What is it that makes a good audio out?

Hi Cavan - This is really an engineering question that is certainly out of my league. I can say that any one of these interfaces / outputs can be better than the other based on its specific implementation. For example Steve Nugent's USB implementations at Empirical Audio are much better than a MacBook's built-in optical output, but the MacBook's built-in optical output may be better than a different USB implementation from someone else.

The most often talked about spec is jitter. There are several techniques for reducing jitter, but I hope someone more learned than I can offer better information based on your question.

Hi Cavan and Chris:

I was just about to ask almost the same question.

I have a few facts, but still wondering which connection type to use.

1.) I downloaded a few albums from Linn to play on my office system. I use a Macbook Pro, and connect to my Bryston B100 DA integrated amplifier with a Toslink cable. So far most of the files have been 24/88.2, one 24/96 album, and one 24/192 track (downloaded from 2L). I must say, the sound quality and detail was far better than I expected, and can see why computer audio can be so addictive.

The 24/192 track would not play at first, then sounded horrible. Of course I know why. First, that size of file requires a cable with a 10MHz bandwidth, and my high end plastic cable is rated at 6MHz. Second, the internal Bryston DAC only accepts files up to 24/96.

So the problem was the cable and the DAC.

2.) I was recently made aware of an additional problem with optical connections. Toshiba makes enough different TOSLINK Optical Modules to fill a bucket. It depends on the use of the component as to which module is selected. Since most consumer TOSLINK cables are limited to a 5 or 6 MHz. bandwidth, why use a module that exceeds that rate of transmission or reception. I'm not sure how accurate this is, but I am told that the modules used for consumer audio are different than the modules for pro audio components. In addition, I have no idea which modules are used on computers. Just check out this PDF file from Toshiba:
http://www.toshiba.com/taec/components/ProdLineGuide/toslink.pdf

Question: For high sample rate files (above 24/96) you can change to one of the new Quartz glass cables. However, when it comes to modules, how do we know if any of the 24/192 files can even be played with TOSLINK ?

3.) Of course the above leads me directly to the DACs. Not all the DACs mentioned on this forum process the input of 24/192.

Questions: However, for the units that do, will their TOSLINK module accept those file rates? Would the manufacturer be able to answer that question? Which computers will transmit the higher sample rate files? Then again, the optical modules may not be a factor, just the cable.

4.) Let us assume that the optical modules can be a factor. That leaves us with SPDIF/Coax, USB, Fire Wire, AES, and BADA (on the Berkeley), and now talk of adding Ethernet inputs on DACs. Can the situation become more complex?

More Questions:

USB: Just since I joined this forum USB has started to appear on DACs. The only limitation is the length of cable. However, I still do not see any specs on which version of USB. It must be at least 1.1, or does the connection include 2, or possibly the new version 3? Just asking, because after the RMAF the discussions about DAC makers dropping Fire Wire in favor of USB was because IEEE was slow releasing the new versions, but USB had released their version first. Seems like a lame answer if the new USB release is not being employed.

Fire Wire: Well it seems Fire Wire has been eliminated from a number of future DACs. That is the understanding I receive from discussions on this forum. But who really knows? Perhaps we are all making too many assumptions?

AES: I have heard conflicting stories about AES/EBU digital cable using XLR connections. Some say it can be used up to 100 meters, other say 10 meters. It would be be really accommodating for my home system because my computer is placed 28 feet away from my audio components (which means I would need to run cables a good 50' long). So, can I run AES cable from a DAC to my pre amplifier with that length without any analog signal problems?

5.) I just purchased a new Mac Pro for my home and was planning on using the Lynx AES16e sound card. Now I have a foolish question: why am I purchasing a sound card with DB 25 pin I/O connectors when the DAC digital input is only (what I assume to be) a single three pin XLR connection? Oh, and just where would I obtain such a cable?

Thanks for your help,

Daphne

Hi Daphne:

A quick response to a few of your questions:

a. Current maximum resolution I have found for optical toslink out of a Mac and any USB is 24/96. Those on the CA site seem to come to the same conclusion:

http://www.computeraudiophile.com/apple_drops_firewire_400_macbook_pro

b. Attached is the link to an optical cable outfit that seems to have a good quality toslink-to-toslink mini cable that has much higher bandwidth than yours (claimed bandwidth at 50 meters is = 40MHz). I have not used their product but they sell long lengths at competitive prices (they seem to be the only player who sells long toslink-to toslink mini cables so technically their prices can not be competitive, rather reasonable, e.g. 10m, for about $50, but we digress):

http://www.lifatec.com/toslink4.html

As Lifatec is both a quartz and "plastic" cable shop I have asked them if they plan on producing an even higher spec Quartz cable.

I am not a fan of "investing" in cables but have found Toslink cables make a huge difference. I currently have a $10, 5m Calrad toslink cable with a $2 Calrad toslink-to-toslink mini adapter and it sounds lousy. I used a 1m cable from Calrad with the same adapter and noticed a huge improvement in sound quality (I just can't sit on top of my stereo gear!). I am awaiting Chris' new budget super-system recommendations this week but will likely upgrade to a long Lifatec cable

c. I have found that the technical specifications relating to digital audio inputs and outputs is difficult to unearth. I think contacting the manufacturer is a good idea. Last week I had a question on the B100-SST (ironically!) and James Tanner from Bryston provided a pretty detailed response in record time, so you might want to drop him a line regarding your Bryston:

http://www.computeraudiophile.com/node/1042

First USB:

There are a number of different implementations ranging from custom, requiring programming and S/W drivers, to chips from TI that are plug-and-play. As you would expect, the more custom it gets, the better it sounds. Most manufacturers for their first try used the PCM270X parts from TI, which sound quite bad, even with low-jitter clocks driving them. Some are changing to the TAS1020, which requires some programming, and others are using completely custom implementations with embedded processors. Most high-end audio USB interfaces are limited to 24/96, but this wil change to 24/192 in 2009. These programmed interfaces can sound excellent, but only if the implementation is good and the clock used has low jitter. It's like all other designs in audio circuits: Just choosing good parts does not guarantee good sound quality.

As for long cables:

AES, Toslink and even USB cables should be as short as possible unless a good reclocker is used after them. For 28 to 50 foot runs I would recommend WiFi using the Sonos or AirPort Express (iTunes). These devices have high levels of jitter, so if you want them to compete with your good transport/CDP, then they will need reclocking. The data itself is delivered bit-perfect and error-free with these networked devices. It is only the jitter that needs to be addressed.

In order to have accurately reproduced digital audio, the D/A converter needs two things:

1) accurate data ( this is easy - errors are rare)
2) accurate timing - this is the difficult one

Jitter, which is inaccuracies in timing, can be reduced by several methods, including asynchronous reclocking, synchronous reclocking, ASRC or Asynchronous Sample Rate Conversion. Each of these has a different level of effectiveness, the synchronous reclocking being the most effective.

Steve N.
Empirical Audio

I did a bunch of internet research on jitter when I was putting my computer audio solution together, in an attempt to discover what to listen for, how serious its effects are and how to reduce/eliminate it. Unsurprisingly, I found a wide variety of viewpoints from users, engineers (design and recording) and scientists. The conclusions I came to, FWIW, are that while there is no doubt that jitter exists, is measurable, and can be reduced, there is plenty of doubt about whether or not humans can hear it.

I came to these conclusions because quite a few pretty savvy people seemed to believe that the presence of jitter, even in very common mid-fi audio, is well below the threshold of human perception, because any and all descriptions of the effects of jitter I got from believers could have been attributed to almost anything in conventional audio, and because I couldn't find a single AB/X test in which listeners identified any difference with jitter reduction equipment in and out of the signal chain. I'm sure jitter solutions do what they say they do. I'm just not at all sure it's audible. Can anyone point to a blind listening test in which jitter reduction (all other things unchanged, including the analog output stages...) is heard consistently enough to beat the margin for error?

Tim

Audibility of jitter is like audibility of different cables. Most self-proclaimed audiophiles systems are not good enough to hear the benefits of either, so there will always be non-believers. In fact, most reviewers systems are not good enough either, even some of the editors of HIFI magazines!!

Most systems have high levels of sibilance and noise, so the benefits of removing jitter are not as great. In most systems though I can hear the difference with the right track. This was demonstrated recently at RMAF in a room where the turntable and the CD player sounded identical. The preamp was "homogenizing" the sound by adding noise and distortion.

This is why I'm really careful about who I send gear to for reviewing.

Steve N.
Empirical Audio

I don't doubt for a moment that some systems resolve better -- even resolve distortions better -- than others. If you're right, and audible jitter is so small that it cannot even be heard in most audiophile systems, reviewers' systems, even Hifi magazine editors' systems, then it's not that jitter is voodoo, it's simply that it's not an issue for almost any of us, and the practical answer to the question posed is "don't worry about it."

Still, the measurements I've seen are so small that I have to wonder how, even with the gear of the gods, humans can hear timing errors measured in picoseconds with ears working in 4/4 time :). Can you tell me what kind of system might reveal this jitter? Do you know of any blind listening tests with highly resolving systems in which the presence/absence of jitter was actually detected often enough to exceed the margin for error?

Tim

Hi Steelman:

Thanks for the information on optical and USB connections.

By the way, I am familiar with Lifatec and use several of their high quality plastic cables. They are very reasonably priced, well made, thin, and light weight, which makes a big difference when connecting to a Macbook Pro. My only complaint is the TOSLINK mini jack used by Apple is the number one least reliable optical connection made. I have no idea why the consumer A/V and computer industries use such poor connectors. A good 90% of the optical fiber market (the telecommunications, instrumentation, and medical industries) all use a variety of locking connectors.

Regarding the Lifatec claim of 40MHz bandwidth. I believe that test was done with an industrial transmitter and receiver, not the type of low level red LED used with A/V TOSLINK applications. Audio optical cables is a small fraction of Lifatec's business. For consumer audio the bandwidth of most plastic cables will max out around 5MHz to 6MHz, but Lifatec uses the very best plastic fibers made and for shorter lengths their bandwidth is definitely more.

"I have found that the technical specifications relating to digital audio inputs and outputs is difficult to unearth. I think contacting the manufacturer is a good idea."

So true Steelman! I keep thinking about the frustration of assembling all the components only to discover that the data is being shaved, or creating major jitter, due to the connections I select. You could say I'm just performing proper due diligence prior to rushing out and making purchases, but once I started to examine the technical specifications I found a void. Naturally I started to ask questions only to discover there are few answers beyond what I already know.

"Last week I had a question on the B100-SST (ironically!) and James Tanner from Bryston provided a pretty detailed response in record time, so you might want to drop him a line regarding your Bryston:"

I saw your post about Bryston, then had to work. By the time I had a chance to respond, I was truly surprised to see a response from Bryston's, James Tanner. Come on, Steelman, what are the odds of asking a simple question and receiving a direct, concise reply from the manufacturer? The only thing I can add is, if you are looking for an integrated amp with average power you cannot go wrong with the B100 or the B100DA. I am currently looking at DACs for my home system and seriously considering the BDA-1 as a possibility.

Daphne

So if there are no equipment standards, and claimed benefits are based
on minimum standards of said equipment, undefined as they are ... where are we?

"To hear the jitter reduction, you'll need a good recording ( good luck with that,
they are few and far between ), brand/model XXX source, brand/model XXX amplification, and brand XXX speakers are minimum requirement." ? Oops, forgot the cables. ;-)

"To fully enjoy your BMW, we recommend removing the speed limiter,
a closed track, and a $3000 set of racing tires." ? Hey, couldn't hurt! ;-)

It's an issue of perspective, and all too often in this audio business,
perspective is woefully absent, sometimes to the point of being nonsensical.

Ha! nonsensical and high-end audio in the same context? No, it can't be true.

I like the post mpmct.

Daphne, I happen to believe that the audio out from the Macbook Pro is excellent. I sometimes use a Monster mini-to-RCA left/right audio cable from the AirPort Express Stereo Connection Kit to connect my Macbook Pro directly to my amplifier without using a preamp. The result is clarity and detail beyond belief. In many ways this is similar to the super sonics espoused by headphone aficionados that just use a good CD player, a good headphone amp and excellent headphones like the AKG 701 or Grado RS1. They firmly believe that music feed from the CD to your ears is simple, direct, pure and revealing.

Maybe my post was a little harsh, but sometimes it's a challenge
to not beg for a little bit of balance.

The recording, the speakers, the room ( unless you have speakers/system that largely takes the room out of the equation ) -- those three things are so huge,
and everything else is so insignificant, relatively speaking.

EG: a $3000 dac feeding $1000 speakers is just craziness. Been there! :-)

And A/B X testing is vital. Totally second Tim's post in that regard.
If no one with mortal ears can *hear* how good it sounds ... well? ;-)

"If no one with mortal ears can *hear* how good it sounds ... well? ;-)"

They don't even have to be mortals. I'd invite the designer/manufacturer/marketer of the jitter reduction box o get a roomful of trained, experienced mastering engineers - not the ones that brickwall everything, but the good ones. Or audiophiles. His own customers, even. I'd let him pick the reference system, and I don't care how expensive it is (though we might want to test a few real world systems to see where the point of no returns is....). But I insist on scientifically conducted blind listening tests, or insist that my skepticism is totally justified. Show me that people can hear something. Anything. If you can't or won't do that and your answer to those who can't hear what you hear is always "your system is not resolving enough," even when that system is "audiophile," even when it is a pro reviewer's reference system, well, I'm sorry, but you're asking for skepticism, begging for disbelief. If someone cries "snake oil!" it is not an empty attack or an insult. It is the closest thing to the truth that we mortals, with mortal ears and mortal gear, have.

Tim

This is first, a double blind test description, that I subsequently
sent to a colleague, who spent his career as a usability expert
at one of the large US telcos. First the test parameters, then
two notes from my colleague, now retired. Each separated with " ........."

................

"How to conduct a double blind test.

Buddy has just spent the equivalent of a trip to the Bahamas on a
pair of interconnect wires. You challenge him to a duel. You tell him
that your $10 cables sound the
same as his "Bahama" wonders. Not wanting to look like an idiot he
goes for the bait. Offer him $100 if he can identify which cable is
which. If he can't he pays you $100.The test is 20 trials of either
cable "A" or cable "B".

Instead of tossing a coin you can go a bit high tech. With your $7
pocket calculator you can use the random number generator. Any
number from 001 to 499 can be cable "A" and 501 to 999 is cable "B".
Should the number 500 appear you can toss it out. Have a third party
poke the calculator and jot down the sequence 1 to 20.

Now the test begins.
Both you and Buddy leave the room. The 3rd party connects according
to the chart A or B. 3rd party now leaves the room. From this point
on no one knows which cable is in the circuit. Do this 20 times.

There can be no communication whatsoever with the 3rd party! No one
has ever been able to determine which is which. By sheer guess
you'll get 10 out of 20. Statistically you must score 17 or so.
After all, if the expensive cable was so much better when sighted
surely you should be able to pick it out by just listening... Buddy
will come up with every excuse you can think of. You can remind him
that he heard the difference when the test was sighted. Now take his
$100 and split it with the 3rd party. Order pizza."

.........................

"Technically, the experimental design is a repeated measures design with 20
trials of measuring one variable that can take on two values: correct or
incorrect. If Buddy got 17 or more right out of 20, the probability of that
happening by chance is calculated with the formula for binomial probability:

http://faculty.vassar.edu/lowry/binomialX.html

With n=20 (trials), k=17 (correct choices), and p=0.5 (probability of a
correct choice by chance on a single trial):, the page shows the probability
(for a two-tailed test under "hypothesis testing") as 0.0026. This is
actually a more stringent test than is usually used. I would be satisfied
with k=15, which would happen by chance with a probability of 0.04, which is
close the traditional value of 0.05.

Actually, when I read the description of the experiment a second time, I saw
a number of flaws, even though the basic idea is correct:

1. The experiment as described is not as sensitive as it could be. People
are best able to disriminate between two choices when they are able to
compare them side-by-side. In the case of audio stimuli, this means hearing
them one after another. A number of protocols have been developed for doing
this. For example, there is the ABA test. In this test, you would hear a
piece of music played with Cable A, then a piece with Cable B, and then a
piece with A or B. The subject's task is to say whether the third case is
the same as the first one they heard or the second. This is not a preference
test, but simply a discrimination test, which is the most sensitive kind of
test: if you can't tell the difference between two cables the question of
which one you prefer doesn't arise.

2. Nothing is said about balancing the number of trials of A and B. The way
it's described, the trials could all have been with A if that's the way the
calculator happened to generate random numbers. That would not be good.
There really should be equal numbers of A and B, tested in random order.

3. Nothing is said about what sounds are played with each cable. These
should be randomized or balanced. In randomization, you would start with 20
selections of music, for example, and randomly assign them to be used with
Cable A or B. In a balanced design, you would start with 10 selections and
use each equally often with Cable A and B. This would probably not be a good
choice because people might judge differently when they heard the second
repetition of each selection during the 20 trials. This would introduce
extraneous noise into the experiment.

4. The sample of people used in the experiment is only one. A better test of
cables would use a much larger sample of people so we could generalize the
results."

.............

Further ...

"I've always believed that much high-end audiophile equipment differentiates
itself in a region of the audio spectrum that the ear can't even detect or
with a time-precision that the brain doesn't process. In other words, a lot
of it is hokum designed mainly to separate audiophiles from their wallets.

( Name redacted ), a guy who used to work at ( redacted, large US telco ) , did a project related to this. He wanted to find out whether there was any audible difference between
records and CDs. At the time, audiophiles were claiming that CDs sounded
worse than records. So ( name redacted ) got together a bunch of music (of all types)
that had both CD and record versions (without re-mastering I believe). He
played them to subjects (college students) using an double-blind ABA
discrimination test, if I recall correctly. The result was that not one
subjects could reliably tell the difference between record and CD. If follows
that whatever preference they might have had for one over the other had
nothing to do with the way the music sounded."

Hi audiozorro:

I also am amazed with the quality audio from my Macbook Pro. I just did not expect it to perform so well.

I have used headphones with a small headphone amp for years. I've had various makes and models over the years, but I put them to use often. I would actually shop for CD players with headphones. I had this leather messenger bag which I would carry my headphones, headphone amp (some required an external power supply so I would also carry an extension cord), various short interconnect cables, and a CD wallet with my reference music. It was easy to move from one CD player to another in the retail stores. Also, a good pair of headphones can be very revealing, and they offer the same reference to compare players.

It was 2002 or 2003 when I purchased a Mark Levinson No. 390s CD player. This player could be connected directly to an amplifier and I used it with headphones often, and it was the second best sounding CD player I have owned to date.

I also like your idea with the Airport Express.

Daphne

Tfarney wrote:
"Still, the measurements I've seen are so small that I have to wonder how, even with the gear of the gods, humans can hear timing errors measured in picoseconds with ears working in 4/4 time :). Can you tell me what kind of system might reveal this jitter? Do you know of any blind listening tests with highly resolving systems in which the presence/absence of jitter was actually detected often enough to exceed the margin for error?"

With the right gear, the differences are more than obvious. I dont expect you to ever believe what I say, but I say it for the benefit of others. I have customers that swear that jitter is actually more important than DAC quality. Jitter and sample-rate are the two things that differentiate digital audio from analog audio. Sample-rate changes the detail, but does not change things like HF sibilance and listener fatigue. These things are primarily caused by jitter. They have been the bain of digital audio from the very beginning.

Systems that are completely modded and have low noise-floors easily reveal jitter. I have numerous customers with open-baffle speakers driven by minimalist tube equipment that can hear everything. No less than $10K preamps are required to get low-levels of noise and distortion IMO, and some of these even add too much noise.

This is one reason why I have launched my new Overdrive DAC. It replaces a $10K+ preamp and a $10K+ DAC with a single device (that can optionally be battery-powered) for as little as $2499. Most audiophiles only dream of $10K preamps, so they are never able to get to the level of real music. It still sounds like a stereo system. The Overdrive enables this dream to come true for them. A true giant-killer.

I put these noise sources in order of importance:

1) Jitter
2) D/A accuracy and distortion
3) pramp noise and distortion

Dont underestimate the ability of your brain to process audio and detect jitter. It's quite a remakable instrument, the brain.

Steve N.
Empirical Audio

Tim wrote:
"They don't even have to be mortals. I'd invite the designer/manufacturer/marketer of the jitter reduction box o get a roomful of trained, experienced mastering engineers - not the ones that brickwall everything, but the good ones. Or audiophiles. His own customers, even. I'd let him pick the reference system, and I don't care how expensive it is (though we might want to test a few real world systems to see where the point of no returns is....). But I insist on scientifically conducted blind listening tests, or insist that my skepticism is totally justified. Show me that people can hear something. Anything."

Well, I consistently have done this at CES for about 7 years now, and even non-believers that enter my suite would leave as believers. Jitter demonstrations are easy and hearing the differences with a system such as the one I used at CES makes it easy. To my knowledge I was the ONLY exhibitor doing jitter demonstrations. I was also the first to use computer driven audio at CES, back in 2002 I believe.

My system at home is even better. Those that have been in my audio studio always leave enlightened. If I could convice a bunch of audio engineers to come and have a listen I would be delighted. They are the worst kind of critics and non-believers, studio engineers. 99% of them dont believe that cables make a difference. Been there done that.

Steve N.
Empirical Audio

If I was a sceptic, with a point to prove, invited to join a blind listening test, I would simply refuse to hear a difference. Simple. I am not a fan of blind listening for this very reason. The sceptic can refuse to accept any technology unless proven by a blind test but the results of a blind test can be easily invalidated by the listener simply declaring not to hear a difference. But there again, that just might be me being overly cynical!

On the subject of audio outputs the best one is the one that sounds the best to you. That is trite, I know, but for me what makes this a wonderful hobby is the journey! The end result is always just the starting point for the next journey. If it was simple and all black and white it would be no fun at all!

My son has just upgraded his system with a gazillion quids worth of new stuff. I listened to it last weekend and my overall impression was that it was 'dark'. I thought it lacked top-end sparkle. That was the only thing it lacked, mind you, everything else was absolutely wonderful! My son was a tad miffed because I think he rather loved how 'smooth' it was!

All roads lead to Rome - it's the fact that there are so many of 'em that makes the journey interesting!

"With the right gear, the differences are more than obvious. I dont expect you to ever believe what I say, but I say it for the benefit of others. I have customers that swear that jitter is actually more important than DAC quality."

What is the 'right gear'? I've read that vague phrase enough times,
let's put some meat on that ghost.
*Everyone* has an opinion, this is not news.
What those random, unmeasured, opinions are worth? ...
just about that much. I maintain the the
Mississippi River is filled with Mountain Dew.
That's my opinion, and I have golden taste buds. Prove me wrong.

Bob ...
"Blind Listening
If I was a sceptic, with a point to prove, invited to join a blind listening test, I would simply refuse to hear a difference. Simple. I am not a fan of blind listening for this very reason. The sceptic can refuse to accept any technology unless proven by a blind test but the results of a blind test can be easily invalidated by the listener simply declaring not to hear a difference. But there again, that just might be me being overly cynical!"

Bob, the term is not "blind listening", and it's spelled: "skeptic".
All double-blind tests in recent history are videotaped, so that professionals
who do this work for a living can assess the possibility of
truculent, impudent, passive-aggressive, childish behavior,
and disqualify patients who are not interested in,
refuse to, or are unable to, behave like adults.
It does happen, as you might guess.

Steve, it's not that I don't believe you; it's not personal. Quite the contrary, I'm an advocate of the most impersonal, dispassionate evidence. The only thing I don't believe, and it's not distrust, but a matter of definition, I suspect, is that you've "done this" at shows for years. A true AB/X test is blind, is accompanied by no knowledge of the products being tested. The listeners just listen to samples and try to differentiate between them. If they cannot do so within the margin of error, it is a statistical fact that the difference does not exist. But it can't be blind at a show. There is more than enough pre-knowledge to introduce a lot of opportunity for psychological bias, something the audiophile world is ripe with.

It is beyond me why manufacturers of cables and other gear that has been the victim of such skepticism haven't conducted such tests and addressed the issue long ago.

And of course it is a moot point anyway. If the price of admission is a $10k preamp (and similar quality components throughout the chain), I'll never be able to join the club anyway. Hardly anyone will.

I am curious, though, how you can rate as you stated:

"I put these noise sources in order of importance:

1) Jitter
2) D/A accuracy and distortion
3) pramp noise and distortion"

If the $10k preamp is required to even hear the jitter, how can the jitter possibly be the most important source of noise?

Tim

Mac has Airport, Pc's are ok via Line-in to Amp, and very conveient. I enjoy my PC as a Music Server or Transport. Cheers!

Wow, you guys really love to jump onto double blind listening tests at every opportunity.

Anyway, just thought I would take this opportunity to point out the problems with listening tests.

When conducting an AB listening test, we are asking humans to make subjective decisions on what they hear so objective results can be recorded. This premise has the potential of containing a number of flaws.

When you have two variables very closely related, the mathematics of probability and statistics can come close to predicting the outcome in advance. Also, the longer the test is conducted with the same subjects the closer the final tabulated results will be with each person. Meaning everyone in the group will come very close to having the same results.

The listening tests in question are not comparing a bookshelf speaker with a floor standing speaker, or a shrill dome tweeter to a mellow ribbon tweeter and asking which sound the subjects prefer. It seems the only time double blind testing comes into question is when subjects are asked to distinguish very subtle differences in sound.

Many who conduct this test have a different method, from the size of the group to several groups, from a single selection of music replayed over and over, to many selections of different types of music. However, a major flaw in double blind listening tests is the duration of the test. The more the subjects are asked to make decisions, the more they become disinterested. The more they listen, the more accepting of the sound, the less of a difference can be distinguished. This is just one of many psychological factors that examiners should take into consideration prior to establishing the testing method.

Another flaw is the sighted test versus the blindfolded (or dark room test). Post interviews from blindfolded tests have revealed some interesting results. I recall reading how sighted people tend to fantasize when listening to music in the dark, therefore ignoring the listening tasks they were requested to perform.

One method that can be used is adding a third or fourth variable. Almost like a placebo effect. Produce a sound that is obviously and measurably better or worse, or both. This method allows the listener to associate A and B with C or D. Which of the two variables in question more closely associates with the better or worse sound? Now that method would make for interesting results.

Now the people on the opposite side of the fence, those who market dubious items which are reported to improve sound quality, actually want doubters to focus on simple double blind testing. They quickly seize the opportunity to point out all the test flaws, thus initiating the process of gaining converts by doing so. The old "being caught in your own trap" scenario. It's amazing how many converts become zealots.

I find it curious that a sampling of methods used to test the blind's ability to hear are not used in audio comparison tests with sighted people. I can only assume the arguments between the medical community and many psychologists who design, perform and interpret the tests have frightened everyone off. Recently, medical research has shown evidence that the seeing lobe of the brain in blind people increases with activity in their daily hearing tasks. As a blind person ages, beyond a young adult, the seeing lobe actually supplements the hearing lobe allowing for increased mental processing of the vast cacophony of sounds surrounding them, a measurable increase over sighted people.

Psychologists will concede that the blind have superior spacial hearing skills - the ability to distinguish the directional source of sounds. However, through various flawed group hearing tests conducted by academics, living in a world of publish or perish, they firmly believe the blind cannot hear and comprehend more details than sighted people. Even in the light of some very revealing scientific proof, many psychologists hold fast to their beliefs (especially those who have written generally accepted theories), and claim any new physical proof to be nothing more than pseudoscience.

This brings me to my blind friend Milton. He teaches music appreciation to blind children, but his specialty is teaching the blind how to navigate through a city environment, and how to communicate to the rare child born blind and mute. It is truly a thankless, low paying job, but Milton was born into a wealthy family and has no financial worries. His sister Sara and I have been close friends for 29 years, and we can recall Milton becoming a devout audiophile at the age of 16. Music and recorded books are his love in life, along with attracting one screwball girlfriend after another.

Milton has two blind friends who are also audiophiles. On numerous occasions they have offered their services for double blind listening tests. They have appealed to audio equipment manufacturers, speaker makers, and even audio cable manufacturers. But no takers. Even though they supply impressive resumes outlining their hearing skills, replies are polite, but hint they should pursue more fruitful endeavors. I have been witness to Milton's ability to hear details few sighted people can come close to matching. He can walk into a room, and with a fair degree of accuracy, describe the components and speakers being used. Absolutely amazing; right on the border of being a magic trick, but nothing more than memory skills.

Not all blind people possess extraordinary hearing. However, many have developed their hearing abilities out of the necessity of survival in the outside world. It is a slow process and one that medical science does not fully understand. On the other hand, Milton will be the first to describe the limitations of human hearing.

One reaches a certain point with their audio system where a 20%, or even a 10%, improvement will no longer be possible. Even prior to that point we have become accustomed to the sonic nuances of our systems. The more we listen, the more we naturally accept and like what we hear. Any small changes can be enjoyable, or an irritating distraction. I like Milton's opinion on audiophiles. To enjoy quality reproduction of music is admirable, but an ever ongoing attempt to wring that last elusive, less than 1% improvement, usually means one is becoming a Zealot, and zeal without knowledge is the sister of folly.

Anyway, I have often wondered what the results would be from a double blind listening test where the subjects asked to determine the sound of one cable over another were a group of blind people with enhanced hearing skills.

Daphne

Angus McKenzie, the noted UK reviewer, was blind. His speciality in print was radio reviewing, often commenting in astonishing detail about the inadequacies of a performance or part of the production chain. Interestingly, one of his final reviews was of a recently introduced DAB radio, which he felt brought the pleasure of radio listening back to him. And having said all that, I'm pretty sure he preferred the old school, good engineering is best approach and left the nuances of cables to younger, more volatile reviewers.

"Anyway, I have often wondered what the results would be from a double blind listening test where the subjects asked to determine the sound of one cable over another were a group of blind people with enhanced hearing skills."

What would it matter to anyone who didn't have 'enhanced hearing skills',
if they could never partake of the claimed enhanced hearing skills?
And following that logic: I have 'enhanced visual skills' versus my blind friends.
What possible difference would that make to any of my blind friends?
They'll probably never 'see' what I 'see'.

A/B X testing requires removal of all visuals, of course.
( Who among us are not momentarily impressed with
good-looking things, or good-looking men in fine Italian suits? )

I repeat: professionals who conduct blind tests for a living, don't
make up the parameters as they go along. I've left enough information
about the process in the long previous post. ( Of course issues of fatigue
are accounted for in professional tests, and etc etc etc ).

The point in all of this is to point out that random opinions, from either
amateurs, or the golden-eared, are what they are: random opinions.
My local hi-end dealer, and friend, will absolutely not submit to any
blind testing of audio gear. And he also wouldn't step into the ring
with Muhammed Ali in the '70s I suspect. And he will not take me
at my word that the Mississippi flows Mountain Dew. But what does he know?

Sorry, Daphne, I didn't get through all that, but I don't think we're focused on nonsense, or even on double-blind testing, we're just discussing a way in which the audibility of something can be reliably tested. All of the caveats you put forth, some of which are valid and some of which do not appear to quite understand either the methodology of objectives of the testing, are addressed with properly conducted tests. The challenge is not conducting an AB/X test that accurately determines if a tweak (yes, they are typically subtleties being tested, there is no need to test the audibility of that which is not subtle) can be heard. I think that may be the key to most of your concerns. Long listening sessions, critical evaluation and subjective judgments are not required. We are only determining whether or not something can be heard at all.

Tim

I wont go down the AB/X rathole. It's like a lie-detector test. Not entirely reliable. And getting the RIGHT people (trained listeners) to come and sit still for hours is virtually impossibe. And then people still find flaws in your method or believe that it is a hoax. Not worth the time and effort unless maybe sponsered and overseen by a third-party such as Stereophile or TAS.

"If the $10k preamp is required to even hear the jitter, how can the jitter possibly be the most important source of noise?"

It's not required. The problem is that there are so many causes of sibilance in the typical system, the preamp only one of them, that many audiophiles dont get the FULL benefit of low jitter once they have it.

We are all at different levels of system perfection. If you are still tuning the sound with different cables for instance, you are not even close.

The reason that I put Jitter as #1 is because of customer feedback. This is what they are telling me, based on experience with MY jitter reducing products. Some of my customers also have world-class systems. They hear what I hear.

Steve N.
Empirical Audio

"professionals who conduct blind tests for a living, don't
make up the parameters as they go along."

These "professionals" however may be limited in the system quality that they are using, the room acoustics or even having good tracks to use. I have heard systems of some high-end reviewers, and so few of these are good enough and that I would never send anything to them for review.

I have read reports of AB/X jitter tests and they are consistently filled with flaws, particularly the ones published by AES. They dont use good enough systems and they dont use good enough tracks. In some cases, they dont even use trained listeners. In some cases, they use contrived jitter stimulus, not related to a music track at all, just random and uncorrelated. The results are totally useless and misleading IMO.

Steve N.
Empirical Audio

Hi Guys - I've been following this one and thinking about it quite a bit. I guess I just don't get the attraction to the "issue" of testing v. not testing etc... The whole thing is so far from kicking back, enjoying music, and the rest of our wonderful hobby. If something sounds good thats all that matters. Test all you want, don't test all you want it doesn't really matter to me. If I listen to a component and it sounds good I'm happy. If a manufacturer claims better sound because of lower jitter I'm cool with that. If I can hear it great, if I can't oh well.

OK, carry on :-)

"Hi Guys - I've been following this one and thinking about it quite a bit. I guess I just don't get the attraction to the "issue" of testing v. not testing etc... The whole thing is so far from kicking back, enjoying music, and the rest of our wonderful hobby. If something sounds good thats all that matters. Test all you want, don't test all you want it doesn't really matter to me. If I listen to a component and it sounds good I'm happy. If a manufacturer claims better sound because of lower jitter I'm cool with that. If I can hear it great, if I can't oh well."

Point well taken, Chris. What we hear is what matters. It all started because someone asked about the difference between the digital output of a Mac and what he would get if he ran it through something like the Trends UD-10. The Emperical Audio box was offered as a superior solution. I guess the question is, "is it?" Heck, I own a UD-10. It re-clocks. Is it better? I don't know that answer, either, but there is a way to determine whether or not a difference can be heard, thus the AB/X discussion. But you're right, in the end it amounts to what you hear, and evidently even one of the best jitter-reduction devices in the business (Emperical Audio's) can only be heard if you have equipment of such a quality as to exceed that of the reference systems of even audiophile magazine editors.

Which makes it a moot point for the overwhelming majority of us.

Tim

"... evidently even one of the best jitter-reduction devices in the business (Emperical Audio's) can only be heard if you have equipment of such a quality as to exceed that of the reference systems of even audiophile magazine editors.

Which makes it a moot point for the overwhelming majority of us."

Exactly.

"If something sounds good thats all that matters. Test all you want, don't test all you want it doesn't really matter to me. If I listen to a component and it sounds good I'm happy. If a manufacturer claims better sound because of lower jitter I'm cool with that. If I can hear it great, if I can't oh well."

Chris, on balance, what does the M.O. you described above cost you annually?
That's a serious question. Would it be unfair to think that you must be pretty flush
to have the attitude about audio equipment purchases that what you're written describes?
Do you just buy, try, and ship to a-gon or similar if you don't like, and eat the difference?
Or do you just live with it, call it one more secondary system, and put it in a back room?
Or, do you get 30 day money-back guarantees, shipping included? If yes, where do you shop please?

I want to be sure I understand what it seems to me you've implied: us consumers
should assume that reviews are 'true', or 'truth-y', and that any and
all manufacturers' claims are too, and that we should buy on that basis,
and expect to take the requisite a-gon downer if we just don't 'hear' it? Yes?

I'm not Chris, nor do I play him on TV, but I'd assume he has good dealers available to him where he lives, does not buy over the internet, and purchases only what sounds good to him. Fair enough. But I also happen to think it's fair to expect the manufacturers of very expensive equipment to show some evidence of the effectiveness of their products in the real world.

Tim

Tim - Pretty much right on. I don't buy anything without hearing it. If there isn't a way to work out some kind of listening session(s) with a product there's no way I'm dropping any cash on it. If I enjoyed reading specs on a nice quiet evening and looking at my audio components then I'd buy anything based on specs and marketing. But, I enjoy listening to music as we all do, so listening to a component is a must. Since there are a plethora of great components available today there is usually a way to work out something where you can listen to a product. Between manufacturers, brick & mortar dealers, and online dealers there should be a way to have a listen. If not, move on to the next component.

mpmct - I think we have a little misunderstanding and we probably agree.

I said "...If a manufacturer claims better sound because of lower jitter I'm cool with that. If I can hear it great, if I can't oh well."

I think you assumed I purchased the equipment already. This M.O. is really before I purchase anything. If I can hear the improvement, great I may purchase the component. If I can't hear it I then I really don't care and I move on, hence the Oh well comment. Manufacturers of all products make claims that not all of us agree with. Fortunately we aren't forced to purchase something that doesn't live up to the claims.

I hope you were being a bit factitious about assuming all reviews are true and we should believe manufacturers claims blindly. That is very far from my belief.

On a side note, there is something to be said for purchasing equipment that may not fully be appreciated at the time of purchase. This is because we usually seek to upgrade components one at a time. Buying a DAC with ultra low jitter may not benefit someone now, but if they upgrade the rest of their system later on then the DAC purchase was money well spent. As long as they have heard the claimed benefits in a dealer's system etc...

I have offered demos in the past. It's a can of worms. Had one customer keep one for 6 months. And offering money-back is a no-win also. The equipment takes about 30 days to fully break-in, so initial listening tests may cause lots of errant returns. No thanks.

If I had a problem selling my jitter reducing devices, then maybe I would resort to these kinds of things, but I've had a 3-5 month backlog for about 2 years now. Just watch Audiogon and see how long one of my devices sits there for sale, when they rarely come up for sale. Usually one day. I dont advertise either. All word of mouth and reviews.

Steve N.

Steve, given the extreme rarity of the system and the customer, as you've described them, who can actually benefit from your products, no, I don't imagine home demos worked well for Empirical Audio. You would have to pre-qualify both the ears of the customer and the resolution of his system to determine that he would actually be able to hear your product before ysubjective evaluation would be meaningful. An impossible challenge.

Fortunately, it seems that you're selling all you can build, that those rare audiophiles whose ears, taste and gear are up to the task of hearing Empirical Audio's products are finding you.

Congratulations.

Tim

Still, the measurements I've seen are so small that I have to wonder how, even with the gear of the gods, humans can hear timing errors measured in picoseconds with ears working in 4/4 time :).

Like all our senses out hearing has a upper and a lower threshold.
Claims about the lower threshold for jitter varies from 250 ns to 20 ps depending on the architecture of the DAC and the frequency.

Sound coming from one direction will reach the ear furthest away later than the closer ear. We can discern approximately 1/500 second time lag.
Hearing differences due to sample rate fluctuations in the order of 250 ns is of course far below the threshold of our hearing.
Having proven now that jitter is inaudible, next question is can we hear two tones of different frequencies at the same time if both of them are loud enough.
The answer is an obvious yes. I think this is the core of the jitter phenomenon. The way sample jitter is mapped into the analogue domain by a DAC creates artefacts like side bands. If the energy of these side band is big enough they become audible.

http://thewelltemperedcomputer.com/KB/BitPerfectJitter.htm

Thanks for the explanation. Do these artifacts manifest themselves in something that is measurable? Never mind. It seems that for me, it is a pretty moot point anyway. Those artifacts are, evidently, minor enough that I would have to first posses a system so resolving as to exceed the reference system demands of many audiophile journalists, then I'd need to train my ears to hear the difference between the presence and the absence of the artifacts. That just doesn't sound like a process that would facilitate the enjoyment of listening to my redbook cds of Live at the Village Vanguard or The Wild, The Innocent and the E Street Shuffle. :)

Tim

I think enthusiasts discussing jitter and it's likely impact on what they hear is a bit like car enthusiasts discussing rocker arm performance in race cars. Both are a tiny part of a jigsaw that gives us the result as music.

In reality jitter is only an issue when A to D 'ing or D to A 'ing, the first isn't usually of concern to hi fi enthusiasts and the latter is well catered for by modern DACs and their attendant circuitry. Therefore if it doesn't sound right, it's best to admit you don't know why.

As far as digital outputs go, they are all as good as one and other for hi fi provided the DAC is designed to cope with the slightly different artefacts each produce.

The optical output looks to become the International standard, it's certainly the most popular now, although people worry that 24/96 is its highest sample rate. We don't think that anyone need worry about it because we've done a series of tests and not been able to hear any difference, so we've involved others by asking them to switch between options using an IR handset and telling us what they hear. These were a cross section of pros, enthusiasts and just passers by, who often hear as much as anyone. After all we're all expert listeners, though we may not be experts at interpreting what we're hearing.

Our experiment was switching the sample rate on an experimental DAC while listening to music. We were able to go from 32 kHz and 16 kHz bandwidth to 24/196 and not hear a difference, which is as science dictates it should be.

I don't doubt some will think we're all deaf, but I hope it re-assures others that the optical output is good enough for anything hi fi, provided a modern DAC that's been auditioned before purchase is used.

These days, IMHO there are far more problems in the rest of a system than in the source.

Ashley

Do these artifacts manifest themselves in something that is measurable?
If you would have bothered to use the link I provided, you can see that a FFT analyzer can measure it

just doesn't sound like a process that would facilitate the enjoyment of listening to my redbook cds of Live at the Village Vanguard?

You can choose between listening to music (recommended) and listening to equipment (some times needed).
Live at the Village Vanguard will always be involving regardless of the equipment because it is good music while boring music in high resz on the highest high end equipment will sound like what it is, boring.

I think enthusiasts discussing jitter and it's likely impact on what they hear is a bit like car enthusiasts discussing rocker arm performance in race cars. Both are a tiny part of a jigsaw that gives us the result as music.

Agreed, that’s the right perspective.

As far as digital outputs go, they are all as good as one and other for hi fi provided the DAC is designed to cope with the slightly different artefacts each produce.

I’m afraid that’s the point, a lot of DACs are build like analogue equipment, accept the input as it is. The modern ones recover the clock. I’m inclined to say that if you hear differences between bit identical source or given a source different cables or transmission protocols, it is a badly designed DAC

I don't doubt some will think we're all deaf, but I hope it re-assures others that the optical output is good enough for anything hi fi,

I know your AMD9.1 offers optical input (Toslink) only. As far as I could judge not to many PC brands offer optical out (Apple, Toshiba). Why not adding a USB? The modern DACs you are using must be able to cope with USB jitter.

Jitter manifests itself as an increase in distortion and or noise in the experiments we did.

We did make a USB only ADM9 but it didn't sell, presumably because the computer is not where the hi fi is. Most customers use Sonos, Apple AE, ATV, Mini or Laptop or PS3. Some use PCs but usually with a streaming device like SB or AE. Two opticals and an analogue covers it for our customers. The problem with USB is that it confines you to a computer in a fast changing market, optical digital gives you a wide choice of sources and I think this is what appeals.

Ashley

Tim wrote: "Fortunately, it seems that you're selling all you can build, that those rare audiophiles whose ears, taste and gear are up to the task of hearing Empirical Audio's products are finding you. "

That's right. It's like fine wine. If the $5 bottle makes you happy and you are not willing to develop a tase for the good stuff, then we are in a different league. If you are after the very best, we can help you.

I'm not interested in being #2 in this game.

Steve N.

"I’m afraid that’s the point, a lot of DACs are build like analogue equipment, accept the input as it is. The modern ones recover the clock. I’m inclined to say that if you hear differences between bit identical source or given a source different cables or transmission protocols, it is a badly designed DAC"

Well, then there are a LOT of badly designed DAC's out there. I have yet to hear one that is not sensitive to different cables, sources etc.., regardless of what the manufacturer might claim, and I've heard quite a few DAC's in my system. I have modded at least 15 different DAC's, including: Benchmark, Birdland, Electrocompaniet, Meridian, Bremen, Audio Note UK, dAck!1 and 2, BelCanto, Northstar, Assemblage, MSB, Musical Fidelity, EAD, Dodson, Perpetual Technologies and others. I believe I have a bit more experience than any of you on this one.

Steve N.
Empirical Audio

Ashley wrote: "We did make a USB only ADM9 but it didn't sell, presumably because the computer is not where the hi fi is."

Let me guess: you used a PCM270X USB interface and it supported a max sample rate of 48kHz. If so, then this is the problem. This chip is the bain of USB audio. There are other chips and designs available that sound good enough to challenge even the very best CD players and transports. Most manufacturers select the 270X for their first foray into USB. Bad choice IMO. The reviews reflect this. Several manufacturers have recently changed from the 270X to a good-sounding chip, including PSAudio and BelCanto.

I have OEMs that use my USB module in their products, including the new MusicStream DAC from UltraFi (first DAC was the iRoc). Now this is a good-sounding USB DAC.

I have to agree that computer audio is still in it's infancy, particularly at the high-end, but mass-marketed devices like the AirPort Express and Sonos are selling like hotcakes.

Steve N.
Empirical Audio

Steve, do you use a Mac or PC?

"That's right. It's like fine wine. If the $5 bottle makes you happy and you are not willing to develop a tase for the good stuff, then we are in a different league. If you are after the very best, we can help you.

I'm not interested in being #2 in this game.

Steve N."

No, my friend, the music is the wine. You make corks. Or perhaps an additive to the cork that creates the slightest change in the way it breathes. I'm happy for you that you're good at it, but it's not the wine, much less the difference between a $5 bottle and the good stuff.

Tim

On another note, corky wine is a serious issue :-)

I use mostly XP OS PC's. I like Toshiba laptops. I use Foobar 0.8.3 and unmap the device to bypass kmixer.

I ocassionally use Mac too.

Steve N.
Empirical Audio

"On another note, corky wine is a serious issue :-)"

Where are those rolling smiley faces when you need them?

Tim