dsi@unccvax.UUCP (02/26/86)
Making a blanket statement such as "phase distortion in audio equipment"
.."is inaudible" is pure bullshit. Notwithstanding several very important
papers done in the area of psychoacoustics existing which tend to prove this
point out, there are nevertheless several virtually unexplored areas of the
effect of phase "distortion" on music.
Theoretically, at least, the RIAA curve should ultimately work in such
a way that the phase response results in a constant time delay for all frequencies.
However, when RIAA compensation is implemented in negative feedback topology,
all that good stuff about summing-node-deviating-from-zero-ohms and the rest
of the distortions we all know, love, and have names for, come into play.
The classic paper on variations of input amplifier impedance and effect on
GROUP DELAY DISTORTION (read this) was written by Tomlinson Holman in 1974,
you can find it in JAES.
Now. . .about this business of "phase" distortion. Are we talking about
channel-to-channel time delay or phase matching, or are we dealing with a
single channel. I doubt the original poster (who made the remark about
not being able behave in a manner which results in phase-discriminating
activity) has ever switched one set of speaker wires in the former case.
Most of the research in this area has been done in the one-source case.
ANY modulating function of ANY kind is subject to not only first order
effects (commonly known as group delay distortion) but second order effects
which are known as those maladies called "differential gain" and "differential
phase." Audio types seem to be unconcerned with this, but those of us
who are in digital television are passionately aware of the problems that
second order phase distortion cause. I have not researched this thoroughly
but consider myself very well read; and have never seen a paper documenting
the effects of second order phase distortion.
'
Seeing that the perceputal apparatus for hearing is much more differ-
entially sensitive than absolute-sensitive (i.e. can you hear a pure tone
and state that it is 70 dBa ?) it would stand to speculation that the ear
may be more concerned with phase changes or gain changes of a particular
function in the presence of other funcitons.
Differential gain/phase distortion is exceedingly difficult to measure
and even as a concept is difficult to formulate because the traditional
meaning came from "to a reference phase." Nevertheless, if instruments
are dynamically changing their phase matrices during actual reproduction
the effects might be those which explain sonic differences in seemingly
"identical" amplifiers.
"Standard" phase distortion (group delay distortion) as a practical
matter causes problems, too. Again, the issue is not whether or not
humans can accurately state "there is a 1.2 radian phase shift at 3000 rad/s"
on such stimulation - though it would be nice to know that conclusively -
but the effect that group delay distortion has when presented to nonlinear
elements, including THE EAR! A classic, though somewhat primitive example
of standard phase distortion would be the overshoot caused by a lowpass
filter followed by a nonlinear element, for those frequencies approaching
the stop band.
Perhaps I am blowing a lot of hot air, but it would seem to me that
the problems that we digital video types have in making systems transparent
are equally applicable to the world of 20 - 20000 hz.
'Nuff said
David Anthony
DataSpan, Inc