SuperQ;626734 Wrote: 
> It sounds like they're using something like audio diffmaker.  It should
> be easy to reproduce their results.

Exactly.  They reach the conclusion that the error signal is
proportional to the time rate of change of the signal being tested. 
Why do you suppose this is?  Let's look at the definition of the
derivative of a time function f(t) (that is, its instantaneous time
rate of change).

df(t)/dt = limit as delta_t approaches zero of (f(t+delta_t)-f(t)) /
delta_t

The idea of their measurement is to introduce the appropriate delay
into one of two signals that are nominally almost equal, except for a
delay, such that when you subtract them, you get almost zero.  But
suppose there's a small error in the computed delay.  Then you get a
time function that looks like this:

g(t) = f(t+delta_t)-f(t)

which is indeed directly proportional to the derivative of f(t), with a
constant of proportionality delta_t.

Notice in their presentation, they do not label the vertical axis of
the graphs of this difference signal, making the results highly
suspect.

But as anyone with experience with DiffMaker will attest, if one
creates two signals that differ by only a delay (best done digitally),
DiffMaker does such a good job of finding the optimum delay before
subtraction that the difference signal is too small to be above the
noise.  This is true even for delays that are not an integer multiple
of the sample period.  It automatically takes care of that too.

So it appears the error signal they're seeing is just from inaccurate
computation of the delay between the two signals.  They're trying to
reinvent the freeware DiffMaker and doing a much poorer job.  Whether
this is deliberate or not I'll leave up to you.


-- 
andy_c
------------------------------------------------------------------------
andy_c's Profile: http://forums.slimdevices.com/member.php?userid=3128
View this thread: http://forums.slimdevices.com/showthread.php?t=87175

_______________________________________________
audiophiles mailing list
[email protected]
http://lists.slimdevices.com/mailman/listinfo/audiophiles

Reply via email to