

By now we should know better, and acknowledge that digital audio is very far from immutable.

Now, by now we should know better than to trust anyone who actually comes to this type of conclusion.

Post scriptum:

Why do these people never understand that the difference is between their ears? They were seeking differences, and they used the good old confirmation bias to find it. Also, humans are not the same all the time. As the day goes on, our ears and our brain change in the way they interpret sound and other stimuli. Maybe they had a cup of coffee before they listened to the drive that sounded “alien,” or maybe they only tested the SSDS after lunch, or after a few beers. This sort of subjective test is simply dumb. These “hypotheses” can be easily tested to prove that data out equals data in, but as long as there are fools willing to be parted with their money, they will keep on testing for unicorns.

28 thoughts on “How Hi-Fi Magazines Write about Cables, Part 17: Hard Disks

  1. A fun article. I appreciate that Kirk distilled the article for us, and selected choice bits of isolated silliness, presenting them with excellent definition in the wordstage. So many impossible assertions, called out in an amusing way.

    And on the subject of call outs, I will offer one for this author. Kirk says, “…and it was louder (which, you know, is physically impossible since 1s and 0s don’t have loudness).” While I can agree that 1s and 0s don’t have loudness, they also don’t have pitch, timber, or rhythm. And yet a digital sound file encodes all those musical aspects and features. So loudness definitely IS represented and stored as 1s and 0s, perhaps even “physically”, depending on your view of how physical bits are on a hard drive or SSD.

    The absurdity isn’t that a digital music stream can’t reproduce loudness variations, but rather conviction on the part of the original review authors, that the same digital stream could produce differences in the loudness of a piece of music as a result of the hard drive from which the digital file is read. An absurdity that Kirk fully understands, of course. I’m sure it was just an aberration in the digital typing stream, caused by the infidelity of the particular bluetooth keyboard that he was using at the time, which allowed this error to creep into the article.

    • Yes, my comment was shorthand for “this is something really, really stupid.” Loudness is part of the data that’s stored, but not in a way that the medium carrying it could change it. It’s not like an LP where, perhaps, engraving a master differently might lead to a slightly different level of volume.

      • Of course with vinyl, the volume is affected by small differences in the Earth’s gravitational field. This may explain why there are more audiophiles in London than in Leeds.

  2. A fun article. I appreciate that Kirk distilled the article for us, and selected choice bits of isolated silliness, presenting them with excellent definition in the wordstage. So many impossible assertions, called out in an amusing way.

    And on the subject of call outs, I will offer one for this author. Kirk says, “…and it was louder (which, you know, is physically impossible since 1s and 0s don’t have loudness).” While I can agree that 1s and 0s don’t have loudness, they also don’t have pitch, timber, or rhythm. And yet a digital sound file encodes all those musical aspects and features. So loudness definitely IS represented and stored as 1s and 0s, perhaps even “physically”, depending on your view of how physical bits are on a hard drive or SSD.

    The absurdity isn’t that a digital music stream can’t reproduce loudness variations, but rather conviction on the part of the original review authors, that the same digital stream could produce differences in the loudness of a piece of music as a result of the hard drive from which the digital file is read. An absurdity that Kirk fully understands, of course. I’m sure it was just an aberration in the digital typing stream, caused by the infidelity of the particular bluetooth keyboard that he was using at the time, which allowed this error to creep into the article.

    • Yes, my comment was shorthand for “this is something really, really stupid.” Loudness is part of the data that’s stored, but not in a way that the medium carrying it could change it. It’s not like an LP where, perhaps, engraving a master differently might lead to a slightly different level of volume.

      • Of course with vinyl, the volume is affected by small differences in the Earth’s gravitational field. This may explain why there are more audiophiles in London than in Leeds.

  3. I can’t wait to read the review of hard disks on movie playback quality. My SSD’s make my everything look like is was produced in 8K. Those 1’s and 0’s sure are imps.

  4. I can’t wait to read the review of hard disks on movie playback quality. My SSD’s make my everything look like is was produced in 8K. Those 1’s and 0’s sure are imps.

  5. I wonder if this discourse has its origin in part in audio production, where the speed of a hard drive can in fact be a factor in the sound. (If the drive is not fast enough, you can get drop outs, clicks and so forth if you are trying to stream too many audio files simultaneously off of one drive.) That production context is a different matter, of course, than serving a single stereo audio track from a hard drive, but perhaps these kinds of stories got started because they heard some audio production folks talking about needing to upgrade their hard drives to avoid sound problems. For listening to commercially produced music these days, my only concern would the noise of the drive itself. (Even today, some are much noisier than others, and SSDs have a general advantage in that respect.)

    • Perhaps, if you have very large files, then speed would be an issue. And I agree; I’d be more concerned about the noise of a drive, or of its enclosure, if there’s a fan in it.

      • It’s not even so much the size of the file so much as how many files you need to access at the same moment. At a certain point you hit either the read limit of the drive (pretty easy to do in a production context on a 5400 RPM drive) or the bus limit (if you are trying to use USB 2 or you have a lot of drives running through a USB 3 hub, for instance). I encounter problems, for instance when I’m running large numbers of virtual instruments all needing to stream a complex succession of samples from the same hard drive at the same time.

        But that is all at the production end, not on the consumer end. As a consumer, I have occasionally had problems streaming video on a USB 2 drive, but I never had problems with audio, and I’ve had no problem with video on USB 3.

        I’m not sure if you’ve addressed this or not, but what’s your opinion on the Apple sound cards (on the various devices) versus an audio interface for driving standard headphones?

        • I don’t know. I use an external DAC, and when I first got one some years ago, it did sound better. But that may just be subjective. I do know that iOS devices are well rated for their sound, that their DACs are considered quite good for portable devices.

  6. I wonder if this discourse has its origin in part in audio production, where the speed of a hard drive can in fact be a factor in the sound. (If the drive is not fast enough, you can get drop outs, clicks and so forth if you are trying to stream too many audio files simultaneously off of one drive.) That production context is a different matter, of course, than serving a single stereo audio track from a hard drive, but perhaps these kinds of stories got started because they heard some audio production folks talking about needing to upgrade their hard drives to avoid sound problems. For listening to commercially produced music these days, my only concern would the noise of the drive itself. (Even today, some are much noisier than others, and SSDs have a general advantage in that respect.)

    • Perhaps, if you have very large files, then speed would be an issue. And I agree; I’d be more concerned about the noise of a drive, or of its enclosure, if there’s a fan in it.

      • It’s not even so much the size of the file so much as how many files you need to access at the same moment. At a certain point you hit either the read limit of the drive (pretty easy to do in a production context on a 5400 RPM drive) or the bus limit (if you are trying to use USB 2 or you have a lot of drives running through a USB 3 hub, for instance). I encounter problems, for instance when I’m running large numbers of virtual instruments all needing to stream a complex succession of samples from the same hard drive at the same time.

        But that is all at the production end, not on the consumer end. As a consumer, I have occasionally had problems streaming video on a USB 2 drive, but I never had problems with audio, and I’ve had no problem with video on USB 3.

        I’m not sure if you’ve addressed this or not, but what’s your opinion on the Apple sound cards (on the various devices) versus an audio interface for driving standard headphones?

        • I don’t know. I use an external DAC, and when I first got one some years ago, it did sound better. But that may just be subjective. I do know that iOS devices are well rated for their sound, that their DACs are considered quite good for portable devices.

  7. Wow. I make my living recording, editing and mixing audio, and that’s one great exercise in creative writing. I worry about reliability, and therefore tend to replace my main audio drive every year. The quality of the converters, source, speakers/headphones is what matters.

  8. Wow. I make my living recording, editing and mixing audio, and that’s one great exercise in creative writing. I worry about reliability, and therefore tend to replace my main audio drive every year. The quality of the converters, source, speakers/headphones is what matters.

  9. I used to think that the type hard drive can in no way what so ever affect the sound in a home audio system. Then I replaced my old Seagate Barracuda HDD with a Samsung SDD because I needed more space and wanted a speedier and more silent drive. Boy was I in for a surprise. The soundstage was noticeably smaller, the bass was muddier, details that used to be clear and noticeable were now buried in the mix. Vocals that used to shine and be up front in the center were now flat and dull. How could this be? I couldn’t believe it. I reformated the drive and made a new copy of the music – didn’t change anything. Could there be something wrong with the SSD? I ran some software checks, but they all showed the disk was fine. Then I tried another SSD (Corsair). Same result. Then I thought maybe I am just somehow fooling my self, it’s all in my head or something…maybe if I install both disks and do some A/B testing I will surely hear that the drive makes no difference. So I did, but this just solidified the observed differences. So to my total surprise, I was forced by the very real evidences to accept the facts. In my high end audio system (Naim streamer, Burmester DAC, Adyton power amp and speakers), the type of disc drive do make a difference in the audio reproduction. There must be a technical explanation for this. I am into music, not the technical aspects of audio reproduction, so I have no theory to offer here.

  10. I used to think that the type hard drive can in no way what so ever affect the sound in a home audio system. Then I replaced my old Seagate Barracuda HDD with a Samsung SDD because I needed more space and wanted a speedier and more silent drive. Boy was I in for a surprise. The soundstage was noticeably smaller, the bass was muddier, details that used to be clear and noticeable were now buried in the mix. Vocals that used to shine and be up front in the center were now flat and dull. How could this be? I couldn’t believe it. I reformated the drive and made a new copy of the music – didn’t change anything. Could there be something wrong with the SSD? I ran some software checks, but they all showed the disk was fine. Then I tried another SSD (Corsair). Same result. Then I thought maybe I am just somehow fooling my self, it’s all in my head or something…maybe if I install both disks and do some A/B testing I will surely hear that the drive makes no difference. So I did, but this just solidified the observed differences. So to my total surprise, I was forced by the very real evidences to accept the facts. In my high end audio system (Naim streamer, Burmester DAC, Adyton power amp and speakers), the type of disc drive do make a difference in the audio reproduction. There must be a technical explanation for this. I am into music, not the technical aspects of audio reproduction, so I have no theory to offer here.

  11. Bjørn, like you, my first reaction is “I couldn’t believe it.” If your careful observations are triggered by something outside of your head, then it must be measurable. How could you confirm your ear/mind observations via a different technology? Obviously, digital data can be corrupted. So the first thing to check is whether the digital files on the two different drives are identical. You can do this with Terminal, among other ways. I’m going to guess that they are the same, bit per bit.

    If so, then there would have to be a difference in the digital data stream that is served to the rest of your music reproduction chain. A couple of decades ago, when a drive couldn’t keep up with the speed at which the data needed to be delivered, it was easy to hear dropouts and other artifacts. It’s unlikely that this could happen with audio from an SSD, but it is testable, if you can capture the digital stream that comes from each drive as the music is played, and copy some of that data to a file. Or record the sound output with something like Audio Hijack, and compare the resulting sound files. Intercutting those sound files would allow you to create a more reliable A/B test. With an assistant, it could be a double blind test.

    It might be simpler to invite three or four friends over, and see if they can tell consistently which drive you are using for playback. If you do any more testing, I would love to read about it here.

    • I’ve been reading up on this subject lately and now have a theory: I think there is no reason to believe that there are any differences between the very data on the drives. I think what is happening here has to to with jitter (digital noise) generated when data are fetched from the drive and fed to the DAC. These drives are fundamentally different in how they store data, the HDD fx. has a 64 Mb cache, the SSD has none, hence there are numerous sources of jitter. And jitter is, as I assume everyone here knows, the main source of the audible quality differences in the analog reproduction of digital audio sources. Jitter is basically aberrations in the time code in the data flow. And these aberrations can result in all sorts of changes in the audio spectrum if not dealt with sufficiently (fx. in the DAC). The S/PDIF standard allows these synchronisation errors up to a certain point without data loss. If the standard had zero tolerance, analog reproduction of digital sources would not be possible with any technology known today.

      My conclusion so far, is that the very construction of the storage medium is a jitter source. Even though the very data are identical when stored on the different drives, digital noise is introduced when these data are sent from the drive to the DAC, and this is why there are audible differences in the analog reproduction.

      • Thanks for your additional information, Bjørn. I had no idea that some/all SSDs have no cache. That sounds like a perfect recipe for timing issues. Keep us posted on your discoveries.

  12. Bjørn, like you, my first reaction is “I couldn’t believe it.” If your careful observations are triggered by something outside of your head, then it must be measurable. How could you confirm your ear/mind observations via a different technology? Obviously, digital data can be corrupted. So the first thing to check is whether the digital files on the two different drives are identical. You can do this with Terminal, among other ways. I’m going to guess that they are the same, bit per bit.

    If so, then there would have to be a difference in the digital data stream that is served to the rest of your music reproduction chain. A couple of decades ago, when a drive couldn’t keep up with the speed at which the data needed to be delivered, it was easy to hear dropouts and other artifacts. It’s unlikely that this could happen with audio from an SSD, but it is testable, if you can capture the digital stream that comes from each drive as the music is played, and copy some of that data to a file. Or record the sound output with something like Audio Hijack, and compare the resulting sound files. Intercutting those sound files would allow you to create a more reliable A/B test. With an assistant, it could be a double blind test.

    It might be simpler to invite three or four friends over, and see if they can tell consistently which drive you are using for playback. If you do any more testing, I would love to read about it here.

    • I’ve been reading up on this subject lately and now have a theory: I think there is no reason to believe that there are any differences between the very data on the drives. I think what is happening here has to to with jitter (digital noise) generated when data are fetched from the drive and fed to the DAC. These drives are fundamentally different in how they store data, the HDD fx. has a 64 Mb cache, the SSD has none, hence there are numerous sources of jitter. And jitter is, as I assume everyone here knows, the main source of the audible quality differences in the analog reproduction of digital audio sources. Jitter is basically aberrations in the time code in the data flow. And these aberrations can result in all sorts of changes in the audio spectrum if not dealt with sufficiently (fx. in the DAC). The S/PDIF standard allows these synchronisation errors up to a certain point without data loss. If the standard had zero tolerance, analog reproduction of digital sources would not be possible with any technology known today.

      My conclusion so far, is that the very construction of the storage medium is a jitter source. Even though the very data are identical when stored on the different drives, digital noise is introduced when these data are sent from the drive to the DAC, and this is why there are audible differences in the analog reproduction.

      • Thanks for your additional information, Bjørn. I had no idea that some/all SSDs have no cache. That sounds like a perfect recipe for timing issues. Keep us posted on your discoveries.

Leave a Reply to Jim Buhler Cancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.