Speaker cable blind listening tests

[hermanv]

If you use the human hearing model that we can only hear 20 to 20K, with monophonic JND parameters, then your statement is entirely correct.

You have missed one of my points..the criteria for judging the audibility of any effects is incorrect. You did not consider at all, the human hearing sensitivity to either ITD or IID, and it is these parameters which we humans use to discern the direction and distance of a sound source..

Many people report significant changes in the imaging qualities of their systems when trying various "audiophile" cables. My room doesn't allow for good imaging, so the selecton process for cables that I like, ends up being based on two criterea; one is clarity, and the second is any reduction of aritifacts in the region of 2KHz to 5KHz. The region where to me many audio sins are commited. It happens to be more or less the peak sensitivity of human hearing and the brain seems quite good at localizing sound origination direction in this frequency band. This may well be learned from locating predators and avoidance of being some creatures lunch, so imaging or sound localization is a good fit.

So, it is moot to declare even simplistic RLC parameters as not having an effect, as your model for what we hear is so inaccurate..Once the criteria for what we hear w/r to soundstage is firmly established, then simple RLC can be re-calculated for the possibility of exceeding the threshold of audibility.
To me there is no distinction when it comes to the RLC model, as I do not consider it as a simple frequency or impedance invariant descriptor. Again, the bulk of discussion here is attempting to fit the wrong human hearing model to the problem..You have fallen into the same boat, as you have also erred in this fashion. I like the fact that you seem to be attacking the issue scientifically, but you have not realized that the hearing model is not up to the task you are working on..

I am looking for an explanantion of what I and others hear. I don't think its the value of the cable capacitance even though it might be the capacitor (I mean the value seems too small to explain the change but the quality of the capacitor might). The problem in these discussions is that the discussion gets continuosly derailed by people who apparently have not heard the effect and therfore feel challenged to prove it doesn't exist. Since you can't prove a negative they turn this on its head and demand proof of the existence of sound quality changes with cables. Irritating, because a search for some kind of proof is exactly the intent. I would like to find a scientific proof and hope some discussions will help lead to one. Many people have measured cables very carefully, the results are published all over the net but the numerical values are quite small (yes, I know so is the effect) but the effects of these reactances are linear whereas what I hear sure sounds like fewer or more products a much more complex interaction requiring the existence of non-linearities. Where in a traditional cable are these to be found? I also find it amusing that one should deny one self the benefits of a superior delivery mechanism when the reason for the superiority is not documented

Peltier and Seebeck effects are not symmetric..so, from physics, if this reaches a high enough level of conversion, there should be some signal non-linearities caused by non linear heat loss... Peltier defines the heat produced as proportional to the current, while Seebeck has the voltage proportional to the temp difference..for high impedance signal runs, the heat across the junction will be miniscule, therefore the seebeck conversion will be very very small, while peltier heat generation will be maximized due to little heating and small thermal gradient...

Again, both effects are incredibly small, and subsequent non linear loss is well below the threshold of detection for audio rate signals...not zero, but many orders of magnitude below the equipment.

Diodes at grain boundaries...hmmm..so far, the test guys here (I'm part of that group) haven't seen any such effects, but we only go up to 30 Kiloamps and look at the microvolt level. As such, only a upper limit of effect can be spoken of..I do feel, however, that this span of range is a tad beyond "high end audio"..

If you wish to discuss "micro-diodes", you have to first get around the fact that grain boundary interfaces are simply lattice discontinuities and the dimensions of the discontinuities are well within the mean free path of the "electrons". IOW, they really don't see the boundaries per se, but will only suffer a relatively minor increase in the number of collisions as a result of lattice defects. Given that at 300K, the mean free path for copper is about 6 orders of magnitude below the grain size, this incremental increase in resistive loss is, shall we say, of little concern?

The effects of the diodes I am visualizing would show up only at micro or nano Amps, the other end of the scale from where youi appear to be instrumenting. I am not trying to build a case that this must be the effect, I am tryng to build a case that some kind of non-linear behaviour would end up as a tidy explanation for what I hear. At least for the cables I like best, complex musical passages show the greatest improvement in resolution or clarity. Once again I am driven towards an intermodulation or product generation explanation.

You have calculated what simple RLC does to the signal, and then applied that result to the current outlandishly incorrect model of human hearing, and decided there was more to "cables" than can be explained using the RLC metrics..

You are guilty of trying to fit your square peg of RLC calculations, into the round hole of human hearing capabilities..ok, I was too subtle...shoot me...

Cable A sounds like this, cable B sounds like that; some people have described the changes as changes in frequency response. The cables I preffer certainly un-mask sounds at the frequency extremes that were inaudible prior to the cable change. This could easily lead you to conclude that the frequency response had changed with the cable change. Going back to those people on the web who have carefully measured cables in a way that highlights RLC effects there is indeed a change in frequency response and the degree of change is the same magnatude range of calculated effects, for example a couple of hundred picoFarads at 10Khz causing a change of a few hudredths of a dB. It is small and linear. I have owned equipment with tone controls there is no comparison of the effect of adjusting a tone control knob and substituting a "good" cable for a "poor" one.

This discussion is not over, not by a long shot...I hope we can continue this, I have good feelings about this..

Thank you.
Cheers, John

I am open to any number of explanations, some feel that is is a human hearing learning curve, curious, because to learn to hear a difference, a difference needs to first be present.

I hear a change, others hear a change. A whole industry has sprung up to service the demand. Certainly some of them are selling snake oil, this is true for most any industry. When I try cables that magazine reviewers say are worth a listen, I find them superior to the garden variety plain everyday consumer interconnects and to zip cord even when the zip cord is an absurdly large guage. Friends and neighbors that are not audiophiles easily hear what I hear, we agree on the effect but do have serious problems with the descriptive language. Human language evolved in an environment absent of electronic aritifacts and our everyday language is poor at describing electronic errors of sound reproduction.

[jneutron]

This may well be learned from locating predators and avoidance of being some creatures lunch, so imaging or sound localization is a good fit.

Agreed. And nothing in the research delves heavily into imaging and localization.

I am looking for an explanantion of what I and others hear.

A good thing.

The problem in these discussions is that the discussion gets continuosly derailed by people who apparently have not heard the effect and therfore feel challenged to prove it doesn't exist.

Perhaps discussions with others..but not me. I have not listened for any difference, having no good system upon which to draw, however, my perception of one is of no concern to the discussions.

I would like to find a scientific proof and hope some discussions will help lead to one.

Concur.

The effects of the diodes I am visualizing would show up only at micro or nano Amps, the other end of the scale from where youi appear to be instrumenting.

A diode is a device with a non-linear I/V curve..as such, it easily spotted, even at the microamp level. I have seen none, however, I personally at a previous job, only tested down to the fempto and atto-amp level...perhaps you need to look lower??
Measurement for "micro-diodes" has failed miserably to find any. Initial positive findings have been shown to be in error...this being due to the failure to baseline the equipment capabilities..what was seen was far below the inherent noise in the equipment, and inaccurately classed as distortion artifacts rather that what they were, equipment artifacts.

I am not trying to build a case that this must be the effect, I am tryng to build a case that some kind of non-linear behaviour would end up as a tidy explanation for what I hear. At least for the cables I like best, complex musical passages show the greatest improvement in resolution or clarity. Once again I am driven towards an intermodulation or product generation explanation.

Since you applied the incorrect human hearing model, you have bypassed the most obvious solution, that being the fact that your metric is incorrect.

You should be driven towards the simplest explanation..not ones which require re-writing physics. Non linearities are definitely way down on the plausibility scale..

I am open to any number of explanations, some feel that is is a human hearing learning curve, curious, because to learn to hear a difference, a difference needs to first be present.

We in fact, need time to re-aquire our sense of imaging when the cues we use for that are changed..there is a "learning curve", more aptly referred to as "settling time".
Two speakers present us with three images for any virtual image in the soundfield..

1. The desired one, which is the result of the sounds reaching the desired ear..left to left, right to right.
2. The right sideband image, the result of the right speaker finally making it to the right ear, providing the temporally accurate image AT the right speaker.
3. The left sideband image.

Because the right and left sideband images correlate absolutely with the desired virtual image in waveform, the human brain learns to ignore the side images, leaving the desired one. However, this is not what occurs in nature..in nature, only the desired virtual image is there, as that is actually where the source is..no sideband images.

When a change is made to your system, any change, if it alters ITD, IID, or placement location relative to the ears, it takes a while for the human mind to re-aquire the ability to "ignore" the sidebands.. this is the settling time..

And, it is not measured using instruments of any kind...

We localize based on ear to ear time and amplitude differences at first order. The time differences are in the low microsecond range, and the amplitude is in the .05 dB range. These are the kinds of differences that are the metrics for localization..and the ones you should be using.

but do have serious problems with the descriptive language. Human language evolved in an environment absent of electronic aritifacts and our everyday language is poor at describing electronic errors of sound reproduction.

Problems with descriptive language????you are a master of understatement.

Cheers, John

[hermanv]

We in fact, need time to re-aquire our sense of imaging when the cues we use for that are changed..there is a "learning curve", more aptly referred to as "settling time".
Two speakers present us with three images for any virtual image in the soundfield..

1. The desired one, which is the result of the sounds reaching the desired ear..left to left, right to right.
2. The right sideband image, the result of the right speaker finally making it to the right ear, providing the temporally accurate image AT the right speaker.
3. The left sideband image.

Because the right and left sideband images correlate absolutely with the desired virtual image in waveform, the human brain learns to ignore the side images, leaving the desired one. However, this is not what occurs in nature..in nature, only the desired virtual image is there, as that is actually where the source is..no sideband images.

When a change is made to your system, any change, if it alters ITD, IID, or placement location relative to the ears, it takes a while for the human mind to re-aquire the ability to "ignore" the sidebands.. this is the settling time..

And, it is not measured using instruments of any kind...

We localize based on ear to ear time and amplitude differences at first order. The time differences are in the low microsecond range, and the amplitude is in the .05 dB range. These are the kinds of differences that are the metrics for localization..and the ones you should be using.

I do not have a problem with using imaging to help ascertain the "cable effect". But I fear it may be a diversion. I am developing a new speaker with a fellow engineer and audiophile. Becasue of costs we built only one prototype and used a passive sum of right and left channels for our "voicing" tests.

As we improved the speaker design with better drivers and a series of crossover improvements we reached a plateau. We had both heard better sound from speakers we couldn't afford (the primary reason we decided to build our own). So we started updgrading the electronics. We bought a better outboard D to A for the CD player, built an all passive pre-amp using so called "high end" passive components and purchased a better (more expensive highly regarded) power amplifier. So during testing we naturally experimented with cabling. Once again a superior cable was quite easy to find in spite of the system already making use of a highly rated and regarded product. Having done all this we exposed another layer or two of the speaker design onion, and our design now competes favorably with speakers that retail for $20,000 and up.

But (drum roll please) all of this was in monophonic. So while imaging may well be improved by more accurate time alignment, better group delay or more equalized cable loss, a good cable just plain improves the sound quality even of a mono channel.

As an engineer I knew all along that "it's just wire". I was stongly pre-disposed to find no such effect but it was easy to hear. So native curiousity is a strong motivator for finding an explanation.

There are two other good reasons to work out why:
1. Once understood, are there other things that can be done to further minimize the negative effects on sound quality?
2. The very best cables have astronomical prices, up to thousands per meter. If the effect is well understood, can you now make a truly good cable for 5 bucks?

Current "high end" cable manufacturer throw all kinds of "magic" into the mix. Ultra pure metals either copper or silver, exotic wire forming techniques (continuos casting, annealing, drawn through diamond dies, etc.) and ultra exotic connector materials or platings. With a poor understanding of the effect how much of this expense contributes to the desired result?

There is a third realated reason to understand the effect of cables on sound quality, currently no realiable means exists of using test equipment to predict which audio products sound best. There was a paradox a few years back when the introduction of Op-amp circuits and high degrees of feedback allowed manufacturers to acheive distortion levels 10 or a 100 times better than previous designs. The resulting amplifiers all sounded awful but measured very well. A whole series of theories quickly sprang up to explain the discrepancy. The remaining two SID and TID (slewing induced distortion and transient induced distortion) were found valid but still rather poor indicators of the end result visa vie how good a given piece of electronic equipment will sound.

I need to stop for today and earn my keep.

[jneutron]

I do not have a problem with using imaging to help ascertain the "cable effect". But I fear it may be a diversion. I am developing a new speaker with a fellow engineer and audiophile. Becasue of costs we built only one prototype and used a passive sum of right and left channels for our "voicing" tests.

As we improved the speaker design with better drivers and a series of crossover improvements we reached a plateau. We had both heard better sound from speakers we couldn't afford (the primary reason we decided to build our own). So we started updgrading the electronics. We bought a better outboard D to A for the CD player, built an all passive pre-amp using so called "high end" passive components and purchased a better (more expensive highly regarded) power amplifier. So during testing we naturally experimented with cabling. Once again a superior cable was quite easy to find in spite of the system already making use of a highly rated and regarded product.

But (drum roll please) all of this was in monophonic. So while imaging may well be improved by more accurate time alignment, better group delay or more equalized cable loss, a good cable just plain improves the sound quality even of a mono channel.

As an engineer I knew all along that "it's just wire". I was stongly pre-disposed to find no such effect but it was easy to hear. So native curiousity is a strong motivator for finding an explanation.

Make two identical low inductance loads with low b dot error...keep them under a nanohenry, and make em 4 ohms.

Then do a side by side differential with speaker wires of various constructions.

You will find lots of differences.

Then look at why two or three drivers would have big issues with wavefront reconstruction at and near the crossover points. Youze is makin a phased array radar....the very low cable reactances cause the crossover points to scan based on frequency...doesn't take much in the way of inter-driver delay to cause it..I have not constructed a model for vertical localization, I don't know where to start on that...just horizontal..

Make the speaker cable as low R as possible, with Z equal to the load, and the lowest effective dielectric constant. It's trivial to get DC effective down to 2.7, I can get that down to 1.05 with some creative "fun"...that will be the best..everything else is making whoopie with the sound..

Compare the RLC for all the cables you tried. You'll find correlation..

With a poor understanding of the effect how much of this expense contributes to the desired result?

Until it is engineered properly, it's nothing but WAG's. And some drive 911's as a result.

Cheers, John

PS..do not fear "diversions". Test them for accuracy of modelling..

[FLZapped]

The effects of the diodes I am visualizing would show up only at micro or nano Amps, the other end of the scale from where youi appear to be instrumenting. I am not trying to build a case that this must be the effect, I am tryng to build a case that some kind of non-linear behaviour would end up as a tidy explanation for what I hear. At least for the cables I like best, complex musical passages show the greatest improvement in resolution or clarity. Once again I am driven towards an intermodulation or product generation explanation.

Visualize this: Commercial comunication repeater systems are licensed up to 350 watts of transmitter power. At the same time, they are running a receiver with a sensitivity anywhere from -113 to -119dBm(0.5 to 0.25uV).

Converting to common parameters we find that 350 watts is +55dBm. This means a differential of 168 to 174dB. So if this noise exists, it is greater than 174 dB down from the generator.

-Bruce

[musicoverall]

Prove it.



Prove any such thing actually exists.



Prove it.



Yawn......prove it.

-Bruce
(Astounded at just how idiot final statement is)

Never fear, Bruce. It's not just your final statement that's "idiot".

Seriously, you remind me of the rest of my class way back when I was in the second grade - a few centuries ago. "Prove it, prove it, prove it". Hmmm... you want ME to prove something to YOU to allay YOUR skepticism? Sure, Bruce, I'll get right on that - because I care SO much!

[FLZapped]

Never fear, Bruce. It's not just your final statement that's "idiot".

I see I was having a good week for typos.

Seriously, you remind me of the rest of my class way back when I was in the second grade - a few centuries ago. "Prove it, prove it, prove it". Hmmm... you want ME to prove something to YOU to allay YOUR skepticism?

You keep making claims, one after another, all you have to do is fill them up so they aren't empty.

Sure, Bruce, I'll get right on that - because I care SO much!

That now puts you in the realm of a troll.

-Bruce

[musicoverall]

I see I was having a good week for typos.


-Bruce

Not to mention trolling.

As far as my making claims, it's ok if you think they're empty... or if you think they're claims at all. My advice to newbies is to listen for themselves. If that's a claim, I'm guilty. And of course my final point was that I couldn't possibly be expected to solve your dilemma of skepticism. I think it's been pointed out many times that no amount of proof will satisfy the naysayers, short of some engineer/scientist type that you know and respect coming out of the woodwork with new measurements. I took some blind cable tests, as evidenced by this thread, and I've done what I needed to do for my own skepticism. I now have the system I want and each component, including cables, adds its own sonic signature. Believe it or not, as you prefer.

[bjornb17]

Come on you two children, break up the fight before i tell the principle!

[krabapple]

It sort of unclear if the pool exists. The author says he has contributed $200 and others would contribute. He also says that he'll pay $1,000 if you go to England , I didn't see an actual address, bank account nor a enforcable contract.

ps. He is asking for +/-0.1dB, sort of reasonable where +/-0.01 dB is not. It is possible that two identical cables would not pass the +/-0.01dB hurdle especially if they are hand made. It also unclear where on could find an instrument that was 0.01 dB accurate. The best I've used has a resolution of 0.01 DB but nowhere near that accuracy (that's about one tenth of 1%; if you read the fine print 2% is common 1% rare for RMS meters).

It's *k*rabapple, and yes, 0.01 was a typo and would be totally unreasonable -- 0.1 dB level matching is what I should have written. Sorry about that.

According to his rules, one could easily cheat. He checks at 100, 1000 and 10,000Hz. A notch filter at 200-800 Hz would pass his rules yet any fool could hear it was there (you'd put it in one of those little boxes MIT cables seem to like so much).

Well, of course any cables that *passed* would be subjected to a test for just such 'cheating'. No one expects to pay out money for a cable that is acting as an equalizer.
The stipulation is that two cables that measure the same, will sound the same.

This offer may have been made in the passion of cable discussions, it seems a little questionable.

You are free to clarify as many details about it as you like, on RAHE.

[krabapple]

After reading krabapple's link to re.audio.high-end, it looks to me like they are saying they can make a zip cord cord sound like any audiophile cable by altering the zip cord (e.g., "add a few passive components to some zip cord to achieve the same FR"). It isn't clear whether they also would want to be able to alter the audiophile cable.

Requiring you to identify the cable correctly 15 times out of 20 to win the money seems to set the p value bar a little high(i.e., 0.02). Usually in hypothesis testing a p value of 0.05 is chosen, which I believe would be 14 out of 20. It's not much difference, but I guess they want an edge in case you get very lucky or hear a difference in the cables.

In science, you want p<0.05, not p=0.05. In any case 14/20 is p=0.058 , which is greater than 0.05. 16/20 would be the minimum -- and even then such a score would arguably justify re-testing (as would 14/20).

Without seeing a full description of their rules, I can't comment further on the cable challenge. Before participating in such a test, however, I would want everything in writing and in the form of an enforceable contract. Otherwise, you may have a difficult time collecting the money if you win.

It's hard to beat someone at their game.

Feel free to post to RAHE to arrange the details of your test, which can of couse be written up when conditions are agreed on. The only 'game' is a the application of standard scientific controls to a listening test.

[krabapple]

Revisiting this thread reminds me to ask, where's jneutron these days? I was very interested in his rigorous investigations of the possible audible differences in cables (even though I suspect he is chasing a wild goose) ...but I haven't seen hide nor hair of him here or on audioholics for months.

[Mash]

My ears were burning, Mr. musicoverall, so I stopped by to quench the fire.

I am glad, Mr. musicoverall, that you held your wire test and that you satisfied yourself while doing so. Your results seemed mildly positive about there being audible differences between your test wires, and I notice that the party-faithful here have given you mildly positive congratulations for your study. I feel that had your results been over-the-top wildly positive the party-faithful here would have given you wildly positive congratulations on your brilliant work. Alternatively, if you had found that there were ZERO differences between the wires in your test then the party-faithful would have told you to get new batteries for your hearing aid, while Estat parsed your report to death with witty or half-witty rebuttals following every parsing. So I think you showed wisdom in what you reported, especially given that your data sample was far too small with too few positives to be statistically positive.

My last visit was 8 March 2005. I spend essentially no time here now because I do not feel like engaging in endless discussions of how many angels can dance on a pin head. Foo-foo dust sermons are irrelevant to me. I have insufficient time now to take our boat out for cruises, and my sound system can reproduce a grand piano well enough without needing foo-foo dust or miracle wires. I just set up HDTV in our bedroom, including having a complete capability to up scale all DVD’s and also VCR tapes into HD. My next step is to add the Velodyne Servo-15 that I have on hand to the HDTV system to support the Magnepans. Time to move on.

I do notice that this little area has become, as some one had once predicted, a moribund ghetto. I suspect many others will ultimately become weary of those sermons to the choir so popular in the other areas……..

Hey, Skep- I saw you were recently here! Good to see that you are still alive! My next little hobby is electric powered R/C airplanes. I really enjoyed flying real airplanes. Did you know that for $600 (on sale) you could buy an electric-powered 72 inch wingspan Predator airplane that will fly for up to 1.5 hours and rise up to 8000 feet above sea level? This plane has an R/C radio with a one-mile range, and you can add a camera that will take, and radio back to you, pictures of that young honey practicing nude sunbathing down the street. Not that you ever would, of course, actually LOOK……Visit Draganfly.com. The $170 Multiplex Easystar is supposed to be the very best trainer for newbies and it is a really slick looking 54 inch wingspan bird. Search the web. This is one hobby where foo-foo dust will never help anyone; in the final result you will either be able to fly your airplane or you will NOT be able to fly your airplane.

[musicoverall]

my sound system can reproduce a grand piano well enough ...

If you're happy, I'm happy. Congrats on having a system that performs to your expectations and for finding the time to enjoy other hobbies.

I, too, am weary of the endless debate. It'll never be settled, I'm sure, hence the "endless" tag. I'm not now, nor have I ever been, interested in recruiting for the "party faithful". People will either hear differences or not. I do and I reported on them. I'm satisfied that my first impressions of the cables (sighted) were essentially confirmed, if not fully quantified in terms of degree. Folks can make of my report what they will. And I don't use "foo-foo dust" anymore because even though it extended the bass of my system, it created a mess on the carpet....

[mystic]

In science, you want p<0.05, not p=0.05. In any case 14/20 is p=0.058 , which is greater than 0.05. 16/20 would be the minimum -- and even then such a score would arguably justify re-testing (as would 14/20).



Feel free to post to RAHE to arrange the details of your test, which can of couse be written up when conditions are agreed on. The only 'game' is a the application of standard scientific controls to a listening test.

I apologize for the late reply, but I overlooked the post. No, I'm not interested in doing one of these tests and probably never will be. My comments were to those forum members who might be interested in accepting the challenge. Since I haven't heard about any members taking part, maybe they think the test would be too difficult to pass or isn't fair and square.

More people might be willing to accept the challenge if the bar were a bit lower.
If you would be satisfied with a p<0.05, how about 13 correct out of 18 trials? I believe that would be p = 0.048. It seems to me 13/18 would be easier than 15/20.

While I doubt individuals conducting such a test would consciously do anything dishonest to affect the outcome, I would find it troubling that they have a stake in that outcome. The fact that they are presenting a challenge suggests bias. Providing full details about the test, including descriptions of possible limitations and weakneses might gain trust.

Regardless, you probably aren't going to get many takers. Do you really expect many members of this forum to go to the trouble and expense? Perhaps rather than trying to run some kind of contest, those making the cable altering claim could do tests on a lot of different audiophile cables themselves and tell us what they have found. It would be interesting to hear about the details. Are such results already available?

[Mike Anderson]

This is one statement I wholeheartedly agree with. Your tests prove nothing.

There weren't nearly enough trials to be statisitically significant. If you flipped a coin 13 times and it came up heads 10 times in a row, that does not demonstrate that the coin isn't fair and that after 1000 tries or 1,000,000 tries it wouldn't be nearly 50%/50%.

I'm a statistician, and I can tell you this is incorrect. You're thinking of the N required for a calculation based on the normal approximation. But here, it's possible to calculate an exact p-value, and the N can be much smaller.

For example, suppose you flip a coin five times, and it comes up heads on all five flips. Under the null hypothesis (that the coin is fair), the chance of that happening is (0.5)^5 = 0.03125, which is significant at the 5% level. So we can reject the null hypothesis.

Here, the calculation is a little more complicated because we have to account for all the different ways a person can get a number of correct guesses. For that we can use the binomial coefficient, or (N choose K).

For example, the number of ways of getting 10 out of 13 is (13 choose 10) = 286. Under the null hypothesis (that the chance of guessing the correct cable is 0.5), the chance of getting any one sequence of 10/13 right is 0.5^13. Since there are 286 possible sequences, the chance of getting 10/13 is 286*(0.5^13).

Now to calculate the p-value we also want to find the chance of getting 11, 12, and 13 right. So the full calculation is:

(13 choose 10)*(0.5^13) +
(13 choose 11)*(0.5^13) +
(13 choose 12)*(0.5^13) +
(13 choose 13)*(0.5^13) = 0.0461 (as posted earlier)

Therefore the result is significant at the 0.05 level, and we can reject the null hypothesis.

Keep in mind, however, that the 0.05 threshold is an arbitrary selection; perhaps the original poster would like to do more trials to see if he can beat 0.01.

[Mike Anderson]

The procedure was flawed. Without rapidly switching, you can't be sure there weren't other variables at work you weren't aware of. Did you just happen to listen to one cable on evenings you came home tired from work? Were there any other coincidences you didn't consider.

Again, this is incorrect.

The whole point of flipping a coin each evening to choose the cable is to avoid these problems. The argument that there were any such coincidences (which would be correlated with cable selection only by chance, since he flipped a coin to select cables) is ruled out with a confidence level of 0.0461.

Any valid attack on the methodology of the experiment has to argue that it wasn't truly blind, i.e. that he was clued into the selection of the cable in some way.

[hermanv]

Hi Mike;

About fifteen years ago Stereophile ran some double blind cable listening test. They used 10 listeners and three of the ten got results beyond chance one of them got a very high percentage correct.

The debate about the validity of the test raged on. A member of The Boston Audiophile Society in particular reasoned that since the test results MUST be random, any listener would eventually score very high so they proclaimed the test as not proving the result that Stereophile said it did. Letters raged back and forth for months (actually this debate rages to this day).

I am an engineer by trade not a statistician, but I am exposed to statistical data. It seems to me that you can't always apply the same rules to data that has a skew as you can to truly random events. If you will allow me an egregious example: A four minute mile was run in 1954, thousands of people could not duplicate the result therefore statistically it never happened.

Not a reasonable example, but illustrative of confusing notions. The right answer to the cable test dilemma of course, is to repeat the test so the sample size is large. The problem is that even a small sample size test takes all day and the participants loose interest, especially those who claim to be able to hear the difference and are wondering why they need to endure hardship to convince anyone else. They are quite happy with their own belief set.

Worse yet are those who interpret the test requirements as follows "If one person can hear a difference then all people must be able to hear a difference." This is equally ludicrous, not everyone is a professional wine taster, I have no problem assuming a professional can taste details in wine I can not.

http://www.bostonaudiosociety.org/ba...bx_testing.htm
http://www.bostonaudiosociety.org/ba...l_thinking.htm

Any comments would be welcome, of course if you disagree with my forgone conclusion then you must be a ___________ (fill in the blank) .

[Mike Anderson]

A member of The Boston Audiophile Society in particular reasoned that since the test results MUST be random, any listener would eventually score very high so they proclaimed the test as not proving the result that Stereophile said it did.

I don't follow this. If what they mean is that do enough tests, and eventually someone will score high just by chance, well then that's why you have to include the N's from all your tests when you calculate your p-values.

I am an engineer by trade not a statistician, but I am exposed to statistical data. It seems to me that you can't always apply the same rules to data that has a skew as you can to truly random events. If you will allow me an egregious example: A four minute mile was run in 1954, thousands of people could not duplicate the result therefore statistically it never happened.

I'm not sure what you mean by "skew" here (that has a technical meaning, it's a defined parameter of a probability distribution); I think you mean "outlier".

The rules you can apply to data depend on how the data are generated. Randomization is a powerful method that allows you to apply some well-defined rules (assuming you do the randomization properly.)

The right answer to the cable test dilemma of course, is to repeat the test so the sample size is large.

Again, the point of my above post is to emphasize that you don't necessarily need a large sample size. If you do five listenings, and the test subject is able to differentiate between the two sources in all five listenings, that lets you reject the null hypothesis at the 0.05 level.

You only need a larger sample if it's hard to detect the difference between the two sources (i.e. the test subject gets numbers closer to 50% correct), or if you want to reject the null hypothesis with a greater level of confidence.

Worse yet are those who interpret the test requirements as follows "If one person can hear a difference then all people must be able to hear a difference." This is equally ludicrous, not everyone is a professional wine taster, I have no problem assuming a professional can taste details in wine I can not.

I would agree with this. The test the original poster performed was to determine whether he could detect a difference between the cables. Certainly, a deaf person would not!

[musicoverall]

Any valid attack on the methodology of the experiment has to argue that it wasn't truly blind, i.e. that he was clued into the selection of the cable in some way.

...the test was essentially double blind because the tester swapped the cables when I was not there and then he left before I got home to listen. Certainly I could have checked out the cables that were connected on any given day. But I saw no point in cheating. I would have been fine with the prospect of hearing no differences - I could have sold the cables and recouped my money. And since I'm not obsessive about being correct all of the time, I found no possible value in peeking. I did understand, however, that I may have been called out on this and accused of lying. I have no problem with that, either - since I was only experimenting for my own edification and not to convince anyone else. I posted my test because I thought others might find it interesting and fodder for discussion, not to convince.

[hermanv]

Hi Mike;

Thanks for your insights. I have been looking for the raw data from the Stereophile tests but have been unable to find it. I have found anecdotal references where it was stated that there were 10 listeners and 16 trials, without the hard data it becomes impossible to analyze who (if anyone) is bending the numbers to agree with their own subjective beliefs.

I have found a lot of conclusions and statements of people claiming "successful" double blind cable comparisons (in this context, sucessful meant the ability to hear a difference) but getting acess to the raw test results using only an internet search has proven most difficult.

I promised I would find support for my belief that cables do in fact sound different and that ABX studies had been published. I remembered reading some of the results, but the only references I can now find do not supply the raw data.

Nevertheless below are some claimed positive tests results. Negative results do abound, but the questions is: is there a difference? If only one person reliably hears that difference then there must be one. The fact that a large number of people get null results doesn't prove there isn't a difference, only that they couldn't hear it.

The first result is this very thread
musicoverall an enterprizing (and brave) individual set up a double blind speaker cable test. Although the results did not meet the 95% confidence they were well above random. This was inspite of the test being made arbitrarily difficult because it was arranged for each single cable test session to occur with a 24 hour interval, this long delay requires a very long acoustic memory..
Speaker cable blind listening tests

From the Cardas site: "Blind Testing"
Q.) Has Cardas ever engaged in any blind testing of its cables?

A.) Sure in the early days many, but now days there is so much product in the field and so much agreement on the character of various cables that setting up a double blind would seem silly to most high-end fanatics. However, the recording industry is very cautious and guarded about changes because the stakes are quite high, so a few years ago there was a double blind test done by what is I suppose the top mastering facility in the world, Grundman Mastering. The type of test is one often used in the industry to sort things out. The reason for the test was not to determine whether or not cables sounded different, that was already a given, the tests were to determine which cable sounded best and why.

In this test new mastering facilities were being built in the US and in Japan and cables were requested from all the leading suppliers from Belden to Transparent. Various cuts were mastered to CD changing only the cables. The CD's were sent to a panel of engineers and others and the results tabulated. None of the listeners knew what the difference was between the cuts. The results were easily quantifiable. I think it took a year to complete all the comparisons, in the end, two of the many cables compared proved profoundly superior to the rest. Grundman is now wired with Cardas Neutral Reference. - George Cardas

I also found the following, again without hard support data:
http://www.geocities.com/jonrisch/i3.htm
......The results I present in this note are based on comparing one cable to another, under blind (or often, double blind) conditions, where levels were always checked for compliance to +/- 0.1 dB, and the comparisons varied one aspect of the cable at a time, i.e., one cable might be identical, except the insulation was foamed PE instead of foamed PP, etc. or the wires were bare copper for one, and tinned copper for the other, etc.

So what I am saying, is that not only did I reliably identify different cables, but I was able to identify a single cable variable at a time. Dozens and dozens of different coaxial and twisted pair cables were listened to, over the course of hundreds of hours of listening tests, stretching over the course of years of effort.

Summary
So in summary I failed in my mission to provide skeptics with hard raw data supporting the contention that cables sound different. For me they do, I am happy with the belief that my system has benefited by careful selection and choice of both interconnect and speaker cables. In my system, cable costs are about 15% of the total investment.

[Mash]

Unless you people are in the business of making and peddling (expensive) “magic audio wires”, why on earth are you so obsessed with this “wires” topic? When I started the Hi-Fi hobby the goal was to accurately RE-produce live performances in our homes because often we could not get to a recital venue or symphony hall. Hence the hobby was called High Fidelity.

Sometime or other this hobby became perverted into everything EXCEPT meeting the goal of reproducing live performances in your homes. One chap seems to have spent maybe $5000 on wires just because they looked nice and he “could afford them”. I spent $500 on matched mono Futterman amps and another $400 to later re-tube my mono Futterman amps and have them modified to incorporate the NYAL auto-biasing scheme. So for $900 I have amps that, when combined with my Tympani, can reproduce concert grand pianos in my home that sound just like the many concert grand pianos I have listened to in recital.

[We Futterman owners also rewired our Tympani to higher resistance, the 4-ohm Tympani we rewired to 16 ohms and the 8-ohm Tympani we rewired to 32 ohms. Talk about realistic reproduction of music in the home!]

I even heard a lecture on how different grand pianos are rebuilt. But people who spend their time sitting around listening to wires are a sad bunch indeed. The goal is listening to MUSIC, people, not wires!

Hey, Skeptic, you should chill at the Darwin awards website and read the (often funny, but not for the star) stories of how so many people improve the human gene pool by doing something stupid to remove themselves from it (and this world). The point being that if people here want to spend their worldly wealth on wires, why not let them? Remember that being a retired millionaire is not going to be that comfortable if EVERYONE is a retired millionaire. So let these dudes waste their money in peace! I remember lots of people who made more money than I once made and now I can buy them and give them to charity. You can do the same. So, chill!

Skeptic, maybe we need “financial Darwin awards” ? Nah- I don’t want the competition when I go to look at more land.

[E-Stat]

Hey, Skeptic...

I think you'll find he posts over at AA now as "Soundmind", mostly in the Outside Forum.

rw

[musicoverall]

Unless you people are in the business of making and peddling (expensive) “magic audio wires”, why on earth are you so obsessed with this “wires” topic? When I started the Hi-Fi hobby the goal was to accurately RE-produce live performances in our homes because often we could not get to a recital venue or symphony hall. Hence the hobby was called High Fidelity.

Sometime or other this hobby became perverted into everything EXCEPT meeting the goal of reproducing live performances in your homes. One chap seems to have spent maybe $5000 on wires just because they looked nice and he “could afford them”. I spent $500 on matched mono Futterman amps and another $400 to later re-tube my mono Futterman amps and have them modified to incorporate the NYAL auto-biasing scheme. So for $900 I have amps that, when combined with my Tympani, can reproduce concert grand pianos in my home that sound just like the many concert grand pianos I have listened to in recital.

[We Futterman owners also rewired our Tympani to higher resistance, the 4-ohm Tympani we rewired to 16 ohms and the 8-ohm Tympani we rewired to 32 ohms. Talk about realistic reproduction of music in the home!]

I even heard a lecture on how different grand pianos are rebuilt. But people who spend their time sitting around listening to wires are a sad bunch indeed. The goal is listening to MUSIC, people, not wires!

Hey, Skeptic, you should chill at the Darwin awards website and read the (often funny, but not for the star) stories of how so many people improve the human gene pool by doing something stupid to remove themselves from it (and this world). The point being that if people here want to spend their worldly wealth on wires, why not let them? Remember that being a retired millionaire is not going to be that comfortable if EVERYONE is a retired millionaire. So let these dudes waste their money in peace! I remember lots of people who made more money than I once made and now I can buy them and give them to charity. You can do the same. So, chill!

Skeptic, maybe we need “financial Darwin awards” ? Nah- I don’t want the competition when I go to look at more land.

I'm obsessed with music and the most accurate reproduction of it, as you claim to be. Since wire is a part of that, I suppose one could argue that I'm obsessed with wire as well. It's part of maximizing the performance of my system. The live performance in my room goal has become closer to a reality with the addition of higher performance cabling. I know you disagree and I wouldn't dream of trying to persuade you otherwise. If you're happy, I'm happy for you.

E-Stat, wasn't Soundmind/Skeptic banned from AA?

[E-Stat]

I'm obsessed with music and the most accurate reproduction of it, as you claim to be. Since wire is a part of that, I suppose one could argue that I'm obsessed with wire as well. It's part of maximizing the performance of my system. The live performance in my room goal has become closer to a reality with the addition of higher performance cabling. I know you disagree and I wouldn't dream of trying to persuade you otherwise. If you're happy, I'm happy for you.

Well said. You might similarly say I'm obsessed with bass traps in that I use them all around my listening room. It's a pity he hasn't had the chance to hear a similar system to his using 20.1s driven by Joule Electra OTL amps using Valhalla throughout as I have. He might be surprised!

E-Stat, wasn't Soundmind/Skeptic banned from AA?

Perhaps, but he did pipe up earlier this month on one of his pet topics, negative feedback. Clearly, he did not find many like minded posters in the audio forums.

rw