James Boyk wrote:




Heyser was a brilliant & inventive guy who invented the machine that
easily measure energy-time curves, and had many subtle & insightful
ideas about audio--all in his spare time from his work at Jet Propulsion
Lab. (Unfortunately, his writing was turgid.) His black box was built to
demonstrate to an AES meeting the flaw with conventional specs; and more
fundamentally, the flaw with attending to specs rather than listening.
Put any standard test signal through it and it measured not very good
but perfect. Put music through it and it sounded atrocious.

The design consisted of a hard-wired connection from input to output
passing through a relay. A bridging circuit could look at the signal. If
it was symmetrical--as all standard test signals then were--the relay
stayed closed and the signal was passed intact. If assymmetrical, the
relay was opened. Since music will look intermittently symmetrical and
not (depending on the time constants of the analysis circuits, etc.) the
relays were constantly opening and closing on music signal. The result
was gross.

I understand that there was *considerable* confusion when he demo'd the
black box at a local chapter AES meeting. It was a vivid way to make his
point.

One inevitable response to this story will be to point out some
asymmetrical test signal that would therefore Not have measured
perfectly. This will miss the point.

James Boyk

Sure; an antecedent of TDS. Good. We've fixed those problems now. But in my
wire tests I verified level matching at 100, 1000 and 10,000 Hz with a sine
wave at the speaker terminals.

No subject was able to demonstrate an ability to reliably distinguish expensive
rca interconnects or expensive speaker wire from junk box rcas or zip cord
purchased for18 cents a foot.

In every case but one the subjects used cables that they personally owned. In
some cases the test was performed using the subjects personal reference system.

In the exception case I assembled a system to test the 'series tweak' idea
(some single tweaks,like speaker wire, may not have audible effect by
themsleves but when coupled with other tweaks they become audible) compared
with a Geak system comprising of decidedly low-to-mid fi components that any
high-ender would look down his nose at.

The Tweak system used the digital output of a Marantz CD-63 cd player, special
interconnects, special networked speaker cables, an audiophile-grade outboard
DAC, a vacumn tube preamplifier, a high-end solid state power amplifier,
speaker wire stand-offs, vibration absorbers, special wire dress throughout and
a aftermarket AC cord.

The Geak system took the analog output of the same cd player, junk box rca
cables, zip cord sold for use in car stereo systems using a 6-foot section for
one channel and a 25-foot section for the other (with the 25-foot cord wrapped
around the AC cords for the other components. Speaker connections were made
with Pamona brand dual bananas. The preamp was a 25-year old $99 Heathkit, the
power amplifier was a 100-watt Parasound that I bought for $200 used several
years prior.

Speaker were the PSACS reference PSB Stratus Minis of which this particular
pair had been measured in the actual NRC anechoic chamber.

Subjects were tested individually and had been asked to bring with them those
cds that MOST dramatically highlighted differences in audio systems.

They were given as long as they felt necessary under open conditions to get a
bead on differences. Most spent about 45 minutes to an hour.

Afterward I asked the subject to leave the room while I flipped a coin and then
connected one of the two systems to the speaker terminals and draped a visually
opague but acoustically transparent over the rear of the speaker cabinet
covering the terminals.

Subjects then had aslong as they wished to decide which system was driving the
speakers. They were also allowed an 'open' refresher anytime they felt
necessary between any given pair of trials.

The speaker connections (me) and scores (subjects) were made on business cards
and dropped into a slot in a sealed cardboard box for each trial. Subjects were
asked to verify the speaker connections visually after every trial.
Inerestingly none of them wanted to.

Subjects were either paid a straight $20 fee OR they could bet $20 of their
own money against $100 of mine that they could identify systems a significantly
significant amout of time. They were free to accept the $100 bet at any time;
meaning if they were feeling confident midway or AFTER completing the session
both payment alternatives were avialable up until the time I opened the box for
scoring.

They were also asked to agree to complete 10 trials at any given session BUT
were allowed to complete any number of trials in any session or follow-up with
any number of sessions they would like at any time within 3 weeks of completing
a session.

One subject extended his session to 16 trials. One subject bet the $20. No
subject was able to reliably identify the systems. ( 9/10; or 12/16)

Every subject initially said they felt they could tell the systems apart under
open conditions but they showed that they were unable to do so with the
simplest of bias-controls implemented.

This is all documented in "To Tweak or Not toTweak" in the June 1998 Stereo
Review. Please note that this experiment was no done on assignment from the
magazine. I conducted it myself to test the series-tweaks idea.

I took a lot of BS when I was initially skeptical about it. I repeatedly asked
one proponent (on-line) to verify it with a controlled test. I even offered to
travel to his place on my dime to watch the confirmation. He reluctantly
'agreed' but when it cam time to make travel arrangements he disappeared.

So I began to implement dubious tweaks and soon found that I was unable to
duplicate the series-tweak effects. Of course, knowing that proponents would
just argue that my ears and/or system wasn't good enough.

So I decided to assemble systems that would be 'different enough' that
some/any benefits/changes of series-tweaks would have to be fairly obvious.

And yes I did use ABX to validate my listening experience with these systems
but chose not to do that for this experiment because I knew that would be the
1st objection from believers.

Note there were no ways for me 'stop' listeners from hearing true differences.
I remained at the rear of the listening room during each trial to make sure
that no peeking was going on but I had no voice in the order/time of program
presentations.

If I were to have had an influence it would have been inadvertant clues
delivered to the listeners and if i were to have purposely or inadvertantly
given 'reverse' clues then subjects would have had 'reverse significant'
results.

Let me finnish (pun intended) by saying that no proponent has delivered a
single replicable experiment that shows that commercially sold audio cables
have any influence, positive or otherwise, on sound quality.

I'm still wondering how one would conduct a bias controlled test of a live mic
feed. But even if there were such differences to be found in production what
difference would that make to the average enthusiast in his home? It might
determine some of the recordings he buys; but it has no effect on what he hears
on his system that is non-program dependent.

"Nousaine" wrote in message

I'm still wondering how one would conduct a bias controlled test of a live
mic
feed. But even if there were such differences to be found in production
what
difference would that make to the average enthusiast in his home? It might
determine some of the recordings he buys; but it has no effect on what he
hears
on his system that is non-program dependent.

It would have to be a live mic feed of a recorded and reproduced signal, as
a live performance would be unrepeatable at a sufficient level of
consistency for the results to be meaningful.

geoff