Re: polytempi in Lilypond - is it possible?

[David Wright]

On Tue 25 Oct 2016 at 21:48:35 (+0200), Bálint Laczkó wrote:
> I would like to engrave a polytempical musical material in LilyPond. I am a
> newbie to this software so I am still learning the basics, but I saw quite
> a few mentions in some forum comments here and there, that LilyPond support
> polymetry and polytempi quite well and folks do it all the time. Now I
> found many useful tutorials/examples/documentation about polymetry but as
> for polytempi I am still stuck in the darkness.

You might have a steep learning curve ahead...

> What I would like to see is similar to this:
> [image: Szövegközi kép 1]
> ...except for the ending, I would like to to it strictly proportionally.
> This is a screenshot from [bach.score], a Max MSP object/package which can
> do polytempi quite well, but it exports pdf files via LilyPond, and somehow
> in this case they don't quite understand each other. Anyway, actually I
> would be more interested in doing it all the way in LilyPond without
> introducing any weird syntax-conversion from another software.

I'm not sure if I'm seeing what you posted; are there two images, or
is the screenshot from [bach.score] the same image as [image: Szövegközi kép 1] 
?

In any case, what I see is five pieces of music which I assume you
want played simultaneously in the precise tempos given. From the point
of view of the performers, producing parts is simple enough, but you
probably realise that.

> Now ideally I would like to see this engraved in a proportional way, so
> that the lower staff has a wider spacing, and so the notes are distributed
> in the two staves illustrating their exact relation to a mutual time-scale
> (so in short: sort of like on the picture above).
> Is there any way I can do this? I only read that some guy did it with
> giving individual horizontal spacings for each staff, but there were no
> examples for that, and while it is not quite the thing (ideally it would be
> like LilyPond understands and calculates the time-differences, and most
> ideally it would render a coherent MIDI representation of it as well), I
> can't even do that one.

To be realistic, I think you have to call this piece what it is: a
synthesis of musical fragments played simultaneously. You can stretch
conventional music notation a long way¹ but I think cartesian graph
paper would be more suited to your purpose—not necessarily physical
paper, a drafting package like Inkscape would do.

If you set each score in strict proportional spacing in one long line
(as long as that particular tempo is maintained) you can then slice
the output into page-width chunks and paste these into Inkscape.
However, you may be disappointed by the precision that proportional
spacing gives you even if you typeset the line as a single measure
and then draw the barlines using Inkscape's ruler.

Having set each score to the appropriate length, you can now assemble
the slices into the synthesis you desire. Quite what the rôle of a
conductor is, when looking at this piece, I'm not sure, because
(a) there's no beat to conduct and (b) I think that grasping five
different unrelated tempi is similar to comprehending two texts being
read into one's ears at the same time: beyond human ability.

There's a fundamental difference between your desire for X-axis
precision of note/barline placement and a notation that is based on
simple subdivisions of a beat. That gap may be unbridgable.
In addition, if the players are human, are they playing with
metronomes to maintain their speed? If not, the precision being
displayed by your printout is not going to be representative of the
sound your hear from the players anyway.

And one last question on the piece: how does the ensemble obey
Beecham's Golden Rule No. 2? :)

¹ The furthest I can think of is Britten's "Curlew River", see
http://scorelibrary.fabermusic.com/Curlew-River-22303.aspx
where groups of performers have independent tempi but use a system of
rendezvous (curlew marks) to maintain order. (There's no conductor.)

Cheers,
David.