Propeller Head Plaza: A reasoned and effectual reply from a buddy of mine... by clarkjohnsen Technical and scientific discussion of amps, cables and other topics.

For Sale Ads FAQ / News / Events

Posted by clarkjohnsen (R ) on September 11, 2007 at 14:16:51

In Reply to: Boston Audio Society Strikes Again! posted by Charles Hansen on September 11, 2007 at 09:44:35:

...who has done some work already along these lines. (Once upon a time he was selected as one of America's "Top Young Scientists".) Many there will be, who won't care to hear this, but give it a shot.

____

I don't know about the specific tests in the Boston silliness, but to my
way of thinking, the fairest test would be to take high quality digital
material (e.g. 96 Khz 24 bits - no non-linear 1 bit SACD stuff) and
make specific versions of it that are degraded. For example, down
convert to 44.1/16 and back up to 96/24. Then play the two files back
through the highest quality available equipment and see if there is any
difference. There will be NOTHING ANALOG that is changing, just the
digital signal itself. Furthermore, there is nothing different about
the digital signal processing into and out of analog. So the difference
between the original and degraded material can be definitively
understood. (And the polarity will be the same as well.)

Now, here's the rub. I did part of this test last year while mastering
an album. Specifically, I went from 44/24 down to 44/16. We conducted a
"triple blind" test. Triple blind, in that not only the experimenter or
the subject knew what was being compared, but even more, none of us knew
that there was actually an experiment underway!

[Note: I developed this same technique in the late Eighties, and presented it in a paper to the 91st AES Convention in 1992; I have taken flak for my impertinence ever since, although none of my antagonists cared to re-analyze my results.]

That is until the
subject (the producer for the album, I was the engineer) complained that
the sound quality was poor. How could this be? The sound quality the
day before had been agreed to be good. What had changed was the word
length. The particular method used to down convert from 24 bits to 16
bits had introduced a low level of distortion -- something that should
have been completely inaudible, but it was actually the difference
between music that was pleasant to listen to and music that was harsh
and unlistenable. Ultimately, we had to get down to 16 bits so we could
make a CD, so after trying several different methods of dithering and
noise shaping we finally came up with a 16 bit version that was acceptable.

Now the theory says that all that changed between the good reduction and
the bad reduction was the low order bit. This was below 0 db spl in the
listening room, at 1 meter from the speakers. (The gain was calibrated.)
We were listening in a room full of computers with fans and whirring
disks, and a noise level of perhaps 45 db or more. So how could the low
order bit have mattered?

Theoretically, it couldn't have mattered. The intellectuals will say we
were deluded. The artists know better. The intellectuals will dismiss
this as "anecdotal evidence". This is, of course, the paramount
buzz-word they use to dismiss results that disagree with Kuhn's "Normal
Science", that is to say their favorite theories.

Now I've thought about this quite a bit -- because this gets to the
heart of the problem that began when digital audio first came into
existence. If one could take tests like this and cast them into a form
that would convince a reasonable fraction of the flat-earthers perhaps
some progress could be made. In other words, make a double blind test
based on the kind of sonic differences that seemed so significant to me
last year.

But I don't think this will work. Why? Because double blind testing
won't be able to distinguish what we were hearing. The problem was that
listening to the polluted music was giving us a headache. It took some
time to recover before we could appreciate the unpolluted music. In a
rapid sequence of blind testing our malaise would have continued and
would have been falsely applied to the unpolluted music as well as the
polluted music.

There may be a way out, but it's very high budget. Perhaps PET scans of
the brain of listeners will show different neural activity when
listening to low and high quality audio. They may give external
("objective") access to some of the subjective aspects of perception
that are not readily available to the consciousness of many subjects.

PS No high end equipment here. A $59 external sound card, plus 20 year
old amplifier and speakers that originally cost under $500. These
differences could be heard on cheap headphones as well.

This post is made possible by the generous support of people like you and our sponsors:

Amplified Parts

---

[ Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2024, Audio Asylum, all rights reserved ]

---

[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]

---