source file: m1524.txt
Date: Tue, 8 Sep 1998 13:02:48 EDT

Subject: Re: MIDI  implementation & Yamaha XG

From: DFinnamore@aol.com

Hi Drew, thanks for the good info!

Drew wrote:

>Pitch bend, I checked with the range at 1-semitone and 
>found no clear pattern in the data values that caused each successive
>pitch change.
>Graham Breed mentioned he does this and
>David Finnamore posted a note cautioning about it, you HAVE to check 
>the frequency of the bends against actual cent frequency equivalents.

For those of us without a frequency meter, old-fashioned beats can be used.
Here's my favorite trick for making sure I'm in tune when stretching the
capabilities of my O5R/W, which, like Drew's Yamaha, uses some tuning-related
SysEx implementations that seem to defy logic:

First, select a tuning table that you are certain about for one channel, like
Pythagorean or 12-tET.  Sound a long note on that channel, tuning it with the
channel's Detune parameter to the target tuning's pitch for that note.
Simultaneously, sound the same long note on another channel using the target
tuning/transposition, and listen for beats between the two notes.  Using a
pure sine wave on both channels is most helpful.

A related method worked well this past week when I Just-tuned (mostly 5-cap
sprinkled with 7-lim consonances) a MIDI file of an old 12-tET orchestral
instrumental of mine, which modulates from Ab to C#m, back to Eb, and then,
over the course of about a minute, through F and E to end in G#.  The G# is a
double syntonic comma - nearly a quarter-tone - south of the original Ab!
Kind of stretches the ear a bit, but it sure is nice to have all those pure,
resonant vertical intervals.

The bass line (almost all roots) was on Pythagorean-tuned channels, and stayed
Pythagorean except at the points of the last two modulations, which were by
15:16, and so needed to be shifted down about 21 cents each time.  After
tuninging the bass tracks properly, I duplicated the bass line on a Just
track, adjusted its channel Transpose and Detune parameters (first by "the
numbers" and then fine tuning by ear) at each mod to keep it in tune with the
Pythagorean track, then used trial and error to find the SysEx messages needed
to set those parameters automatically during playback.

Thankfully, Korg's pitch bend implementation seems to be perfectly consistent,
allowing the use of the formula cents*8192/P.B.Range to tune individual notes
on the fly on monophonic tracks.  It's surprising to me that Yamaha's
implementation is less reliable.

Those cheesy brass and wind samples that most of us have to put up with sound
a lot more realistic when they're in tune!  Maybe that's a good place to start
when demonstrating the value of tuning to other keyboard players.

David J. Finnamore
Just tune it!