Re: MIDI keyboard

[Richard Shann]

On Tue, 13 Aug 2013 11:50:16 -0400
Carl Peterson <address@hidden> wrote:

> On Tue, Aug 13, 2013 at 11:19 AM, Richard Shann
> <address@hidden>wrote:
> 
> >
> > here is your problem. You are hoping that the timing of your
> > keypress could be interpreted and a duration of note estimated from
> > it. Such systems have been tried many times, and are offered by
> > programs that don't care if you succeed or not, as long as you buy
> > the program. They don't work because of the subtleties of timing,
> > rests and notation (consider, 1/4 note tied to 1/8 note is the same
> > duration as dotted 1/4 note).
> > Well, I would like to be proved wrong; the moment you hear of a way
> > of doing it I promise I will implement it in Denemo: everything is
> > there just waiting for someone to invent the algorithm.
> >
> 
> Richard,
> ICBW, but I think that *usually*, 4. vs 4~8 depends on the context
> and the time signature. For instance, I was told to break and tie
> notes if they cross the midline of a duple or quadruple measure (so
> "c4 c4. c8 c4" would be written as "c4 c4~c8 c8 c4" in 4/4 and "c8 d
> e4 f8 g" as "c8 d e~e f g" in 6/8), but there are others that are
> largely stylistic (such as whether to break a quarter note if it
> crosses any beat at all).

yes, I just chose an example at random, there is no 1-1 relationship
between performance and notation; for any entry system to be useful it
has to be highly reliable, fixing mistakes has to be counted as a very
high penalty for any entry system. That is why Optical Music
Recognition (OMR, see Audiveris for example) is still slower than
entering music by playing in, usually.

> 
> One option would be to have a MIDI-entry mode and notate based on
> actual durations (i.e., notate a 4. if that was what was played),
> then present it to the user to review with a popup of some sort to
> allow for alternate notations (e.g., show c4~c8 or c8~c4 [depending
> on where the beat is] as an alternate to c4.) before entering into
> the score proper.

The set of alternate notations in music is very large. For music that
sticks to a reasonably small set of idioms presenting these to the
program first and playing them on the MIDI keyboard to teach the
program how you play them sounds like a better bet. This would be a
similar task to the OMR, and if someone
creates a library that does this I'll be the first to use it.

If you can read music fluently and have a lot of music to enter
sequentially into LilyPond then Denemo gives you a way of leveraging
your sight-reading skill to enter the music by allowing you to enter it
"in music time" - that is you can keep track of where you are in the
music entry process because you are reading and playing the music as
music, not as a set of letters with numbers, dots, apostrophes etc. But
if I could cut it down to a single play through instead of two, I would
be even happier.

Richard