String coupling.

[Robert Scott]

Richard Brenkne suggests:

>I personally am convinced that if ETD would provide us information 
>about more then one partial at a time (as in 3 or 4) we could learn 
>an awfull lot about this process.

Here's the difficulty, Richard.  For an ETD to display tuning information
about several partials, that ETD would need to have a target frequency
for each of those partials.  Where should we get that target from?
Current ETDs all use a tuning curve to specify the target frequencies.
The tuning curve specifies the cents offset from a zero-stretch
equal-tempered scale.  The tuning curve also specifies which partial to
use for each note.  So we only have a target specified for one partial
for each note.  If we were to display tuning error for several partials
we would need to have an enhanced tuning curve that specified the
stretch for each partial.  How would we go about generating such a
curve?  We could use models of inharmonicity like the SAT (FAC method).
But if any manual tinkering is required, can you imaging the complexity
of the editing screen?  Instead of just one curve, there would be
3 or 4 curves.  Would you edit them separately, or would one editing
operation move all curves simultaneously?  How would you make allowances
for measured inharmonicity?

Once you figure out how to generate a multi-partial tuning curve,
then in principle it would not be hard for the RCT or TuneLab to 
have several spinners or phase displays (although, I would probably
have to stop supporting 486's).

The desire to see several partials at once may have research appeal,
but I doubt it would be of much day-to-day use.  If you don't mind
using the same stretch numbers, you can quickly change the partial
in TuneLab using the F3 and F4 function keys to take a peek.  The
only context in which this multiple partial thing came up is in 
regard to unisons.  And even the most ardent ETD fan will probably
still tune unisons by ear (except possibly for the high treble
where multi-partials are not an issue).

If you really want to play around with multi-partial, do as
Steve Fairchild has suggested and get several ETDs together, each
set for a different partial.

-Robert Scott
 Real-Time Specialties