ETD Displays

[Robert Scott]

Richard Brekne wrote:

 >Robert Scott said he could do it for
 >something under 3000 bucks last year...

Before anyone starts writing me a check, I need to qualify the
context of that discussion.  Richard was explaining his concept
of how an ETD could be used to directly compare two notes, and
how a multi-partial display might assist in that effort.  I never
did fully understand the practical application of Richard's
ideas to a procedure for tuning (not that it is so unintelligible,
but rather due to my limited experience in practical tuning).
My estimate of the effort involved in this hypothetical software
modification was for the capability to display additional
partials simultaneously.  I did not address the issue of how
each partial might be referenced.

For the record, I don't think the processing of the phase display
in TuneLab is a very large fraction of the total real-time
processing. The spectrum display, for example, takes much more.
If there were, say, four phase displays, each linked to a different
partial, it would not add too much to the total computing load.

Although not directly related to the issue of multi-partial displays,
the idea of direct interval tuning seems to be coming up in this
thread.  If I understand Richard's description:

 >take A4.... play it and use the puters controls to stop the
 >dancing dial. Now play A5 and tune it till that same
 >dancing dial stops. You have now exactly matched whatever A4
 >partial the etd was set for in the first place.

This seems to be what Bill Bremmer and others have been calling
direct interval tuning.  While an ETD can be used this way, it
seems to me like using a nail gun to drive nails by swinging the
tool like a regular hammer.  It ignores the real power and
potential of an ETD.  If all you want to do is compare two
notes, your ears are better suited to that task than any ETD.
The power of an ETD is in its ability to encompass the
entire scale at once and give you an absolute reference for
every note so that you do not develop cumulative error.

I have thought a little about how a multi-partial display
might fit within the existing structure and usage of an ETD.
Suppose that you start with an existing tuning curve.  This
defines an offset for one partial of every note of the scale.
This tuning curve could be extended to provide offsets for
several partials for every note provided the ETD has a model
of the inharmonicity for every note.  That way, if you switch
partials, the offset is corrected for inharmonicity.  By the
way, this type of partial-switching compensation, while not part
of TuneLab 97, is provided in the soon-to-be-released TuneLab Pro.

Once you have an offset defined for every partial of every note,
it is conceptually possible to use this information to provide
a multi-partial display.  I expect that if this option were
available, it would not be used very much.  When using an ETD,
one generally tunes each note to a single partial.  Sometimes
with dead bass strings, the selected partial may be too weak to
register, so you might resort to plucking the string or
temporarily selecting a different partial.  The point is, the
type of piano where this becomes an issue is hardly the kind of
piano that would warrant a long involved and detailed examination
of all the partials.  So the multi-partial display seems to be
a solution where there is no problem.  But I would be glad to
hear differing views.

-Robert Scott
  Real-Time Specialties