Re: Pitch-bent note's release-phase

[Graham Breed]

Richard Sabey <address@hidden> wrote:

> Since sending my initial enquiry to this list I have
> indeed written an app which confirmed to me what you say.

Right.  I also have a file that shows Timidity will share
the tuning of a note even between channels, so that F and
F# come out the same in meantones.  I've updated my script
to work around this by setting each channel to a different
tuning table.  The result is at least as good as pitch
bends now.

> Thanks also for that Python app. I plumped for your
> Python v.3.2 version and installed Python v.3.2
> accordingly. 
> 
> >There is a problem, in that if two notes are assigned to
> the same MIDI key (even in different voices) they'll have
> the same tuning.  That makes it possible for C and C# to
> become confused.
> 
> That's a pity. I was hoping to use temperaments with many
> (31? 34? 41? 53?) pitches per octave, and create chords
> with pretty small intervals, perhaps smaller than a 12et
> semitone...

Until they get smaller than a semitone, there won't be a
problem because (touch wood) no plausible function is going
to assign two pitches more than 100 cents apart to the same
key number.  When the intervals get narrower than that, you
have to start thinking about assigning the pitches to
different channels, or transposing so that the rounding
comes out right.  When it becomes a real problem is when we
can suggest Lilypond be made more complex to work around it.

One thing that does work is to "transpose" pitches
into a tuning table with more than 12 notes.  That
restricts the range of each voice but you still get 4
octaves of 31 notes, for example.

> >It would make sense if
> the pitch were rounded to the *nearest* equally tempered
> value instead.
> 
> That'd only move this problem elsewhere.

Yes, but it's easier to predict that F# and Gb are going to
round to the same pitch, and so avoid them in the same
chord, than that F and F# will be a problem.

> I've seen Lilypond apply downward pitch bend even if this
> means using a MIDI note whose name is different from the
> name in the .ly.

It does?  I don't know what's going on then.


                     Graham