The State of USB Audio by Alan Taffel

[BobH]

(I see what you mean, Eloise!)

 

Leaving aside the pedantry over my choice of the word 'stream', I did concede the likely viability of well designed usb dac's. (Granted, it can be made to be very good indeed - or so the reviews of well implemented usb dac's would indicate.). I didn't mention a usb-->spdif connection so, yes, we may as well let it lie.

 

As for the rest of it, well, I thought it perfectly clear that this was my experience and IMHO. You don't have to agree, of course, but that doesn't invalidate my preferences. I happen to think that usb is far from ideal as an audio interface. I happen to think that coax spdif via a decent soundcard beats it. I happen to think that the networked implementations I have listened to beat it. I happen to agree with Mr Taffel's findings.

 

Just thought I'd mention it.

 

[cfmsp]

 

"It appears that there's an orchestrated effort on the part of one or more manufacturers of USB DACs to attempt to discredit Alan Taffel's article."

 

Wow, this takes the cake. Rather than admit that they could have done a better job, they drum up a conspiracy theory to account for a groundswell of outrage.

 

As a fairly vocal proponent of Firewire as an interface (but no financial interest), I have NOTHING to gain by the success of USB audio, indeed, Async USB is competition to the advantage that Firewire holds over S/PDIF in the computer audio sweepstakes. yet, I was outraged by the claims made by Alan Taffel.

 

Count me as another former subscriber.

 

Clay

 

[idiot_savant]

Quote : "I think I tend to agree with Mr. Taffel. Not his methodology, or the broad sweeping statements, but with the general notion that usb, as an audio interface, is rubbish."

 

you are, of course, welcome to your opinions, which I respect.

I'm just pointing out that the advantages ( of ethernet, firewire and USB ) cannot be fully expressed if your DAC just accepts SPDIF ( with, e.g. no clock back ), and that slating one for no technical reason over the other ones is plainly bad journalism.

EDIT I've just re-read your post, and can _tell you_ that the USB implementation you have is rubbish.

never trust a company that can't even spell the greek god correctly ( sorry ). USB adaptive is the worst jitter interface in the world...

/EDIT

 

My less than subtle hint is that whilst the greatest factor in early proliferation of USB is, for example mice, keyboards and printers, the 2 biggest factors in early proliferation in TCP/IP are military usage and pornography. They've both moved on, probably?

 

I'm here all week,

 

your friendly neighbourhood idiot

 

 

[The Computer Audiophile]

Just an FYI that some of the very expensive Network connected DACs are really Ethernet to S/PDIF converters. Inside the box the stream is converted right to S/PDIF. Take it for what it is.

 

We should also keep this thread focussed on Alan's article instead of turning this into an infighting thread where no progress is made :~)

 

[idiot_savant]

I'll back out,

 

your friendly neighbourhood idiot

 

[The Computer Audiophile]

No worries. Everyone here has some great info to offer. It's just a matter of keeping us on track :~)

 

[barrows]

Interesting thread... It seems Alan's TAS article really did a disservice to the audio community. While I agree with his findings regarding what he did test, I suspect that his findings would have been quite different if he tested an Ayre QB-9, or Wavelength Cosecant in an optimized system, and the fact that he did not even mention the possibility that there are other USB implementations available that could provide better performance is just plain poor journalism.

If Alan's approach is to provide an overview of what is available, it would seem the inclusion of units using ASYNC protocol would be necessary. Alan's supposition that computer audio should be plug and play to achieve the best possible results is quite absurd, considering that TAS is an audiophile magazine, which stresses the importance of proper system setup, exotic audio cabling, exotic vibration management devices, CD treatment fluids, etc!

It is interesting to note that TAS Editor, Robert Harley, reviewed the Reference Recordings Hrx format using a highly optimized computer audio system (custom built PC with SSD, Lynx soundcard and Berkeley Audio Alpha DAC-TAS January 2009 issue): this system was much more sophisticated than is necessary to get good performance from, say, an Ayre QB-9. Additionally, suggesting that there are no audiophile Firewire DACs is just plain wrong-the Weiss Minerva is specifically an audiophile Firewire DAC, distributed through traditional audiophile dealers, and promoted at audio shows like CES.

I hope that soon TAS will consider the errors it made in printing this incomplete/innaccurate article, and will consider writing an informative article, showing its readership how to actually get excellent performance from a computer based playback system. I would suggest that they reconsider dealing with Wavelength Audio, and allow Gordon Rankin to help them setup a simple system, that any audiophile could easily recreate (like, MacBook, 4GB RAM, decent USB cable, Cosecant) and explain in this article how to rip, store, and playback files to get the best performance. This stuff is not rocket science, and setting up a system like this is way easier (and more affordable) than setting up a good turntable/cartridge/phono stage! An article like the one I just described would provide something of great value to the TAS readership, and the audio industry as a whole. It is no secret that audiophiles in general are an aging, dwindling, population, and the sooner the mainstream audio press embraces the possibility of high quality audio playback from computer audio sources, the better the chance of promoting high end audio to the IPod generation.

 

[satkinsn]

idiot wrote:

 

"my take on the whole PC/Mac being FAST enough is pretty simple - if the CPU can't process enough audio, it doesn't degrade in a subtle way. Packets ( which are probably of the order of 100ms or so ) will not reach the output in time - this won't cause "a collapse of soundstage" or "an increase in sharpness" - it will cause a big pop, bang or crackle - the data is there (good), or it's gone missing or is late ( bad )- imagine stitching together bits of tape 1/10 of a second long with variable gaps in it. Smaller packets ( which require MORE processing time ) will case smaller pops, but it will be obvious."

 

Exactly what I was wondering about. If you're right - and I suspect you are - doesn't it throw into question the need for a high end computer for music playback, assuming you do the rest of it right?

 

tks,

 

s.

 

[IamKirk]

This whole argument kind of seems like it's being mad at a newspaper for not telling you what happened 2 hours ago when the internet can. Why would you care what magazines say in this age when you can go online to Computeraudiophile.com and talk to someone that made the product.

 

[JohnB]

Does the article really surprise anyone? I recently started getting TAS and to be honest, just don't find it to be that good or interesting - not the cover to cover reading I used to do of my audio and A/V mags. Maybe I don't click with the reviewers (i.e. similar taste in music, types of gear), maybe it is the fact so much of what they review feels so esoteric and outrageously expensive (though I'm sure if I had the money to burn I might very well be buying some of the gear they review but alas my pockets are much emptier these days). Truthfully this isn't just TAS, it feels that way with most of the audio (Stereophile) and A/V mags I get (Sound & Vision and Home Theater). I remember the days maybe 5 or 6 years ago when I'd read these things cover to cover and the coverage just felt "right". Now days it seems to be so much vinyl and gear so far out of my price range I've just lost interest in them - and they really don't even seem to want to touch computer audio or computer-based A/V systems, both areas I find myself really excited about. Even their blasting of iPods as crappy basically cutting off coverage there frustrates me, instead I feel they should be trying to help us achieve the best possible sound from these devices. While an iPod might offer crappy sound it along with MP3's has really helped me rediscover music and has gotten me excited about music again and from that to created a computer based music system originally starting with my ripped mp3's but then I re-ripped everything to AppleLossless to get the best fidelity I could - and now I'm constantly on this site looking at how I might improve upon my current set-up.

 

I hope the article on USB audio is just the beginning, that maybe they are seeing computer-based audio as a growing segment and one worth helping build and inform but my gut tells me my dissatisfaction will continue and I will continue to be outside the demographic they are trying to reach. Sad, I really used to enjoy these magazines...

 

 

John

 

[lxgreen]

I have been wading through all these comments about how USB is the way to go if "done right". I currently stream by audio files via a NAS to a squeezebox to a modded music hall dac and it sounds very nice, almost as nice as my vinyl from my rega p3/24 and high quality tube preamp and amps. I am willing to experiment with a USB solution but need a better road map to configure the computer than all of the posters here seem to imply is so critical. Can someone give me better details of what this entails? i am currently using a PC running XP. thanks

 

[ampman]

Chris has hit the nail on the head with his unbiased reviews of USB DAC's. TAS as in most publications are biased to whatever their hidden agenda is. It is crazy that in TAS they bash USB sound and in the next article they praise USB DAC's.

 

 

[souptin]

Is this available to read online anywhere? I followed the link earlier in this topic, but it took me to another forum talking about the article, not the article itself. It's a US magazine, not easy to find here in the UK.

 

FWIW, the UK hifi press have an annoying tendency to skirt around computer audio. Even when they review one of the increasing number of components with computer connectivity they'll say things like "oh and there's a usb port, but we didn't really test it" etc. Often they don't even say if the usb port is for iPod, memory stick, or audio. It's as if they're in denial of the whole computer thing, which I believe is a mistake as many people have already said.

 

As to the accusations of bias and hidden agendas... weeeellllll yeah I suppose. Editorial and advertising have always had an awkward relationship. As most of you guys will probably be aware, it is quite possible for a print magazine to make money and not charge its readership anything. Think of the trade magazines you get delivered free to your door or work regardless of whether you ask for them or not. In fact one of the main reasons for there being a cover price is that it adds perceived value to the advertisers (ie they can charge more for a page of advertising). Not just in print. I do believe there are one or two ads right here on CA *cough - thinks of the days before Adblock* and it doesn't make me completely distrust everything I read here.

 

[Gag Halfrunt]

Newspapers and magazines CAN survive against online sources, if and only if they resort to actual journalism. In the US - where newspapers tend to be a combination of re-runs of AP/Reuters newswire items and their own posturing - papers are dying back at great speed. In contrast, the Daily Telegraph in the UK has been running something approaching journalism recently (it broke the 'Moatgate' politicians' expenses scandal) and has seen a huge rise in circulation. And this is in a country where there's a huge publicly-funded news website (the BBC) competing for free.

 

With TAS, I see no such journalistic intent. Instead I see nothing but keeping the advertisers sweet.

 

 

 

[manisandher]

It seems that the biggest issue that most people have with Alan's piece is that he didn't seem to include async USB DACs in his review.

 

If you look at JA's jitter measurements in Stereophile, you will see that the Wavelength's jitter measurements are beaten quite easily by DACs with spdif inputs, e.g. the Benchmark. Spdif simply has better jitter performance than both USB and async USB.

 

I'm not surprised that Alan preferred spdif...

 

Of course the problem then becomes, "How do connect the DAC to the computer?"

 

Mani.

 

[solidsqual]

"If it measures good and sounds bad, -- it is bad. If it sounds good and measures bad, -- you've measured the wrong thing." Daniel R. von Recklinghausen

 

[manisandher]

The fact that Alan did not include an async USB DAC isn't a big deal in my opinion.

 

If an async USB DAC sounds good, then it's not because it's producing increadibly low jitter. Yes, it seems async USB is better than 'standard' USB in terms of jitter, but it's still worse than well-implemented spdif. (Again, compare the Wavelength to the Benchmark in Stereophile.)

 

And why am I harping on about jitter? Well, isn't that exactly what async USB is all about reducing? (And it still can't beat good old spdif!) If I'm wrong, and there are other sonic benefits to using async USB, please let me know...

 

Mani.

 

 

 

[CG]

OK, you are wrong.

 

The Stereophile measurements seem to show some additional problem taking place, perhaps from the test set-up, since there's power supply remnants that appear in the output plot. According to the text, these disappeared when using a different battery powered computer.

 

A number of the SPDIF solutions use an asynchronous sample rate converter to reduce jitter. Large amounts of jitter reduction is usually needed, because the SPDIF interface itself is fraught with limitations. The data is buried within the clock stream, so changes in the data cause jitter in the clock stream unless a lot of conditions are carefully met. Which they almost never are. So, ASRC is used. That really *does* reduce the jitter, but then the actual data bit stream is modified to make this work. Is that a good thing? There are other pretty effective solutions as well, but they are rarely used.

 

To be fair, and please nobody skewer me for saying this, it's quite likely that many if not most people won't notice or care. Other parts of their overall system may mask any negative effects of jitter. Or, more likely, they themselves can't even tell the difference. Based on people's background, learning, and genetic disposition they may not be sensitive to whatever side effects poor jitter performance brings. How else do you explain the wide variation of what people think is the "Absolute Sound" (haha) to them? From horns and SET amplifiers to kilowatt push pull solid state amplifiers driving Wilson X3's, they all sound different. Yet, they deliver meaningful sound reproduction to a wide range of people.

 

People obviously respond differently to these wide ranging systems. If the various SPDIF and other solutions were not good enough for a number of people, then they'd disappear from the marketplace.

 

[manisandher]

@CG,

 

You need glasses!

 

You said:

"OK, you are wrong.

 

The Stereophile measurements seem to show some additional problem taking place, perhaps from the test set-up, since there's power supply remnants that appear in the output plot. According to the text, these disappeared when using a different battery powered computer."

 

They don't disappear - they're still there, but now at around -114dB (as opposed to -95dB!). In any event, this is still poorer performance than the Benchmark using spdif, which was my point!

 

Now, is the Benchmark using some sort of jitter reduction? Yes, for sure. But that's allowed, no?

 

Mani.

 

[CG]

I do need glasses, but not for this.

 

If you're good with the Benchmark DAC, great for you. I mean that.

 

 

[manisandher]

The highest level jitter measured for the Benchmark is around -121dB. This IS lower than -114dB... whether you agree or not.

 

I.e. the Benchmark's spdif performance (measuring jitter) is better than the Wavelength's async USB performance.

 

FACT!

 

Mani.

 

[CG]

I guess you win - I concede. I'll email Wavelength, Ayre, dCS, and the various Firewire guys right now to tell them how off base they are. Perhaps they can have a fire sale before anybody else catches on.

 

To me, what form this jitter takes and how it's attained is more important, at least in my value system. But, as they say, YMMV.

 

I don't think I have anything more to offer in this thread, either.

 

 

 

[manisandher]

I love being right! Actually, the only time I've ever been wrong was once when I thought I was wrong... but I was actually right.

 

FWIW, I am a 'Firewire guy' and have absolutely no vested interest in whether USB is good or not. But I do like facts...

 

Mani.

 

[idiot_savant]

Mani,

 

there are a couple of problems comparing these two sets of measurements. Firstly, as alluded to by CG is that the benchmark uses an ASRC to reduce jitter. It does this by doing some convoluted maths that changes jitter from being predictable sidebands ( i.e. visible equally spaced above and below the fundamental ) into unpredictable tones, or noise, or some other facet dependent on the precise ratio of the two clocks, the jitter, and the ASRC chip used.

If you look closely at the benchmark measurements ( january 2008 issue ), you will notice the 229Hz square wave there - so, you can see the 33rd harmonic at 7.5k or so, the 35th harmonic at 8k or so, etc. etc. Now, on the benchmark, my contention would be that the ASRC has converted the jitter/clock differences so that the related distortion is spread around - are there spuriae of around the same level as the harmonics we expect? Could these be jitter artefacts, and in total, could they, in fact sum up to be more than the wavelength? I would contend, also that they are more likely to persist and move around than the wavelength ( I don't know, I don't have either product ).

This is where we come to the second part of the problem - the wavelength is unfortunately on the noisy side due to it's valve output stage, so it's hard to actually really measure the jitter using this technique,

 

your friendly neighbourhood idiot

 

[rlodad]

Barrows,

Maybe we should ask Mikey Fremer to do a computer audio USB/Firewire set-up DVD?

 

O.K. Maybe not...

 

:^)