Cyber ears

[Robert Scott]

Kevin Riggs writes:

>There is at least one person with hearing problems that claims
>he can hear the beats when enhanced but not otherwise.

I don't doubt it.  But even with his hearing problems, he is
still using his own patern matching ability to pick out the
beats.  I can believe there are many ways to enhance the sound
to make it easier to hear beats.  I was only saying that it
would be very hard to leave the human totally out of the loop.


On the same subject, Richard Brekne writes:

>But the computer can sample to notes, and all relavant partials.
>Compare the frequences of coincidents and calculate the difference,
>and finally present the resulting data in terms of bbs.

>Does this sound do-able ?

Yes, if you sample the notes separately.  You would have to
play one note alone, wait for it to decay (because an accurate
sample requires a long sample period) then play the other
note, then wait for it to decay.  Then the computer could
calculate the beat rates without actually detecting them
directly.  But this makes it very inconvenient to use in
tuning.  You would have to set the tuning pin, then play your
two notes as described above, then see the beat rate, then
re-adjust the tuning pin.  I can see making no more than one
adjustment every 10 seconds.  But in aural tuning, you often
make adjustments several times per second, and maybe, when
you are very close, one adjustment every two seconds.  It
would take you all day to tune a piano if you had to wait
for the computerized "beat calculator".  You could cut
the time roughly in half by letting the computer know
which of the two note you are tuning.  Then it would only
have to sample the fixed note once while sampling the
note being tuned after every tuning adjustment.  But even
that would give you only one tuning adjustment every 5 seconds,
which is more than double the time it takes to tune aurally.

What the computer cannot do is sample all the partials of
two notes being played simultaneously.  The coincident
partials are too close together to be detected separately.

I think the effect you are trying to achieve is to remove the
note-by-note decision-making function completely from the ETD
and place it solely with the technician.  You can just about
do that now with any ETD.  Suppose you want to tune a C-G fifth
by tuning the G.  If you use the ETD to measure the offset of
the a certain partial of the C, then you could calculate (by
hand) the desired offset of a coincident partial in the G.
This calculation would be based on the desired beat rate
for this coincident partial.  Enter that offset into the ETD
and tune the G using the ETD normally.  Of course, all that 
calculation and offset entering is tedious and not really 
practical, but at least it gives the idea for experimentation.
But tuning that way bypasses one of the biggest advantages of ETDs.
They have no cumulative error.  Direct interval tuning does.  
That's why aural tuning has so many cross checks - to compensate 
for cumulative error.

-Robert Scott
 Real-Time Specialties