Significant sound quality differences between digital music storage technologies | Hacker News Reader

Significant sound quality differences between digital music storage technologies

enjoythemusic.com

29 pointslelf44 comments

People are passing this off as silly, but there's a valid concern here - whether the lossless compression algorithms can actually infect the source material based off of how it interacts with the hardware. I've noticed this myself while coding - late at night, my software works great when it is all uncompressed and spread out in my IDE, but after I commit it (undoubtedly to remote hard drives that compress my code), it's pulled by others, and sometimes those other people experience bugs. The bug didn't happen for me when it was uncompressed in my IDE. But after the compression/decompression cycle of committing/pulling, it's obvious that the code sometimes gets infected by the hard drives themselves. If you're in a silent room when you push to github, you can actually hear the alterations if you listen really, really hard.

That is a very serious problem I often face myself. I found that having a cactus near the monitor influences the way information is encoded and perceived. When a person that does not have a cactus tries to run my code, they could face cactus not found and sometimes even cactus panic issues. Solution of course is to bring my cactus to their workstation; it is obvious that a reboot is required so that workstation could recognize new peripheral plantware. Working with servers is much harder, but waving a cactus in front of terminal window and changing font color to green had been proven to work.

I see what you did there.

These tests are all horribly flawed and are in effect measuring the same error: edge cadence. When you pack the bits on a drive they are compressed and if you pull them back off and send them directly to your DAC you haven't give the bit edges a chance to expand back to their natural state. Hence the distorting edge cadence. The only way to truly remove this effect is to stage the bits in a ram disk and let them rest. Minimum of two minutes but if you can, five minutes allows full expansion of the soundscape. Then you send the stream from the ram disk on to your DAC and, voila, no edge cadence and prefect sound every time!

Too bad we missed our chance to demo the new ($2,499) edge-cadence eliminating audiophile quality PCI-E cache (with pure gold connectors) SSD at CES. It's all point-to-point wired with 12 gauge cryo-treated OFC solid conductor wire sheathed in Techflex for the best possible quality. We also have a balanced model coming out which uses three separate 1x PCI-E slots for L/R/ground to decrease cross-talk (and obviously jitter).

If it's not oxygen-free copper wire, with gold mirror surface, I'm not buying.

/sarcasm

We need to talk, this could be complementary to our $5,430 air stabilizing chamber prevent turbulence to create lens effect that boost the interference of cosmic rays and still get through the techflex. We also offer the option to replace the air inside the chamber with a neutral gaz. Its helps to prevent the changes in temperature and therefore thermal expansion of the copper wires and gold plating to create unwanted variation in signal.

Why is nobody addressing the neutrino coloration issues at the heart of the issue? We offer a private listening space ($599 an hour) unlike any other, lined with a few light-years of lead (substantially compressed) to prevent a minority of neutrinos from creating aberrant distortions in source material. Together with the vibration-reducing space-time deformation we create by putting this much mass in one place, we've heard from several noted audiophiles that this is the only true way to reveal the music the way the artist intended.

Finally someone caught on to this problem.

140 years ago my great-grandfather dug a palladium mine anticipating exactly these problems with internal disk drive components causing imperfections in audio playback. The steers grazing the grass above the mines were fed gold putty so as to increase the conductivity of their excrement. Over the years this has resulted in superconductive minerals in the earth beneath my house.

As a result, for most of my life I have enjoyed unmolested audio, because we store our files on handmade disk drives whose internal components were hewn of this superconductive mineral. There was one time in 1994 when a cousin, who'd briefly slipped the bonds of his reeducation room, tried to use copper solder from RadioShack to fix a transistor to a PCB for use inside one of my drives. We haven't had any issues with imperfect sound once we killed him.

Sometimes we can hear a train whistle from down the road, though. That's annoying.

This is a joke right?

"Penguin Café Orchestra's Union Café has an altogether more natural recorded acoustic. On Scherzo and Trio QNAP1 promoted the leading edge piano transients, following through with a lighter, brighter instrument tone – possibly Steinway-like? The same piano had more lower mid body on QNAP2 and slightly softer hammer impact, perhaps more like a Bosendörfer.

That hint of glaze on QNAP1 also showed an impaired subjective noise floor elsewhere. In hi-ﬁ parlance, QNAP2 had the blacker silences and deeper spaces between notes. If anything, this track highlighted a fundamental shift in timbre between the storage sources. This wasn't the gentle tweak of a DAC's digital filter option; we felt it was more akin to changing loudspeakers. System sound was improved as if the DAC itself had been upgraded, say from a £500 to a £2000 model."

I can only assume this is a joke pointed at audiophiles, the sort who promote the 192khz/24 bit music.

http://xiph.org/~xiphmont/demo/neil-young.html

My first thought was that this post is making fun of people who pay thousands of dollars for audio cables.

I'm think this is yet another case of Poe's law at work. As far as I can tell, this was published in a bonafide audiophile magazine and that the authors were dead serious.

Here are some skeptics discussing the article: http://forums.naimaudio.com/topic/listening-to-digital-music...

Poe's law.

"QNAP1 promoted the leading edge piano transients, following through with a lighter, brighter instrument tone – possibly Steinway-like? The same piano had more lower mid body on QNAP2 and slightly softer hammer impact, perhaps more like a Bosendörfer."

I suspect just using an off the shelf hard drive would sound more like a Yamaha.

"Fellow computer audio enthusiast (and Naim PR person) Stephen Harris and I launched into some preliminary listening tests, to establish under reasonably controlled conditions if audible, repeatable differences could really be heard. We readily confirmed that the final sound quality is influenced not only by the choice of network player, DAC, digital cables, or indeed many other long-recognized factors, but additionally — and quite markedly — by the manner in which we now store large quantities of our music at home."

First, the nature of the tests conducted is not adequately described. Were they double blind ABX? Second, they confirmed a sound difference between digital cables, which many others have utterly failed to do in double-blind ABX tests. This suggests to me that their testing methodology is deeply flawed and the author should be disregarded unless he posts more details.

IF there is a significant difference in the quality of sound between one brand of disks in a NAS and another, then something is seriously broken or badly designed in the DAC.

On the other hand, I don't trust those who don't do controlled experiments to have unbiased conclusions on audio gear.

Way back when I cared about audio, I was listening to some high end (~10k/component) cd player + power supply at a Naim dealer. The sales guy was raving about the sound, but it was just wrong to me. The soundstage was just wrong. Weirdly sideways. After a going back and forth a few times, visual inspection proved the speakers were wired out of phase. So, not controlled experiment, and not a magazine columnist doing the miswiring, but if you can't tell when the basics aren't right, how can you expect to hear nth order effects?

Are audiophiles the most irrational consumers? They seem to be quite a bit more sillypants than the new agey alt-med crowd. Homeopathy is almost certainly just a fancy placebo, but the attempts I've read at putting a modern theory under homeopathy (e.g. water having state memory) actually sound less cuckoo than the stuff I see on audiophile sites.

Assuming this is sincere, I find the most striking thing about it to be the mismatch between the high quality of the writing and the extreme silliness of the conclusions.

Have you tried reading the article over gigabit ethernet? I find that, as compared to wifi, it makes the writing slightly more dry but adds punch to the conclusion.

This test obviously could benefit from a Blackbody Ambient Field Conditioner (MSRP $1,323, now discontinued)

http://www.lessloss.com/blackbody-p-200.html

Let us not forget the importance of vibration control for digital devices! http://shop.mapleshadestore.com/products.asp?dept=127

I use one of these under my laptop and my code compiles faster and runs a lot quicker!

TCP buffers, and caching eliminate any argument about Jitter that one could pose on this issue. Timing of file transfer and playback rate on the DAC are not linked, so this study is absurdly flawed.

If these people ever learned the meaning of "double blind" would their heads explode?

They know the meaning, and actively argue against it:

http://www.stereophile.com/content/listening-143

http://www.theabsolutesound.com/forums/threads/13/

I have also noticed that the content of blog articles varies slightly depending on who manufactures the ram that I'm using - related issue?

Surely this is a joke? Not only are the bitstreams coming from the hdd digital, they're also buffered and decoded. I can see different decoders affecting the sound, but not the hdd.

So the measurement of output quality is just... the reviewer's ear?

Moving along.

Trained, experienced individuals with excellent hearing can be used as "detectors" quite well in blind studies. That's not the issue.

The real crux of the matter is that the article is an example of Poe's law. It's satire, but very well done.

Why do you think this article is satire?

Transposing Clarke's law: sufficiently advanced satire is indistinguishable from sincere bullshit. (And, of course, vice versa.)

Have these guys heard of caching?

The usual dogma among those with a passing familiarity with digital audio seems to be to the effect of, "Bits are bits - it doesn't matter where they come from."

This, as it turns out, is not entirely correct. Jitter can be a substantial problem in digital transport and does have well-understood effects on overall sound quality, and certain DAC designs are much better at jitter rejection than others.

Here, it's a bit of a stretch, but not entirely outside the realm of possibility for jitter to be affected at varying levels by the storage system in use, though there are a lot of variables here (router, UnitiServe, and Naim NDX) that may not have been adequately isolated.

That being said, I don't deny that there's a substantial amount of 'snake oil' in the audiophile industry, and reviews like this should be considered with some skepticism - but that shouldn't mean we dismiss them outright.

I would, of course, like to see tests indicating these differences, as - if they exist - they will be quantifiable with the right equipment, e.g., an Audio Precision analyzer.

No, jitter is not a problem. It is more than a stretch to go from "jitter as a theoretical problem" to "jitter as an actual problem" - it is absurd. Don't even give this credence.

The only way to make jitter audible is to do it deliberately, to the extent that the DAC would have difficulty even locking to it. At that point, it would likely end up dropping samples, and we're not really talking about digital audio any more.

Audio marketing has been using crap like "digital jitter" as one of their snake oil sales tools for as long as digital audio has been commonplace. Don't give it even the slightest amount of publicity.

I used to design digital electronics in a past life. When the slave has locked onto the master clock, and remains there locked for a long time, it's game over. The slave becomes effectively a clamp-down filter on the master's real or imaginary deficiencies. Yes, the master does matter still, but a quality, stable slave acts as a damper.

The slave's own jitter tendencies might be important - but then again, where do they end up? Mostly in the analog /dev/null sinkhole - the sampling frequency is far above the audio spectrum, the jitter components are too, mostly (unless it's low-freq drift), and the output of your DAC has a steep downpass filter. Jitter components are obliterated.

That's assuming that jitter would matter at all even if it was not filtered out. But if a DAC is flickering that badly that you could hear it, it's probably a piece of trash anyway.

Make sure the actual digital-to-analog conversion (the last link in the digital chain) is done well, and don't worry about anything else. Yes, you can screw up DAC design and execution, but nowadays most DACs are very good. This is not your grandpa's radio.

You must be unfamiliar with the AES and other papers that describe audible jitter and the threshold at which it starts to become audible. See e.g.

"Eric Benjamin and Benjamin Gannon, "Theoretical and Audible Effects of Jitter on Digital Audio Quality", Preprint 4826 of the 105th AES Convention, San Francisco, September 1998"

"The Effects of Sampling Clock Jitter on Nyquist Sampling Analog-to-Digital Converters, and on Oversampling Delta-Sigma ADCs, 87th Convention of the Audio Engineering Society, October, 1989"

Ashihara and Kiryu, etc (2005)

https://www.jstage.jst.go.jp/article/ast/26/1/26_1_50/_pdf

I am quite familiar and in fact that paper concludes that jitter is not detectable in normal consumer equipment:

"About 25% of the listeners detected jitter when its size was 500ns. When it was 250ns, however, no listener could discriminate the sounds."

...

"Nishimura and Koizumi made attempts to measure actual jitter of various DA systems during reproduction of music signals. They could not detect any jitter larger than 3ns in their measurements."

...

"So far, actual jitter in consumer products seems to be too small to be detected at least for reproduction of music signals"

...

CONCLUSION "The threshold values seem to be sufficiently larger than the jitter actually observed in various consumer products."

You should point out that typical jitter on the digital interface will be in the order of few 100 PICOSECONDS...

and is totally unrelated to any jitter on the storage system (unless it's so slow that it can't keep up with the DAC, which is unlikely to happen unless it's a 20th century hard drive or connected by a wireless/internet connection).

Rule of thumb: when discussion of vanishingly small effects as they regard human perception (especially when applied to mere entertainment) devolve to requiring analysis "quantifiable with the right equipment", we've pretty much left the realm of human perception and the cost to obtain further improvements are pure waste.

Yep, I was just having this conversation with a friend recently, he was asking about high definition audio, high end usb cables, power conditioners other things like that. I own a good system to listen to audio from the computer (a Hegel H80) so he was asking me what I thought.

I was having a hard tome to convince him that at one point you're better off spending money on better recordings and more music than going after the imperceptible (or not existing) differences through these kinds of equipment "upgrades".

My point wasn't about effects being so small they're imperceptible - it was that human perception is notoriously subjective (particularly in the audiophile world where people tend to want to hear differences even if they don't exist), and an AP analyzer (or similar) would provide a much more rigorous way of identifying if an effect was present.

As far as perceptibility is concerned, ABX testing is a better method for that, and it really doesn't appear the individuals in this article conducted their listening tests via that method.

But here's the thing: The clock source that goes into your DAC isn't influenced by the pattern of packets arriving from your storage system, it's typically a standalone oscillator sitting in the vincinity of the DAC and its controller in the shiny converter box... (for Pro-Audio it will be a rather dull 19" rackmount box, for audiophile stuff it will be brushed aluminum and have wooden knobs...).

The only bad thing that can happen with the sound quality is that the computer isn't fast enough to keep up with supplying the controller next to the DAC (often in consumer hardware everything is integrated in one piece of silicon, though) with data, which is normally called a buffer-underrun. The ADC then doesn't know what to do and might produce clicks due to the sudden frames of silence inserted, some might just replay the last frame over and over again (stuttering or harsh noises). This is obviously audible to a huge extent.

And yes, there are synchronization issues, and sometimes devices resample to bridge synchronization barrier., You'll be able to configure master/slave relationships, distribute word-clocks, ... but for the simple task of listening to a FLAC file of a Jazz Concert, those complications just don't apply.

Reply on news.ycombinator.com