Re: midipu

[Monz writes, at http://www.ixpres.com/interval/dict/midipu.htm:]
>At the setting for finest pitch-bend resolution, a semitone is divided
>into 212 = 4096 pitch-bend units, or midipus. Thus there are 4096 * 12
>= 49152 midipus in an "octave", so the midipu measurement system may be
>thought of as 49152-EDO tuning, with a midipu being one degree in
>49152-EDO.

No, that's not right. At the DEFAULT setting, with the complete range
of 2^14 values mapping to +/-2 semitones, that is correct. But the MIDI
spec allows for overriding the default range, both larger and smaller.

[Manuel Op de Coul wrote:]
>I find your "midipu" definition misleading since there is no
>requirement in the MIDI specification for this unit to actually
>represent an interval of 1/49152 octave. Because the actual
>value depends on the pitch bend range. Moreover there is no
>standard way to set the pitch bend range. There's only a
>(I think non-mandatory) assumption that the range is from
>minus to plus one 12-tET whole tone when the instrument is
>switched on or reset.

Yes, it's my understanding that the MIDI spec is carefully vague about
actually requiring adherence in this regard. BUT, there is a suggested
convention in the spec for setting the bend range:

B0 65 00 (Registered Parameter Number, course)
B0 64 00 (Registered Parameter Number, fine)

B0 06 02 (data entry slider, high bits (coarse))
B0 26 00 (data entry slider, low bits (fine))

(repeat for each channel). The above example instructs the range to be
the usual default, plus and minus 2 semitones and no cents.

JdL