more MIDI tuning madness

i really find the "official" MIDI webpages
to be no help at all.

look at "Message #5" on this page

http://www.midi.org/about-midi/tuning_extens.shtml

----------- begin quote from official MIDI webpage ----------

Message #5

SCALE/OCTAVE TUNING DUMP, 1 byte format

F0 7E <device ID> 08 05 bb tt <tuning name> [xx] ... chksum F7

F0 7E Universal Non-Real Time SysEx header

<device ID> ID of target device (7F=all devices)

08 sub-ID#1 = "MIDI tuning standard"

05 sub-ID#2 = "SCALE/OCTAVE TUNING DUMP, 1 byte format"

bb bank: 0-127
(described as 1-128 in MIDI Tuning Specification)

tt tuning preset number: 0-127

<tuning name> 16 ASCII characters

[xx] frequency data for C,C#,... B (12 bytes total)
00H means -64 Cent
40H means +/- 0 Cent
7FH means +63 Cent

chksum see "Checksum calculation", below

F7 EOX

----------------------- end quote -------------------

the part under consideration here is the byte
designated "[xx]".

they don't really mean +/- 64 *cents*, do they?

shouldn't it be +/- 64 hexamus, as i've defined "hexamu"?

if you agree with Aaron and don't like my definition,
call it whatever you want ...

but my point is: it should be +/- 100 cents, in
the form of units which are 100/64 = 1 & 9/16
= 1.5625 cents each. correct?

the following message, #6, is definitely correct
(look at the part describing "[xx yy]"), and
really makes #5 seem wrong:

----------- begin quote from official MIDI webpage ----------

Message #6

SCALE/OCTAVE TUNING DUMP, 2 byte format

F0 7E <device ID> 08 06 bb tt <tuning name> [xx yy] ... chksum F7

F0 7E Universal Non-Real Time SysEx header

<device ID> ID of target device (7F=all devices)

08 sub-ID#1 = "MIDI tuning standard"

06 sub-ID#2 = "SCALE/OCTAVE TUNING DUMP, 2 byte format"

bb bank: 0-127
(described as 1-128 in MIDI Tuning Specification)

tt tuning preset number: 0-127

<tuning name> 16 ASCII characters

[xx yy] frequency data for C,C#,... B (24 bytes total)
00H 00H means -100 cents (8,192 steps of .012207
cents)
40H 00H means 0 cents (equal temperament)
7FH 7FH means +100 cents (8,191 steps of .012207
cents)

chksum see "Checksum calculation", below

F7 EOX

The minimum and maximum offsets are approximately 100 cents.

----------------------- end quote -------------------

and to make things even more confusing, that last line
about "100 cents" only appears under Message #6, and
in the regular font for the webpage text, which is
different from the font used for the message specs.

i understand why it says "approximately" 100 cents,
thanks to Aaron's explanation that the upward pitch-bend
cannot misses the next MIDI-note frequency by 1 unit.

but it's not clear whether this line is supposed to apply
only to Message #6 or to both #5 and #6.

if it *doesn't* apply to Message #5, then is there
*really* a new MIDI tuning spec which allows
+/- 64 cents?

... or is this simply yet another error on the
official MIDI pages, which seem to be full of mistakes.

H E L P !

-monz

oops ...

> From: "monz" <monz@attglobal.net>
> To: <tuning@yahoogroups.com>
> Sent: Monday, July 07, 2003 7:37 AM
> Subject: [tuning] more MIDI tuning madness
>
>
> <snip>
>
> The minimum and maximum offsets are approximately 100 cents.
>
> ----------------------- end quote -------------------
>
>
> and to make things even more confusing, that last line
> about "100 cents" only appears under Message #6, and
> in the regular font for the webpage text, which is
> different from the font used for the message specs.
>
> i understand why it says "approximately" 100 cents,
> thanks to Aaron's explanation that the upward pitch-bend
> cannot misses the next MIDI-note frequency by 1 unit.

the "cannot" should have been deleted:

... the upward pitch-bend misses the next MIDI-note
frequency by 1 unit.

-monz

It all looks correct to me.

>they don't really mean +/- 64 *cents*, do they?

This is similar to what some Roland synths have.
If there's hardware that only can do this kind of tuning,
then it makes sense to define a message to support it.

>but my point is: it should be +/- 100 cents, in
>the form of units which are 100/64 = 1 & 9/16
>= 1.5625 cents each. correct?

No that would only lower the resolution.

>and to make things even more confusing, that last line
>about "100 cents" only appears under Message #6, and
>in the regular font for the webpage text, which is
>different from the font used for the message specs.

Yes, it only applies to message #6. This is the one
Aaron was referring to. It has lower resolution than
the bulk dump, I guess they didn't want to add another
byte with 6 unused bits.

Manuel

--- In tuning@yahoogroups.com, "monz" <monz@a...> wrote:
> i really find the "official" MIDI webpages
> to be no help at all.

Pretty grisly, aren't they? This is why I was suggesting doing an
explanation that made more sense. As we have just seen, even the
experts have been confused by this!

> the following message, #6, is definitely correct
> (look at the part describing "[xx yy]"), and
> really makes #5 seem wrong:

#5 and #6 seem to be describing different systems, with #6 having the
12 * 8192 = 98304-et Aaron seems to be using. I've been trying to
clear up, at least, what the MTS standard is, which uses twice that.

What a mess!