Interesting Article about the Audio Press

[Guest]

<font color='#000000'>I just wanted to chime in on a couple of things.

I certainly didn't want to accuse anyone here of &quot;whoring&quot; to the advertisers. There is a high degree of professionalism on this site which I certainly appreciate. Perhaps it has something to do with the fact that you are degreed audio engineers. This is not something the journalism majors at the magazines can generally claim and the professional ethics of these two groups are usually quite different. I greatly appreciate that you can explain much of this stuff clearly to someone with only two semesters of college physics well behind him. I just wanted to raise a caution about how the appearance of impropriety could arise and an advertisement That I found could be confused as an article.

A note on ABX testing. I have yet to see a good ABX test in the audio field. My backround is medicine (radiology) and I have previously done a fair amount of benchwork research. I have also participated in several blinded ABX studies comparing image quality between various competing imaging equipment especially ultrasound. The difficulty falls into the unpleasant (for me) realm of statistics. To obtain a valid sample size a large number of trials are required when the difference between A and B is small. If one flips a coin twenty times and gets 12 heads is it unfair or just random luck? How about 1200 heads in 2000? I once participated in a study to compare three ultrasound machine's image quality. 5 colleagues and myself each looked at 600 ultrasound images and evaluated them blinded as to which machine they came from. This required a commitment &nbsp;of over 30 hours each as well as requiring two hundred patient's livers to be scanned on each of the three machines. The cost was significant (it was footed by General Electric who finished third of three). Most audio companies don't have this sort of research budget even if they had the desire. University audio engineers may try this but sufficient grant dollars may be hard to come by compared to the grant dollars available for say a cancer drug trial.

I think the results would be fascinating but a valid trial would need over 100 and probably 200 subjects to truly prove A was better than B unless we compare a boom box to a $50,000 system. Also remember, in medicine we want what works best and while there is some bias at times from the corporate world, the good work is peer reviewed and is often a multicenter trial. In audio the industry has a much greater influence. Unless I see it in JAES I would have doubts just like I don't believe in any of those cola taste tests validity.</font>

 

[Rob Babcock]

<font color='#8D38C9'>You make some good points, Dan.  I do think, however, that some of the tests were done with a statistically adequate amount of subjects.  Also, many subjects repeated the test many many times.  As you point out, a large amount of data is needed when the difference is small- I just find it humorous that the audio press claims the differences are &quot;not subtle&quot;, &quot;nite and day&quot;, etc etc then can't discern those differences with any statistical reliability.  If it was as plain as the nose on your face, you'd think 10 out of ten would be easy.  Or at least 8 of 10.  I've never even seen a reliable 7 of 10.

Again, I'm not saying they all sound alike.  But even if they do no amount of statistical or empiracle evidence would change someones mind.  We're touching on articals of faith, not scientific inquiry.  The notion that every single item of gear has its own &quot;soul&quot; is fundamental to the Audio Holy Writ- any suggestion to the contrary is heresy.  And of course that notion would be pretty threatening to those whose livelihood is based on spinning purple prose to describe those nonexistant differences.</font>

 

[Rip Van Woofer]

<font color='#000000'>Earlier today I added this to my previous post. Perhaps I should have posted it separately:

&quot;There is an AES paper by David Clark titled &quot;Ten Years of A/B/X Testing&quot;, AES preprint #3167, that might be relevant to Jeff's question. AES preprints are $5 ea., available on their Website. I just ordered a copy.&quot;

I think ten years worth of tests should provide a statistically valid sample, even if it is an aggregate of many smaller individual tests. Anyone can buy an AES paper for five bucks--check it out. The paper also appeared in the JAES, I think. An earlier paper by Clark describes the ABX comparator and the testing methodology if you're interested in that as well.

Audio Engineering Society Website

(I hope the patients got their livers back after scanning!

)</font>

 

[A. Vivaldi]

<font color='#000000'><table border="0" align="center" width="95%" cellpadding="0" cellspacing="0"><tr><td>Quote </td></tr><tr><td id="QUOTE">Again, I'm not saying they all sound alike. &nbsp;But even if they do no amount of statistical or empiracle evidence would change someones mind. &nbsp;We're touching on articals of faith, not scientific inquiry. &nbsp;The notion that every single item of gear has its own &quot;soul&quot; is fundamental to the Audio Holy Writ- any suggestion to the contrary is heresy. &nbsp;And of course that notion would be pretty threatening to those whose livelihood is based on spinning purple prose to describe those nonexistant differences.
</td></tr></table>Your right. Even if it's 100% true that almost all amps sound the same it wouldn't matter anyway. If one truly thinks or believes that the amp they bought sounds better, than it does. If the act of simply spending $2000 on a amp that looks cool and is built like a tank makes it sound better, then it's well worth it.</font>

 

[A. Vivaldi]

<table border="0" align="center" width="95%" cellpadding="0" cellspacing="0"><tr><td>

Rob Babcock : <font color='#000000'>Pete Townsend personally oversaw the remix of the new MC SACD version of &quot;Tommy&quot;- but he's been nearly deaf for many years!

  I've even heard him discuss his hearing loss in interviews.</font>

<font color='#000000'>That Tommy SACD is awesome and is the best one I've bought so far, so mabye Pete Townshend did &quot;oversee&quot; it's production, but did he &quot;overhear&quot; it? &nbsp;

</font>

 

[Rob Babcock]

<font color='#8D38C9'>Hmmm...I don't have that one yet. &nbsp;Too many things out right now with too little 'casheesh.&quot;

&nbsp; I really want it, though...maybe I'll have to splurge.</font>

 

[Guest]

<font color='#000000'>The JAES paper is on the way. Thanks for the tip Rip! I'll be curious to see the methodology. Multiple tests work fine for achieving statistical significance if they are performed in a reasonably similar manner. The statistics required are called a meta analysis for those who care. They have been done with hundreds of thousands of women in multiple studies for example to assess the efficacy of mammograpghy for example. Doing a meta is well beyond my abilities however. Oh and Rip, while those livers were merely borrowed we did take 500! livers and sliced em and scanned six ways from Sunday. The patients didn't need them, we gave them all new ones (well slightly used). I can't eat liver to this day.

</font>

 

[Mudcat]

<font color='#000000'>Bringing it back to Steve and Jeff.

I'm neither broke nor mostly broke, just

flatbaroqueandberserk

apologies to Roy Harper

and if you've listened to Roy Harper, then you would understand the folkjokeopus. &nbsp;Please stop me before I ramble on like a bullinamingvase

</font>