Re: [Rosegarden-user] Note modification

[Russell S. <rc....@sy...>]

Chris,

I think I can see your point. You are correct that the score is recorded
from midi. As a result there is the problem you suggest below with
quantization of the notes.

This got me thinking about other parts of my recording. I have attached
two png files. One of the notated midi and the other of the matrix view
of the same segment.

There are four bars shown (the fourth is only a small part of the bar).
At the end of the second bar a note starts around the third beat and a
second note starts on the fourth beat. I would expect this to be notated
as a note tied to a chord which is tied to a note. The chord would be
the duration of the shorter note, with the tied end notes the duration
of the "tails" of the longer note. Not very readable I agree, but it
does accurately reflect what is happening. 

As you can see in the notation view this is not what happens. The
impression given by the notation is that these notes follow each other.
It's a bit hard to count it out but I suspect if you do the notes don't
add to 4 beats if you assume they follow each other.

Am I correct in thinking the notator is not working correctly?

Russell.


On Mon, 2002-12-30 at 16:37, Chris Cannam wrote:
> On Monday 30 December 2002 20:08, Russell Smiley wrote:
> > I'm just wondering if someone can explain what is going on with the
> > note modification. I have attached two png screen captures
> > illustrating the problem.
> 
> Interesting.  I think what's happening is that the first note in the 
> first screenshot is not actually at the start of the bar, it's about 
> a quaver later.  So when you try to modify it to the semibreve, the 
> semibreve doesn't actually fit in the bar and is therefore split into 
> a double-dotted minim in that bar and a quaver in the following one, 
> which in turn causes the split-and-tie of the note in that bar.
> 
> While it's fairly easy (once you know how Rosegarden is thinking) to 
> see what the problem is here, it's harder to see how one should solve 
> it generally and elegantly.  The data presumably came from a MIDI 
> recording, and it's not always very safe for Rosegarden to make too 
> many assumptions about the precision of recorded data, although it 
> could do a better job than it does now.  In practice the solution is 
> usually to quantize harder, but obviously that screws up your 
> performance timing.  The Smoothing slider on the notation window is 
> supposed to do a careful non-destructive quantize for notation 
> purposes alone, but it's unlikely at the moment to deal with that 
> situation well (although I have a few ideas about how it could be 
> improved).
> 
-- 
Russell Smiley
rc....@sy...