[ In reply to Dan Wilson's letter in Digest 980605 ]

MIDI enables one machine to send short messages to another that can
turn a note on, or turn it off (yes, separate messages), or select a
tone color, or effect some other change.  A Note-On message does include
not only a keynote (one of 128 pitches) but an intensity indicator
(1 of 128 levels of intensity, depending on the velocity with which the
key went down).  MIDI can send as many as 1000, 2000, even 3000 such
messages per second, so it can turn on a note (theoretically) in 1/1000
of a second.  That's fast enough that we mere human performers don't
notice a lag.  Not unless there's a _huge_ amount of stuff being sent
over the MIDI cable when we're trying to play a chord.  We tend to pay
little attention to the "problem" at that level.

By the way: no sound starts until the key hits bottom_.  That "1/3 of
a second" that a key is descending is spent counting milliseconds in
order to decide how loud to play (longer = softer); the note is not
begun until that timing is completed.  Obviously, that timing "lag"
would require a performer to make a prediction as to when his key hits
bottom in order to play with "good ensemble."

But Mr. Wilson's point is well taken.  If a machine had to respond
strictly to music control data coming over a wire, there could be a
buildup of the sort he mentions, not necessarily from MIDI problems but
from solenoid delays, speed-of-light between London and California (for
example), software note processing in each computer and synthesizer,
and Internet packet processing delays.  So mechanical synchronization
could be an acute problem.

Actually, the problem is similar to pipe organs that could have an
amazing delay between console key-press and start of sound from a pipe.
For ensemble purposes the fix is the same in both cases: a human
performer.  We develop a sense of pulse so that we may predict with
uncanny accuracy the millisecond at which we should start a tone.
That's how really good performing groups play well together; not by
start-together-at-some-point-A-and-meet-at-the-pass (never works) but
by _listening and adjusting_ continuously (only way it can work).

Michael J. Babcock